POWER LINE ISSUE DIAGNOSTIC METHODS AND APPARATUS USING DISTRIBUTED ANALYTICS

Abstract

Apparatus, system, and method for diagnosing status of electrical line performance by receiving and analyzing a plurality of electrical line data from a plurality of lines includes a collection of internet of things sensors, a communication network, local firmware boards, a data hub, a computation engine, a machine learning engine, and an internet of things server operatively connected to the machine learning engine.

Description

CROSS-REFERENCE TO RELATED PATENT APPLICATIONS

This patent application is a 35 USC 120 division of co-pending U.S. patent application Ser. No. 14/956,403, entitled “Distributed Internet of Things Based Sensor Analytics for Power Line Diagnosis,” filed 2 Dec. 2015 and published as US 2017/0160328 on 8 Jun. 2017; the priority of the '403 application is claimed under 35 USC 120.

This patent application is also a 35 USC 120 continuation-in-part of U.S. patent application Ser. No. 16/253,462, entitled “Real Time Machine Learning Based on Predictive and Preventive Maintenance of a Vacuum Pump,” filed 22 Jan. 2019; the priority of the '462 application is claimed under 35 USC 120. The '462 application is a continuation of U.S. patent application Ser. No. 14/628,322, noted immediately below.

This patent application also claims the priority of U.S. patent application Ser. No. 14/628,322 entitled “Real Time Machine Learning Based on Predictive and Preventive Maintenance of a Vacuum Pump,” filed 23 Feb. 2015; the priority of the '322 application is claimed under 35 USC 120, through the '462 application.

FEDERAL FUNDING STATEMENT

Not Applicable—this invention was conceived and developed entirely using private source funding; this patent application is being filed and paid for entirely by private source funding.

INCORPORATION BY REFERENCE

Applicant hereby incorporates by reference the disclosures of the following United States patent publications: US 2019/0154032; US 2019/0154469; US 2019/0113280; US 2019/0191287; US 2019/0113281; US 2017/0011298; US 2016/0245279; US 2017/0051978; US 2016/0313216; US 2016/0291552; US 2016/0245686; US 2017/0178030; US 2018/0077522; US 2017/0160328 and US 2016/0245765.

Applicant hereby incorporates by reference the disclosures of the following U.S. Pat. Nos. 9,826,338 and 9,823,289.

FIELD OF THE INVENTION

This invention generally relates to the internet of things and more specifically to internet of things enabled distributed computing and fault diagnosis in power lines involving power quality parameters such as sag, swell, flickering, surge, and harmonic distortion.

DESCRIPTION OF THE INVENTION

The internet of things is a network of uniquely-identifiable “things” that are able to communicate data pertaining thereto over the Internet, where the communicated data form a basis for manipulating operation of the “things”. The “thing” in the internet of things can be virtually anything that is animant or, which moves or operates (such as a machine), or which changes state (such as a plant). For example, the “thing” could be a person with a heart rate monitor implant, a farm animal with a biochip transponder, an automobile having built-in sensors to alert the driver when tire pressure is low, or any other natural or man-made entity that can be assigned a unique IP address and provided with the ability to transfer data over a communication network, which network is typically the Internet. If all the entities in an internet of things network are machines, then the internet of things is referred to as a “machine to machine” internet of things.

An entity becomes a “thing” of an machine to machine internet of things when the entity is equipped with one or more sensors capable of (i) capturing one or more types of data pertaining to the “thing”, (ii) segregating the data if applicable, (iii) selectively communicating each segregation of data to one or more fellow “things”, and (iv) receiving one or more control commands from one or more fellow “things”. Control commands for one “thing” are based on data received from the fellow “things”. Executing the control commands results in the manipulation or management of the operation of the entity which is the receiving “thing”. In an internet of things-enabled system, the “things” may manage themselves without any human intervention.

U.S. Pat. No. 9,052,216 B2 discloses an energy measurement system which measures electrical parameters, such as line-to-line voltage/current, line to neutral voltage/current, total apparent power, reactive power, active power, fundamental and harmonic total energy per phase, fundamental and harmonic reactive energy per phase, active energy per harmonic frequency per phase, reactive energy per harmonic frequency per phase, and fundamental and harmonic active energy per phase.

