PREDICTION METHOD, PREDICTION DEVICE, AND PREDICTION PROGRAM

A prediction method of performing heat source demand prediction in a space having a predetermined air conditioning control region, includes predicting a required heat quantity in the air conditioning control region by using a predetermined parameter related to a surrounding environment of the air conditioning control region and a setting value of an air conditioner set in the air conditioning control region as inputs, and predicting a heat source demand of the entire space from the predicted required heat quantity for each air conditioning control region.

Skip to: Description  ·  Claims  · Patent History  ·  Patent History
Description
TECHNICAL FIELD

The disclosed technique relates to a prediction method, a prediction device, and a prediction program.

BACKGROUND ART

Air conditioning equipment in large facilities such as buildings regulates the temperature of the buildings on the basis of an operation scenario. In an operation of air conditioning equipment, a heat source device generates hot water or cold water as a heat quantity, and sends the generated hot water or cold water to each air conditioning control region (air conditioning use portion), and each air conditioning control region generates hot air or cold air with an air conditioner by using the hot water or cold water. The air conditioning control region is a partial region of a building and consumes the heat quantity obtained by the heat source device. When calculating an operation scenario for the heat source device for satisfying a set temperature set for the entire large facility, it is necessary to predict a heat source demand to be used in a heat source in advance. The operation scenario for the heat source device is planned to reduce the power consumption in consideration of startup power and operation power while satisfying a predicted heat source demand. There are two types of methods of heat source demand prediction: a physical model using a simulator and a learning model for constructing a certain mathematical model through machine learning or the like.

In the case of using the physical model, it is necessary to set a large number of specific parameters based on a configuration of a building, and there is a problem that highly specialized knowledge is required to determine a setting value, and when there is an error in the setting value, an error of a prediction value also increases. In the case of using the learning model, since prediction is performed on the basis of the past heat consumption amount for each building, parameter tuning as in a physical model is unnecessary. Non Patent Literature 1 discloses a technique for predicting a heat source demand by using machine learning.

CITATION LIST Non Patent Literature

Non Patent Literature 1: Matsunaga, K., Kaminaga, M., Nagata, K., Kawamura, T., Nakamura, R., Kunitomo, O., & Sasajima, K. Air conditioning and heat source device cooperative control by Senems through co2-saving community planning by smart energy network (Report 14). Proceedings of the Annual Conference of the Society of Heating, Air conditioning and Sanitary Engineers, Japan, B-7, 29-32. (2016).

SUMMARY OF INVENTION Technical Problem

However, the technique in Non Patent Document 1 uses only external parameters of a building such as the outside air temperature or the outside humidity, and does not consider the surrounding environment of the air conditioning control region, for example, settings of an air conditioner and a heat source of a moving object typified by a human. Thus, in a case where there is a pattern in which the surrounding environment of the air conditioning control region is different even a day on which a parameter outside the building is the same, the accuracy of the heat source demand prediction decreases. As a result, in a case where a large heat source demand is predicted, energy saving performance is deteriorated due to acquisition of unnecessary heat quantity, and in a case where a small heat quantity source demand is predicted, comfort is deteriorated due to shortage of heat quantity. That is, there is a problem that a set temperature cannot be achieved, and an indoor air temperature becomes close to an outside air temperature and a humidity becomes close to an outside humidity.

The disclosed technique has been made in view of the above circumstances, and an object thereof is to provide a prediction method, a prediction device, and a prediction program capable of performing appropriate heat source control by predicting a heat source demand for each air conditioning control region.

Solution to Problem

According to a first aspect of the present disclosure, there is provided a prediction method of performing heat source demand prediction in a space having a predetermined air conditioning control region, the prediction method including predicting a required heat quantity in the air conditioning control region by using a predetermined parameter related to a surrounding environment of the air conditioning control region and a setting value of an air conditioner set in the air conditioning control region as inputs; and predicting a heat source demand of the entire space from the predicted required heat quantity for each air conditioning control region.

Advantageous Effects of Invention

According to the disclosed technique, appropriate heat source control can be performed by predicting a required heat quantity for each air conditioning control region.

BRIEF DESCRIPTION OF DRAWINGS

FIG. 1 is a block diagram illustrating a hardware configuration of a prediction device.

FIG. 2 is a block diagram illustrating a configuration of the prediction device of the present embodiment.

FIG. 3 is a diagram illustrating an example of each portion of a building and an external service according to input data of the prediction device.

FIG. 4 is a flowchart illustrating a flow of a learning process by the prediction device of the present embodiment.

FIG. 5 is a flowchart illustrating a detailed flow of a parameter determination process.

FIG. 6 is a flowchart illustrating a detailed flow of a model learning process.

FIG. 7 is a flowchart illustrating a flow of a prediction process by the prediction device of the present embodiment.

FIG. 8 is a diagram illustrating a verification result by verification A.

FIG. 9 is a diagram illustrating a verification result by verification B.

 DESCRIPTION OF EMBODIMENTS

Hereinafter, an example of an embodiment of the disclosed technique will be described with reference to the drawings. In the drawings, the same or equivalent constituents and portions are denoted by the same reference numerals. Dimensional ratios in the drawings are exaggerated for convenience of description, and may be different from actual ratios.

First, an outline of the present disclosure will be described. The technique of the present disclosure proposes a technique for estimating a heat demand by predicting a demand for a heat source for each air conditioning control region on the basis of a surrounding environment of each air conditioning control region and calculating a sum of the demands for the heat sources predicted for respective air conditioning control regions. In the existing technique, a heat source demand of the entire building in which all air conditioning control regions are put together is predicted. In contrast, in the technique of the present disclosure, a heat source demand of the entire building is predicted by predicting a required heat quantity for each air conditioning control region from parameters related to a surrounding environment of each air conditioning control region and a setting value of air conditioning equipment. Examples of the parameters related to the surrounding environment include weather data and people flow data that will be described later, but any data having thermal energy that influences a target air conditioning control region may be further used.

Hereinafter, a configuration of the present embodiment will be described.

FIG. 1 is a block diagram illustrating a hardware configuration of a prediction device 100.

As illustrated in FIG. 1, the prediction device 100 includes a central processing unit (CPU) 11, a read only memory (ROM) 12, a random access memory (RAM) 13, a storage 14, an input unit 15, a display unit 16, and a communication interface (I/F) 17. The constituents are communicably connected to each other via a bus 19.