WIPO publication WO2014089567A2 discloses automated monitoring of various sensors including sensors that measure power, voltage, current, temperature, and humidity of power sources as well as notification triggers and alarms sent to a cellular phone.

Chinese patent 203,588,054U discloses a wireless network sensor monitoring system used in a power environment based on the internet of things.

Chinese patent 102,539,911A discloses smart metering systems, large master systems, digital substations, and small “smart metering” sensors operating in an “internet of things” environment.

U.S. patent publication 2012/0213098 A1 discloses use of an internet of things analyzer measuring voltage, current and resistance as a multi-meter.

U.S. Pat. No. 8,447,541 B2 discloses energy monitoring devices communicating with energy aware appliances having an embedded energy monitor and connected network equipment, such as a router or a hub and a server.

None of this prior art discloses distributed computing, power factor calculations and use of “big data” technologies to diagnose power quality issues of a power line, to reduce the cost of electronics measuring those issues and offering scalability for a large number of measuring points.

Moreover, the aforementioned prior art fails to address measurement of electrical parameters affecting power quality of power lines in large numbers over a large area. An organization with multiple locations around the world with multiple electrical lines to be monitored may be too huge to handle by the aforementioned prior art. The aforementioned prior art fails to address scale of data, frequency of data, and calculation capabilities residing in a single location.

There exists the need for a solution to the aforementioned problems. The instant invention addresses those problems.

SUMMARY OF THE INVENTION

This invention embraces methods and apparatus for distributed power line diagnosis and in one of its aspects provides a method of predicting electrical issues by collecting information through a processor, where the information is one or more electrical line readings produced by one or more internet of things sensors, and transmitting the collected information over a communication network together with one or more pre-selected electrical line readings to a machine learning engine. In this aspect of the invention, the method further includes visualizing one or more electrical line issues through processor based analysis using a big data engine and indicating detected electrical line issues through a user dynamic interface. An alarm is set through a processor for indicating the predicted electrical line issues.

In another aspect of the invention, a distributed power line diagnosis system includes firmware receiving a plurality of electrical line data over a wide ranging communication network, a real time data processing system communicating with a plurality of distributed databases, a local firmware board, a data hub, an internet of things server, a multi-classification machine learning engine operatively associated with the internet of things server, a display module associated with one or more processors, a user interface, and an alarm module. A power line issue is mapped into a depiction on a user interface. The power line issue is determined based on a computation performed at one or more of the local firmware board, the data hub, and the internet of things server. The alarm module raises an alarm when a pre-set condition is met or breached. Further, the alarm module is operatively associated with the multi-classification machine learning engine.

In yet another one of its aspects this invention provides a method of distributed power line diagnosis which includes receiving one or more electrical readings through a firmware computation engine and transmitting output of the firmware computation engine through a communication network to a data hub computation engine. Output of the data hub computation engine is transmitted through the communication network to a big data server.

The methods disclosed herein are desirably implemented in the form of a machine-readable medium embodying a set of instructions that, when executed by a machine, cause the machine to perform one or more of the operations disclosed herein; however the methods of the invention as disclosed herein may be implemented in other manners, known to those of skill in the electrical power line diagnosis art. Other features will be apparent from the accompanying drawings and from the detailed description that follows.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a diagrammatic representation of a distributed power line diagnosis system in accordance with the invention.

FIG. 2 is a diagrammatic representation of a data processing system for processing sets of instructions for performing the methods of the invention.

FIG. 3 is a process flow diagram detailing a method of predicting electrical line issues according to the invention.

FIG. 4 is a circular gauge for depicting a condition of an electrical line.

FIG. 5 is a process flow diagram detailing a method for distributed power line diagnosis according to the invention.

FIG. 6 is a front view of a portable electronic device displaying results of a power line diagnosis according to the invention.

FIG. 7 is a bar graph schematically depicting results of instantaneous measurement of active power and reactive power during three phase power transmission.

FIG. 8 is a collection of three bar graphs schematically depicting results of three separate instantaneous measurements of power factor during three phase power transmission.

FIG. 9 is a collection of three bar graphs schematically depicting average voltage, average current and average power over a period of days during three phase power transmission.

FIG. 10 is a pair of bar graphs schematically depicted average energy in kilowatt hours, for one week and for one day of the week, showing a breakdown as between active power and reactive power.

FIG. 11 is a schematic representation of a circular gauge presenting six (6) different parameters involved in the electrical power line predictive analytics and distributed diagnoses aspect of the invention.