The CPU 11 is a central processing unit, and executes various programs and controls each unit. That is, the CPU 11 reads the program from the ROM 12 or the storage 14, and executes the program by using the RAM 13 as a work region. The CPU 11 performs control of each of the above-described constituents and various types of operation processing according to the programs stored in the ROM 12 or the storage 14. In the present embodiment, a learning processing program and a prediction processing program are stored in the ROM 12 or the storage 14.

The ROM 12 stores various programs and various data. The RAM 13 temporarily stores programs or data as a work region. The storage 14 includes a storage device such as a hard disk drive (HDD) or a solid state drive (SSD), and stores various programs including an operating system and various data.

The input unit 15 includes a pointing device such as a mouse and a keyboard, and is used to perform various inputs.

The display unit 16 is, for example, a liquid crystal display, and displays various types of information. The display unit 16 may function as the input unit 15 by adopting a touch panel system.

The communication interface 17 is an interface for communicating with another device such as a terminal. For the communication, for example, a wired communication standard such as Ethernet (registered trademark) or FDDI, or a wireless communication standard such as 4G, 5G, or Wi-Fi (registered trademark) is used.

Next, each functional configuration of the prediction device 100 will be described. FIG. 2 is a block diagram illustrating a configuration of the prediction device of the present embodiment. Each functional configuration is achieved by the CPU 11 reading a learning program and a prediction program stored in the ROM 12 or the storage 14, loading the programs to the RAM 13, and executing the programs.

As illustrated in FIG. 2, the prediction device 100 includes, regarding the learning process, a learning data collecting unit 110, a learning data storage unit 112, a parameter determination unit 114, a parameter storage unit 116, a learning model creation unit 118, and a learning model storage unit 120. The prediction device 100 includes, regarding the prediction process, a prediction data collecting unit 130, a prediction data storage unit 132, a demand prediction unit 134, and a prediction result storage unit 136. The processing unit related to the learning process and the processing unit related to the prediction process may be divided into separate devices and configured as a learning device and a prediction device, respectively.

The parameter determination unit 114 includes a first variable creation unit 210 and a first model creation unit 212. The learning model creation unit 118 includes a second variable creation unit 220 and a second model creation unit 222. The demand prediction unit 134 includes a third variable creation unit 230.

Here, a mode of input data that is input to each of the learning data collecting unit 110 and the prediction data collecting unit 130 of the prediction device 100 will be described. FIG. 3 is a diagram illustrating an example of each portion of a building 50 and an external service related to input data of the prediction device 100. In the example illustrated in FIG. 3, the building 50 includes a building and energy management system (BEMS) 52 and each air conditioning control region 54, and has a people flow detection sensor 56 for each air conditioning control region 54 (the reference numeral of the air conditioning control region 54 is used only in the description of FIG. 3, and the reference numeral will be omitted hereinafter). The BEMS 52 manages an energy use amount based on use energy for each air conditioning control region 54. The use energy is a type of energy generated by a heat source (for example, an air conditioner), and in the present embodiment, the BEMS 52 manages an energy use amount of hot water and cold water. The term “based on use energy” means that data is separately managed between hot water and cold water. The BEMS 52 measures weather data for each air conditioning control region 54. The people flow detection sensor 56 in each air conditioning control region 54 detects a people flow in the air conditioning control region 54 and measures people flow data. The weather data for each air conditioning control region 54, a setting value of the air conditioner for each air conditioning control region 54, and the energy use amount based on use energy for each air conditioning control region 54 are input to the learning data collecting unit 110. Predicted weather data that is a predicted value of the weather data, predicted people flow data that is a predicted value of the people flow data, and a scheduled setting value of the air conditioner for each air conditioning control region 54 are input to the prediction data collecting unit 130. A people flow predictor 60 predicts the predicted people flow data on the basis of the people flow data from the people flow detection sensor 56. The predicted weather data is predicted by using an external weather service 62. An example of data input to the prediction device 100 has been described above. The technique of the present disclosure will be described by exemplifying a case where a required heat quantity of the air conditioning control region 54 in the building 50 is a prediction target, but is not limited thereto, and it is also assumed that a prediction target space is expanded and used for prediction of the demand for a district heat source. In a case of handling a district heat source, it is assumed that the supply of heat sources to respective buildings existing in the district is collectively managed. In a case where the demand for the district heat source is predicted, the technique of the present disclosure is applied by replacing the building 50 with a district and replacing each air conditioning control region 54 with a building.

The learning process and the prediction process based on the input data will be described below.

Learning Process

A configuration of a processing unit related to the learning process will be described.

The learning data collecting unit 110 collects learning data necessary for determining a time delay parameter and creating a learning model, and stores the learning data in the learning data storage unit 112. The learning data is collected as time-series data of actual measurement values serving as an explanatory variable and an objective variable. The learning data is present for each air conditioning control region. The learning data to be collected is described in Table 1 below, and an example of a storage format of the learning data storage unit 112 is shown in Table 2.

TABLE 1 Data Specific example Application Weather data *Consider change in time Outside air temperature corresponding to air conditioning control region Explanatory variable Outside humidity corresponding to air conditioning control region Solar radiation amount corresponding to air conditioning control region Wind speed corresponding to air conditioning control region People flow data *Consider change in time Number of unique persons present in air conditioning control region Explanatory variable Stay time (sec) of person present in air conditioning control region Setting value of air conditioner Room temperature setting value of air conditioner in air conditioning control region Explanatory variable Air volume setting value of air conditioner in air conditioning control region Supply air temperature setting value of air conditioner in air conditioning control region Air conditioner operation status in air conditioning control region Use energy amount Cold water energy use amount used in air conditioning control region Objective variable Hot water energy use amount used in air conditioning control region

TABLE 2 Date and time Weather data: outside air temperature People flow data: number of people Room temperature setting value Cold water energy use amount Hot water energy use amount ... 2020-01-01 09:00:00 10.0 100 20 0 50 2020-01-01 09:10:00 10.2 110 20 0 60 ...

In the example of Table 1, the learning data is classified into weather data, people flow data, a setting value of an air conditioner, and an energy use amount. An application of these learning data is divided into an explanatory variable and an objective variable. An application of the weather data, the people flow data, and the setting value of the air conditioner is an explanatory variable, and an application of the energy use amount is an objective variable. The weather data and the people flow data are determination targets of a time delay parameter for considering the influence of a change in time, and an average value per unit time is obtained by using the time delay parameter. The average value thus obtained is used as an explanatory variable in consideration of the time delay. The weather data and the people flow data that are targets for creating an explanatory variable in consideration of the time delay are examples of time-considered data of the technique of the present disclosure. The people flow data is not necessarily required, and sensor data from which the number of humans present (the number of humans that was present) and further a sum of quantities of heat radiated by humans present per unit time can be obtained may be used.

Table 2 is an example of the learning data stored in the learning data storage unit 112. Although omitted in Table 2, it is assumed that learning data is collected every unit time as time-series data for one month from January 1 to 31, 2020. The unit time is set to every 10 minutes. Table 2 illustrates an example in which an outside air temperature is collected as the weather data the number of people is collected as the people flow data, and a room temperature setting value is collected as the setting value. A cold water energy use amount and a hot water energy use amount are set for cold water and hot water as use energy.

A specific example of the weather data is a weather data group of an outside air temperature, an outside humidity, a solar radiation amount, and a wind speed corresponding to the air conditioning control region. It is assumed that a weather data group corresponding to each of the air conditioning control regions is collected by the BEMS 52. For example, in the case of outside air information, sensor data acquired by a weather sensor closest to the air conditioning control region may be used. If the outside air information is acquired for each air conditioning control region by the BEMS 52, the sensor data may be used. However, since it is supposed that it is difficult to collect a weather data group for each air conditioning control region, for example, a weather data group for the building 50 (BEMS 52) of a weather service 62 may be uniformly handled as a weather data group for each air conditioning control region.

A specific example of the people flow data is the number of unique people (the number of people) per unit time present in the air conditioning control region or a stay time of a person per unit time present in the air conditioning control region. In the case of the number of people, the number of people present during the unit time may be measured by the people flow detection sensor 56. As the number of unique people, for example, the number of people who were in the region within the unit time may be counted. Regardless of how long a person stayed, counting is performed in a case where the person was present in the region. For example, a case is assumed in which a certain people flow detection sensor 56 tracks a person in a region by assigning an ID to the person. In this case, it is assumed that the following ID = 1 and ID = 2 are detected within the unit time. For ID = 1, it is assumed that (1-1) ID = 1 was detected, (1-2) ID = 1 stayed in the region for several minutes, and (1-3) ID = 1 moved to the outside of the region. For ID = 2, it is assumed that (2-1) ID = 2 was detected, and (2-2) ID = 2 was detected to have moved to the outside of the region in several seconds. In this case, the number of unique persons is counted as two. In the case of the stay time of a person, for example, an average value of “persons × stay time (seconds)” for respective persons present in the air conditioning control region for the unit time may be measured by the people flow detection sensor 56.

Specific examples of the setting value of the air conditioner include a room temperature setting value of the air conditioner in the air conditioning control region, an air volume setting value of the air conditioner in the air conditioning control region, a supply air temperature setting value of the air conditioner in the air conditioning control region, and an air conditioner operation status in the air conditioning control region.

Although specific examples of data that can be used as explanatory variables have been listed, only some specific examples may be used. For example, data used as an explanatory variable related to weather information may be only the outside air temperature or a combination of the outside air temperature and the outside humidity. It goes without saying that explanatory variables may be defined by using specific examples other than those listed. In the case of weather data, other weather-related data that is an external parameter of the building and influences a heat quantity used in the air conditioning control region may be used. For the people flow data, data of another object serving as a heat source that influences the air conditioning control region present in the building may be used.

The parameter determination unit 114 determines time-considered data for each air conditioning control region, that is, a time delay parameter for each of the weather data and the people flow data. The time delay parameter is a maximum time tmax to be considered before a prediction target time, and a time width Δt indicating a time interval at which an average value from the maximum time tmax to the prediction target time is desired to be obtained. The parameter determination unit 114 determines the time delay parameter by performing processes of the first variable creation unit 210 and the first model creation unit 212 as internal processing, and stores the time delay parameter in the parameter storage unit 116. Details of a parameter determination process of the parameter determination unit 114 and processes of the first variable creation unit 210 and the first model creation unit 212 related to the parameter determination process will be described later in the description of the actions.

In a case where the weather data, the people flow data, and the like are used as explanatory variables, it is necessary to consider the influence of the change in time for each space of the air conditioning control region, that is, how much time delay the space of a certain air conditioning control region is influenced by the explanatory variables. For example, it is considered that a space adjacent to the outside of the room is easily influenced by the outside air temperature immediately, that is, the time delay is shortened. On the contrary, the time delay is considered to be long in a space where there is a passage between the outside and the room.

In the present technique, in order to take into account the influence of such a time delay, “a plurality of average values before the prediction target time” is used as an explanatory variable. For example, when prediction is performed at 12:00, an average value from 9:00 to 10:00, an average value from 10:00 to 11:00, and an average value from 11:00 to 12:00 are used. Since a parameter for determining during which time an average value is to be obtained is required, the maximum time tmax and the time width Δt are determined as time delay parameters.

The learning model creation unit 118 learns a prediction model for use energy for each air conditioning control region by using the learning data and the time delay parameter stored in the parameter storage unit 116, and stores the learned prediction model in the learning model storage unit 120. The prediction model learned here is a model for predicting the required heat quantity of the air conditioning control region. The learning model creation unit 118 learns the prediction model by performing the processes of the second variable creation unit 220 and the second model creation unit 222 as internal processing. Details of a model learning process of the learning model creation unit 118 and processes of the second variable creation unit 220 and the second model creation unit 222 related to the model learning process will be described later in the description of the actions.

Next, the actions of the learning process of the prediction device 100 will be described.

FIG. 4 is a flowchart illustrating a flow of a learning process by the prediction device 100. A learning process is performed by the CPU 11 reading a learning processing program from the ROM 12 or the storage 14, loading the learning processing program in the RAM 13, and executing the learning processing program.

In step S100, the CPU 11 as the learning data collecting unit 110 collects learning data and stores the learning data in the learning data storage unit 112.

In step S102, the CPU 11 as the parameter determination unit 114 determines a time delay parameter of each of the weather data and the people flow data for use energy for each air conditioning control region, and stores the time delay parameters in the parameter storage unit 116.

In step S104, the CPU 11 as the learning model creation unit 118 learns a prediction model for the use energy for each air conditioning control region by using the learning data and the time delay parameters, and stores the learned prediction model in the learning model storage unit 120.

Next, the parameter determination process in step S102 will be described with reference to a flowchart of FIG. 5. In the parameter determination process of the present embodiment, as an example, a case where the time delay parameter is determined by performing cross verification of an explanatory variable of the learning data by separately setting a verification day and a training day will be described. FIG. 4 is a flowchart illustrating a parameter determination process. In the parameter determination process in FIG. 4, the time delay parameter for use energy that is used for each air conditioning control region may be determined. The use energy that is used for each air conditioning control region may be appropriately determined from data of an objective variable as an application of the learning data. For example, if an energy use amount of hot water (or cold water) for the air conditioning control region is only 0, the parameter determination process for the hot water (or cold water) may be omitted, and in a case where there is an input value other than 0, the parameter determination process may be performed. That is, the parameter determination process may be performed in a case where the use energy is used in the air conditioning control region.

In step S1100, the CPU 11 as the parameter determination unit 114 acquires the learning data stored in the learning data storage unit 112.

In step S1102, the CPU 11 as the parameter determination unit 114 selects an air conditioning control region that is to be a processing target.

In step S1104, the CPU 11 as the parameter determination unit 114 selects a combination of the time delay parameters corresponding to the air conditioning control region that is a processing target, and outputs the selected combination to the first variable creation unit 210.

A sample of the combination will be described. For the weather data, the outside air temperature and the number of people in the people flow data are used. Assume that a combination is tried with setting times of the maximum time tmax = [60 min, 120 min, 180 min, 240 min] and the time width Δt = [10 min, 20 min, 30 min, 40 min, 50 min, 60 min] in each of the weather data and the people flow data. The number of combinations of setting times of the maximum time tmax and the time width Δt set is 4 × 6 = 24, and the number of trials is 24 × 24 = 576 since the combination is changed depending on the outside air temperature and the number of people. A total number of trials is 1152 since a verification model of an energy use amount of cold water and a verification model of an energy use amount of hot water are tried, and if it takes 10 minutes in one trial, all trials are ended in 8 days, which can be said to be a feasible range. Consequently, for example, values as shown in Table 3 below are determined. The same setting time of the maximum time tmax and the same setting time of the time width Δt are used in the weather data and the people flow data, but different setting times may be used in each piece of data. The setting time of the maximum time tmax and the setting time of the time width Δt are examples of a setting time for the maximum time and a setting time for a time width of the technique of the present disclosure.

Table 3 shows an example of a case where the maximum time tmax and the time width Δt, which are time delay parameters, are determined for each of cold water and hot water as use energy. These time delay parameters are used in a prediction model for cold water and a prediction model for hot water. Although the corresponding air conditioning control region is not shown in Table 3, the time delay parameter in Table 3 is determined for each air conditioning control region.

TABLE 3 Model Weather data: outside air temperature People flow data: number of people tmax Δt tmax Δt Prediction model for cold water energy use amount 240 40 240 50 Prediction model for hot water energy use amount 180 50 60 40

In step S1106, the CPU 11 as the first variable creation unit 210 obtains an average value of the time-considered data by using the time delay parameter of the combination selected in step S1104, and creates time-series data of the explanatory variable including the average value.

Here, inputs and outputs of the first variable creation unit 210 will be described. The average value of the time-considered data of the selected combination obtained by the first variable creation unit 210 is an example of a predetermined calculated value of the technique of the present disclosure. In addition to the average value, a weighted average value, a median value, or the like obtained by weighting each setting time may be used.

Table 4 shows inputs and outputs of the first variable creation unit 210. The inputs and outputs shown in Table 4 are common to the second variable creation unit 220 and the third variable creation unit 230.

TABLE 4 Input Time-series data used for explanatory variable (application in table 1 is data of explanatory variable) Time delay parameter of explanatory variable considering time delay (maximum time tmax considering time delay and time width Δt of which average value is calculated) Output Time-series data of explanatory variable • Explanatory variable considering time delay: values created as plurality of average values before certain time • Explanatory variable not considering time delay: original value • Time label

The time-series data of the explanatory variables in the output shown in Table 4 includes an explanatory variable considering a time delay, an explanatory variable not considering a time delay, and a time label. The time label indicates a prediction target time, and is set for each time interval of the time width. The first variable creation unit 210 outputs a value created as a plurality of average values before a certain time for an explanatory variable considering a time delay and outputs an original value for an explanatory variable not considering a time delay among the time-series data of the explanatory variables. Here, a method of calculating an explanatory variable considering a time delay will be described. For each of the outside air temperature and the number of people as explanatory variables, the setting time is set to i = 1, 2, ..., n (nΔt ≤ tmax < (n+1) Δt), and an average value of the date and time t-(i-1)Δt to the date and time t-iΔt is created as an explanatory variable considering a time delay for an explanatory variable of the date and time t. By performing this for all i, an average value for each setting time i is obtained. The calculation of the average value does not include a value of the date and time t-(i-1)Δt, but includes a value of the date and time t-iΔt.

Table 5 shows an example of time-series data (learning data) used to create an explanatory variable as an input of the first variable creation unit 210.

TABLE 5 Date and time Weather data: outside air temperature People flow data: number of people Room temperature setting value ... 2020-01-01 09:00:00 10.0 100 20 2020-01-01 09:10:00 10.2 110 20 ...

Table 6 shows an example of the time delay parameter of the combination selected in step S1104 as an input of the first variable creation unit 210. Table 6 shows an example of a combination in which the maximum time tmax is 120 minutes for the outside air temperature and 60 minutes for the number of people, and the time width Δt is 60 minutes for the outside air temperature and 30 minutes for the number of people.

TABLE 6 Date and time Weather data: outside air temperature People flow data: number of people Maximum time tmax considering time delay 120 min 60 min Time width Δt of which average value is calculated 60 min 30 min

Table 7 shows an example of the time-series data of the explanatory variable as an output of the first variable creation unit 210. The example in Table 7 is an example of a case where the outside air temperature and the number of people considering time, and explanatory variables of the room temperature setting value not considering time and a time label are represented with January 1 as the verification day.

TABLE 7 Date and time Weather data: outside air temperature People flow data: number of people Room temperature setting value Time label Average value from 0 to 60 min Average value from 60 to 120 min Average value from 0 to 30 min Average value from 30 to 60 min ... 2020-01-01 9.5 8.6 90 50 20 9:00 09:00:00 9.6 8.7 92 52 20 9:10 2020-01-01 09:10:00 ...

The process of the first variable creation unit 210 has been described above.

In step S1108, the CPU 11 as the parameter determination unit 114 sets a verification day and a training day. Here, for each day of the learning data, an arbitrary day is set to a verification day and other days are set to training days. In the repetitive processing, each day of the learning data is selected once as a verification day, and the processing is repeated until all days are selected as verification days (a determination process in step S1116 that will be described later).

In step S1110, the CPU 11 as the parameter determination unit 114 outputs time-series data of the explanatory variable and the objective variable for the set training day to the first model creation unit 212. For the explanatory variable for the training day, the time-series data of the explanatory variable for the training day among the time-series data of the explanatory variable created in step S1106 may be included in the outputs, and the time-series data on the verification day may be excluded. The same applies to the objective variable.

In step S1112, the CPU 11 as the first model creation unit 212 creates a verification model by using the time-series data of the explanatory variable and the objective variable for the training day.

Here, inputs and outputs of the first model creation unit 212 will be described. The verification model created by the first model creation unit 212 is a model used for cross verification. Hereinafter, a model for parameter determination created by the first model creation unit 212 will be referred to as a verification model, and will be described separately from a prediction model created by the second prediction model.

Table 8 shows inputs and outputs of the first model creation unit 212. The input and output shown in Table 8 are also common to the second model creation unit 222.

TABLE 8 Input Time-series data of explanatory variable and objective variable Output Model learned by using input data

The time-series data of the explanatory variable and the objective variable of the input shown in Table 8 is time-series data on the training day with respect to the first model creation unit 212. That is, the verification model learned by using the time-series data on the training day is output as the output. The first model creation unit 212 creates a model learned by using the time-series data of the explanatory variable and the objective variable. For an algorithm, a regression model that can predict a continuous value from a plurality of explanatory variables may be selected, and for example, the Random Forest disclosed in Reference Literature 1 may be used.

[Reference Literature 1] Breiman, L. (2001). Random Forest. Machine Learning, 45(1), 5-32.

As shown in Table 9, the verification model learned by using the input data in Table 8 is created as a verification model in which the input is an explanatory variable and the output is an objective variable. The objective variable in Table 9 is a prediction result of the hot water use energy, and indicates an output of the verification model for the hot water. In a case where the cold water use energy is also predicted, a verification model for the cold water is also created.

TABLE 9 Explanatory variable Objective variable Date and time Weather data: outside air temperature People flow data: number of people Room temperature setting value Time label Hot water use energy Average value from 0 to 60 min Average value from 60 to 120 min Average value from 0 to 30 min Average value from 30 to 60 min ... 2020-01-01 9.5 8.6 90 50 20 9:00 50 09:00:00 2020-01-01 9.6 8.7 92 52 20 9:10 60 09:10:00 ...

The process of the first model creation unit 212 has been described above.

In step S1114, the CPU 11 as the parameter determination unit 114 calculates a predicted value of the objective variable for the verification day by using the verification model created in step S112.

In step S1116, the CPU 11 as the parameter determination unit 114 calculates an absolute error between the predicted value and the measured value for the verification day.

In step S1118, the CPU 11 as the parameter determination unit 114 determines whether or not all the days of the learning data have been used as the verification days. In a case where it is determined that all the days have been used as verification days, the process proceeds to step S1120. In a case where it is determined that all the days have not been used as verification days, the process returns to step S1108, another verification day is selected, and the process is repeatedly performed.

In step S1120, the CPU 11 as the parameter determination unit 114 calculates a score for the combination selected in step S116 by using an average value of the absolute errors obtained for the respective verification days in step S1104 as a score.

In step S1122, the CPU 11 as the parameter determination unit 114 determines whether or not scores have been calculated for all combinations. In a case where it is determined that the scores have been calculated for all the combinations, the process proceeds to step S1124. In a case where it is determined that the scores have not been calculated for all the combinations, the process returns to step S1104, the next combination is selected, and the process is repeatedly performed.

In step S1124, the CPU 11 as the parameter determination unit 114 determines an optimal time delay parameter for the use energy, and stores the determined time delay parameter in the parameter storage unit 116. The optimal time delay parameter may be a time delay parameter of a combination having the best score among the scores calculated for the respective combinations. Consequently, the time delay parameter for the air conditioning control region that is a processing target selected in step S1102 is determined. The determined optimal time delay parameter is the optimal maximum time and time width of the disclosed technique.

In step S1126, the CPU 11 as the parameter determination unit 114 determines whether the time delay parameter has been determined for all the air conditioning control regions. In a case where it is determined that the time delay parameter has been determined for all the air conditioning control regions, the process is ended. In a case where it is determined that the time delay parameter has not been determined for all the air conditioning control regions, the process returns to step S1102, the next air conditioning control region is selected, and the process is repeatedly performed.

The parameter determination process in step S102 has been described above.

Next, the model learning process in step S104 will be described with reference to the flowchart of FIG. 6. In the model learning process, the prediction model is learned by using the determined time delay parameter. In the model learning process, similarly to the parameter determination process, a prediction model for the use energy that is used for each air conditioning control region may be created.

In step S1200, the CPU 11 as the learning model creation unit 118 acquires learning data and a time delay parameter. The learning data is acquired from the learning data storage unit 112, and the time delay parameter is acquired from the parameter storage unit 116.

In step S1202, the CPU 11 as the learning model creation unit 118 selects an air conditioning control region that is to be a processing target.

In step S1204, the CPU 11 as the second variable creation unit 220 obtains an average value of the time-considered data of the learning data by using the acquired time delay parameter, and creates time-series data of the explanatory variable including the average value. The time-series data of the explanatory variable and the objective variable is output to the second model creation unit 222. The time-series data of the explanatory variable created by the second variable creation unit 220 is different in input from the case of the first variable creation unit 210. The second variable creation unit 220 targets all the time-series data of the learning data, and the first variable creation unit 210 targets the time-series data on the training day. Since a method of creating the time-series data of the explanatory variables is similar to the process of the first variable creation unit 210 in step S1106, the description thereof will be omitted.

In step S1206, the CPU 11 as the second model creation unit 222 generates a prediction model for the use energy for the air conditioning control region that is a processing target by using the time-series data of the explanatory variable and the objective variable, and stores the prediction model in the learning model storage unit 120. Since a method of creating the prediction model is similar to that of the first model creation unit 212, the description thereof will be omitted.

In step S1208, the CPU 11 as the learning model creation unit 118 determines whether or not prediction models have been generated for all the air conditioning control regions. In a case where it is determined that prediction models have been created for all the air conditioning control regions, the process is ended. In a case where it is determined that prediction models have not been created for all the air conditioning control regions, the process returns to step S1202, the next air conditioning control region is selected, and the process is repeatedly performed.

As described above, according to the learning process of the prediction device 100 of the present embodiment, it is possible to predict a required heat quantity for each air conditioning control region and learn the time delay parameter and the prediction model for enabling appropriate heat source control.

In the parameter determination process, trying all combinations of tmax and Δt that are time delay parameters is merely an example. Any method may be used as long as an appropriate time difference in which the influence of the environment outside the building propagates can be obtained for each air conditioning control region. As an example, a range of a setting time may be set in advance for tmax and Δt, and parameter search may be performed through Bayesian optimization or the like within the range. For example, a range of 60 to 240 for tmax and a range of 10 to 60 for Δt are set as setting times.

As a method of considering a time delay, the following is conceivable in addition to the “plurality of average values before the prediction target time” described above. (1) A value at a certain point before the prediction target time is used. For example, a value at 11:00 is used when prediction is performed at 12:00. (2) Values at a plurality of points before s prediction target time are used. For example, a value at 10:00 value and a value at 11:00 are used when prediction is performed at 12:00. In the technique of the present embodiment, as a result of verifying these, “a plurality of average values before the prediction target time” with the highest accuracy is used. A verification result will be described later.

Prediction Process

Next, a configuration of a processing unit related to a prediction process will be described.

The prediction data collecting unit 130 collects prediction data and stores the prediction data in the prediction data storage unit 132. As described above, the prediction data is predicted weather data that is the predicted value of the weather data, predicted people flow data that is a predicted value of the people flow data, and a scheduled setting value of an air conditioner for each air conditioning control region. Table 10 below shows a description of the prediction data to be collected, and Table 11 shows an example of a storage format of the prediction data storage unit 132.

TABLE 10 Data Specific example Forecast value of weather data Outside air temperature forecast value corresponding to air conditioning control region Outside humidity forecast value corresponding to air conditioning control region Solar radiation amount forecast value corresponding to air conditioning control region Wind speed forecast value corresponding to air conditioning control region Predicted value of people flow data Predicted value of number of unique persons present in air conditioning control region Predicted value of stay time (sec) of person present in air conditioning control region *Predictor is prepared Scheduled setting value of air conditioner Room temperature scheduled setting value of air conditioner in air conditioning control region Air volume scheduled setting value of air conditioner in air conditioning control region Supply air temperature scheduled setting value of air conditioner in air conditioning control region Air conditioner operation plan in air conditioning control region

TABLE 11 Date and time Weather data: outside air temperature forecast value People flow data: number-of-people predicted value Room temperature scheduled setting value ... 2020-02-01 09:00:00 10.0 100 20 2020-02-01 09:10:00 10.2 110 20 ...

The prediction data in Table 10 is classified into predicted (forecast) weather data, predicted people flow data, and a scheduled setting value of an air conditioner. Each piece of the data corresponds to the learning data in Table 1. Table 11 is an example of the prediction data stored in the prediction data storage unit 132. Prediction data is stored as time-series data from Feb. 1, 2020. The example of Table 11 shows a case where an outside air temperature forecast value is collected as predicted weather data, a predicted number of people is collected as predicted people flow data, and a room temperature scheduled setting value is collected as a scheduled setting value. In the prediction process, the predicted weather data and the predicted people flow data are time-considered data.

The demand prediction unit 134 predicts a required heat quantity for the use energy for each air conditioning control region by using the prediction data stored in the prediction data storage unit 132 and the stored time delay parameter. The predicted required heat quantity for each air conditioning control region is stored in the prediction result storage unit 136. Consequently, a sum of the required heat quantities predicted for the respective air conditioning control regions is obtained as a prediction result of the heat source demand of the entire building. The prediction data and the time delay parameter are examples of predetermined parameters related to a surrounding environment of the air conditioning control region of the technique of the present disclosure.

Next, actions of the prediction device 100 will be described.

FIG. 7 is a flowchart illustrating a flow of a prediction process by the prediction device 100. The prediction process is performed by the CPU 11 reading the prediction processing program from the ROM 12 or the storage 14, loading the program to the RAM 13, and executing the program.

In step S200, the CPU 11, as the prediction data collecting unit 130, collects the prediction data and stores the same in the prediction data storage unit 132.

In step S202, the CPU 11 as the demand prediction unit 134 selects an air conditioning control region that is to be a prediction target.

In step S204, the CPU 11 as the demand prediction unit 134 acquires a time delay parameter and a prediction model corresponding to the air conditioning control region that is a prediction target. The time delay parameter is acquired from the parameter storage unit 116, and the prediction model is acquired from the learning model storage unit 120.

In step S206, the CPU 11 as the third variable creation unit 230 obtains an average value of time-considered data of the prediction data by using the acquired time delay parameter, and creates time-series data of the explanatory variable in the prediction process, including the average value.

Table 12 shows an example of the time-series data of the explanatory variable in the prediction process, which is an output of the third variable creation unit 230.

TABLE 12 Explanatory variable Date and time Weather data: outside air temperature forecast value People flow data: number-of-people predicted value Room temperature scheduled setting value Time label Average value from 0 to 60 min Average value from 60 to 120 min Average value from 0 to 30 min Average value from 30 to 60 min ... 2020-02-02 9.5 8.6 90 50 20 9:00 09:00:00 2020-02-02 9.6 8.7 92 52 20 9:10 09:10:00 ...

For example, in a case where it is desired to obtain a predicted value of the hot water energy use amount on 2020/2/2, the time-series data of the explanatory variables as shown in Table 12 is created from the time-series data for one day of 2020/2/2 of the prediction data and the time delay parameter for predicting the hot water energy use amount. An example in Table 12 is time-series data of explanatory variables for one day of 2020/2/2. The time-series data of the explanatory variable created as described above is input to the prediction model for predicting the hot water energy use amount, and a predicted value is obtained as an output. The same applies to a case where it is desired to obtain a cold water energy use amount predicted value.

In step S208, the CPU 11 as the demand prediction unit 134 uses the time-series data of the explanatory variable for the air conditioning control region that is a prediction target as an input to the acquired prediction model, and predicts a required heat quantity for the use energy on the basis of an output from the prediction model. The prediction result for the air conditioning control region that is a prediction target is stored in the prediction result storage unit 136.

In step S210, the CPU 11 as the demand prediction unit 134 determines whether or not the required heat quantity has been predicted for all the air conditioning control regions. In a case where it is determined that the required heat quantity has been predicted for all the air conditioning control regions, the process is ended. In a case where it is determined that the required heat quantity has not been predicted for all the air conditioning control regions, the process returns to step S202, the next air conditioning control region is selected, and the process is repeatedly performed.

As for the required heat quantity, a required heat quantity is predicted for each of cold water and hot water as use energy. In this case, an explanatory variable may be created for each type of use energy for each air conditioning control region. An average value for each setting time of the time-considered data is obtained by using the time delay parameter as an explanatory variable. The time delay parameters (the maximum time tmax and the time width Δt) are learned for each type of the use energy. For each air conditioning control region, an explanatory variable for each type of use energy is used as an input to the prediction model, the required heat quantity is predicted, and the heat source demand of the entire building is predicted. The prediction model is also learned to predict a required heat quantity for each type of use energy.

As described above, according to the learning process of the prediction device 100 of the present embodiment, appropriate heat source control can be performed by predicting a required heat quantity for each air conditioning control region.

Verification Result

Verification A and verification B will be described for the effects actually verified by using data regarding an air conditioning control region of a building.

In verification A, an effect in which the setting value of the air conditioner is included in the explanatory variable was checked. The effect in which the setting value of the air conditioner is included in the explanatory variable was verified by comparing with a result of the setting value of the air conditioner not being included in the explanatory variable.

First, explanatory variables as shown in Table 7 were created as explanatory variables including the setting value of the air conditioner. Subsequently, as explanatory variables not including the air conditioner setting value, explanatory variables excluding the “setting value (room temperature) of the air conditioner” therefrom were created. The values in Table 3 were used as time delay parameters.

An absolute error of the predicted value was obtained according to the following methods 1 to 6 for the created two types of explanatory variables. The results are shown in Table 13 and FIG. 8.

1. In the time-series data, any one day is set to a verification day and other days are set to training days.

2. A model learned from data on the training days is created.

3. A predicted value for the verification day is calculated from the created model.

4. An absolute error between the predicted value and a measured value for the verification day is calculated.

5. 1 to 4 are repeated until all days in the data are verification days.

6. An average value of the absolute errors obtained in 4 for all the verification days is set as an absolute error of the model.

TABLE 13 Cold water energy use amount prediction Hot water energy use amount prediction Presence of setting value of air conditioner 43.97 29.84 Absence of setting value of air conditioner 61.14 37.75

In both models, the absolute error was reduced by the setting value of the air conditioner being included in the explanatory variable, and the accuracy was improved.

In verification B, the effect in which an explanatory variable considering a time delay of the outside air temperature in the weather data and the number of people in the people flow data are included was checked. The effect in which the explanatory variable considering the time delay is included was verified in comparison with a result in the case of not considering the time delay. Results of explanatory variables considering other time delays other than the “plurality of average values before the prediction target time” were also compared. The other time delays are a value at a certain point before the prediction target time and values at a plurality of points before the prediction target time.

For comparison, the following four patterns A to D were created as explanatory variables of the outside air temperature in the weather data and the number of people in the people flow data. The other “air conditioner setting value (room temperature)” and “time label” are common.

A is a pattern of a plurality of average values before the prediction target time. This is the pattern of the above embodiment. As inputs, a plurality of average values before the prediction target time were used for the outside air temperature, and is a plurality of average values before the prediction target time were used for the number of people. The values in Table 3 were used as time delay parameters.

B is a pattern that does not consider a time delay. As inputs, a value at the prediction target time was used for the outside air temperature, and a value at the prediction target time was used for the number of people.

C is a pattern of a value at a certain point before the prediction target time. A value before the prediction target time was used for the outside air temperature, and a value before the prediction target time was used for the number of people. As to how many minutes before a value is to be used, an optimum value is employed similarly to the cross verification method shown in the parameter determination unit 114.

D is a pattern of values at a plurality of points before the prediction target time. Values at a plurality of points before the prediction target time were used for the outside air temperature, and values at a plurality of points before the prediction target time were used for the number of people. As to how many minutes before a value is to be used, an optimum value is employed similarly to the cross verification method shown in the parameter determination unit 114.

An absolute error was obtained for the created four types of explanatory variables according to the same method as in verification A. The results are shown in Table 14 and FIG. 9.

TABLE 14 Cold water energy use amount prediction Hot water energy use amount prediction A: plurality of average values before prediction target time 43.97 29.84 B: not consider time delay 51.10 31.17 C: value at certain point before prediction target time 47.63 30.42 D: values at plurality of points before prediction target time 44.89 29.87

It was confirmed that the model of the pattern B that does not consider the time delay has the lowest accuracy, and the accuracy is improved by considering the time delay in the outside air temperature and the people flow. Above all, it was confirmed that the pattern A in which “the plurality of average values before the prediction target time” is the most accurate.

As described above, in the method of the present embodiment, the people flow and the setting value of the air conditioner are included in the explanatory variables for the prediction of a required heat quantity of the air conditioning control region. This makes it possible to improve the accuracy of prediction of the heat source demand in a case where there are a plurality of patterns of stay and movement of people in the air conditioning control region and in a case where a frequency of change in the setting value of the air conditioner is high even on a day when environmental data such as the outside air temperature and the outside humidity related to the surrounding environment of the building is the same.

The outside air temperature in the weather data, the number of people in the people flow data, and the like, are handled as time-considered data in which values past the prediction time are included in the explanatory variables in consideration of the influence of the time delay on a space. Consequently, it is possible to improve the accuracy of the heat demand prediction by taking into consideration the influence of the time delay caused by the outside air temperature and the people flow different for each air conditioning control region on the space.

As the time delay parameter, various time delay patterns are verified by using a machine learning method to obtain a time delay parameter with the smallest prediction error. By using the machine learning model while avoiding the problem of the physical model, it is possible to expand application destination and improve the accuracy since it is not necessary to determine a parameter. The problem of the physical model is a method of obtaining a time delay from a physical model on the basis of a parameter of a space such as specific heat or air density of the space, a heat quantity of a human body, and the like, and is a problem that accuracy decreases in a case where it is difficult to determine a parameter and the parameter deviates from the real world.

The learning processing program or the prediction processing program executed by the CPU reading software (program) in the above embodiment may be executed by various processors other than the CPU. Examples of the processor in this case include a programmable logic device (PLD) in which a circuit configuration can be changed after manufacturing such as a field-programmable gate array (FPGA), and a dedicated electric circuit that is a processor having a circuit configuration exclusively designed for performing specific processing, such as an application specific integrated circuit (ASIC). The learning processing program or the prediction processing program may be executed by one of these various processors, or may be executed by a combination of two or more processors of the same type or different types (for example, a plurality of FPGAs or a combination of a CPU and an FPGA). The hardware structure of these various processors is, more specifically, an electric circuit in which circuit elements such as semiconductor elements are combined.

In the above embodiment, the aspect in which the learning processing program or the prediction processing program is stored (installed) in advance in the storage 14 has been described, but the present invention is not limited thereto. The program may be provided in a form stored in a non-transitory storage medium such as a compact disk read only memory (CD-ROM), a digital versatile disk read only memory (DVD-ROM), and a Universal Serial Bus (USB) memory. The program may be downloaded from an external device via a network.

With regard to the above embodiment, the following supplementary notes are further disclosed.

Supplement Note 1

A prediction device that performs heat source demand prediction in a space having a predetermined air conditioning control region, the prediction device including:

  • a memory; and
  • at least one processor connected to the memory, in which
  • the processor is configured to:
    • predict a required heat quantity in the air conditioning control region by using a predetermined parameter related to a surrounding environment of the air conditioning control region and a setting value of an air conditioner set in the air conditioning control region as inputs, and
    • predict a heat source demand of the entire space from the predicted required heat quantity for each air conditioning control region.

Supplement Note 2

A non-transitory storage medium storing a prediction program that is executable by a computer to execute a prediction process and performs heat source demand prediction in a space having a predetermined air conditioning control region, the prediction program causing the computer to execute:

  • predicting a required heat quantity in the air conditioning control region by using a predetermined parameter related to a surrounding environment of the air conditioning control region and a setting value of an air conditioner set in the air conditioning control region as inputs; and
  • predict a heat source demand of the entire space from the predicted required heat quantity for each air conditioning control region.

Claims

1. A prediction method of performing heat source demand prediction in a space having a predetermined air conditioning control region, the prediction method causing a computer to execute processes of:

predicting a required heat quantity in the air conditioning control region by using a predetermined parameter related to a surrounding environment of the air conditioning control region and a setting value of an air conditioner set in the air conditioning control region as inputs; and
predicting a heat source demand of the entire space from the predicted required heat quantity for each air conditioning control region.

2. The prediction method according to claim 1, wherein

the required heat quantity is predicted by using at least one of weather data related to a periphery of the air conditioning control region and people flow data related to a person present in the air conditioning control region as the predetermined parameter, the weather data and the people flow data being time-considered data in which an influence of a change in time is considered.

3. The prediction method according to claim 2, wherein

a predetermined calculated value of the time-considered data for each air conditioning control region is obtained by using a maximum time that is a consideration target before a prediction target time and a time width set between the maximum time and the prediction target time, learned in advance for the time-considered data, and
the required heat quantity is predicted by using the predetermined calculated value and the setting value as inputs.

4. The prediction method according to claim 3, wherein

the maximum time and the time width are learned according to a predetermined estimation method by using a setting time for the maximum time and a setting time for the time width that are determined for each piece of the time-considered data.

5. The prediction method according to claim 4, wherein

the maximum time and the time width are learned by, as the estimation method, performing cross verification for each combination of the setting time for the maximum time and the setting time for the time width, and estimating the maximum time and the time width that are optimal among the combinations.

6. The prediction method according to claim 3, wherein

the required heat quantity is a required heat quantity for each of two or more types of use energy,
the predetermined calculated value of the time-considered data is obtained for each type of the use energy by using the maximum time and the time width obtained for the use energy for each air conditioning control region, and
a heat source demand of the entire space is predicted by predicting the required heat quantity by using the predetermined calculated value for each type of the use energy and the setting values as inputs for each air conditioning control region.

7. A prediction device that performs heat source demand prediction in a space having a predetermined air conditioning control region, the prediction device comprising:

a demand prediction unit that predicts a required heat quantity in the air conditioning control region by using a predetermined parameter related to a surrounding environment of the air conditioning control region and a setting value of an air conditioner set in the air conditioning control region as inputs, and predicts a heat source demand of the entire space from the predicted required heat quantity for each air conditioning control region.

8. A non-transitory, computer-readable storage medium storing a prediction program for performing heat source demand prediction in a space having a predetermined air conditioning control region, the prediction program causing a computer to execute processes of:

predicting a required heat quantity in the air conditioning control region by using a predetermined parameter related to a surrounding environment of the air conditioning control region and a setting value of an air conditioner set in the air conditioning control region as inputs; and
predicting a heat source demand of the entire space from the predicted required heat quantity for each air conditioning control region.
Patent History
Publication number: 20230349579
Type: Application
Filed: Sep 18, 2020
Publication Date: Nov 2, 2023
Applicant: NIPPON TELEGRAPH AND TELEPHONE CORPORATION (Tokyo)
Inventors: Naoki ARAI (Tokyo), Kazuaki OBANA (Tokyo), Keisuke TSUNODA (Tokyo), Sotaro MAEJIMA (Tokyo), Midori KODAMA (Tokyo), Nobuhiko MATSUURA (Tokyo)
Application Number: 18/026,341
Classifications
International Classification: F24F 11/64 (20060101);