DETAILED DESCRIPTION OF THE INVENTION

As used in this application, and as generally used by those of skill in the art, “big data” is a term used to refer to large data sets. These data sets are so large and complex that traditional computers and data processing systems are inadequate to handle these data sets.

With reference to FIG. 1 of the drawings, it is manifestly clear that this invention provides apparatus for diagnosing status of electrical line performance by receiving and analyzing a plurality of electrical line data from a plurality of electrical lines. In that regard, the invention proceeds with a collection of machine wearable internet of things sensors, one sensor operatively connected to each of the electrical lines of interest. Each sensor senses one or more parameters respecting the electrical line, including current, voltage, power factor, harmonic distortion, swell, surge, sag, active power, reactive power, and frequency of the electrical power carried by the electrical line of interest. The apparatus of the invention yet further includes a communication network 102, a collection of local firmware boards 106, 108, one for each of the internet of things sensors, with each local firmware board being operatively connected to an associated internet of things sensor, for receiving the sensed electrical line parameter data from the associated internet of things sensor.

The apparatus further includes a data hub 114, operatively connected to the local firmware boards 106, 108, and receiving therefrom data sensed by the machine wearable internet of things sensors. The apparatus still further includes a computation engine 118 that is operatively connected to local firmware boards 106, 108, data hub 114, and an internet of things server 112, for classifying different types of faults in the electrical power line as evidenced by the data sensed and supplied by the sensors for the electrical power line data, using quadratic hyperplanes and transformed variable space techniques at local firmware boards 106, 108, or data hub 114, or, optionally, at internet of things server 112, and computing at the local firmware boards 106, 108 and/or at the data hub 114, and/or at the internet of things server 112, values of parameters for a selected type of fault, such as sag, which fault is pre-defined by a user.

The apparatus still yet further includes a machine learning engine 104. Internet of things server 112 is operatively connected to machine learning engine 104 and serves to analyze the sensed electrical line data by comparing the sensed data to prior condition data indicative of acceptable operation, and raising an alarm if the sensed data deviates from the prior condition data, which is indicative of acceptable operation, by more than a pre-selected amount. The apparatus optionally includes a gauge for displaying a visual indication of the performance status of one or more of the selected parameters, with the gauge accepting user-based intuition about the parameter state when displaying visual indication; specifically, the user can place a marker or other indicia on the gauge to mark a level of a displayed parameter as surprising or as indicating that an alarm should be sounded or other action taken.

It is still further evident from the drawings, particularly from FIG. 1, that the invention in one of its aspects, embraces a method for determining status of electrical line performance where the method includes the step of positioning a plurality of machine wearable internet of things sensors in operative communication with a plurality of electrical lines, with one internet of things sensor for each line. The method further proceeds with sensing electrical line data, using the collection of internet of things sensors, sensing one or more electrical parameters including current, voltage, power factor, harmonic distortion, swell, surge, sag, active power, reactive power, and frequency of the electrical power carried by the associated electrical line. The method yet further proceeds by providing a collection of local firmware boards 106, 108, one board for each of the internet of things sensors, with each board being operatively connected to the internet of things sensors, with the boards serving to receive sensed electrical line data from the associated internet of things sensors.

The method yet further proceeds by receiving from the local firmware boards 106, 108 the electrical line data sensed by the internet of things sensors on a data hub 114 operatively connected to the local firmware boards 106, 108, providing a cloud-based internet of things server 112 and computation engine 118 operatively connected to the data hub 114, classifying different fault types in the electrical line data using quadratic hyperplane techniques in transformed variable space at the local firmware boards 106, 108, at the data hub 114, and/or at the internet of things server 112, and computing at the local firmware boards 106, 108 and/or at the data hub 114, and/or at the internet of things server 112 selected data parameters of the electrical power line of interest, again preferably using quadratic hyperplanes in transformed variable space.

In this method aspect of the invention, the method proceeds with analyzing the sensed electrical data by comparing the sensed data to pre-selected parameter values of prior data established using quadratic hyperplanes in transformed variable space, which values are indicative of acceptable operation. Optionally, the method proceeds with raising an alarm if the sensed data deviates from the values of the prior acceptable operation data by more than a pre-selected amount and further by optionally displaying a visual indication of the performance state of the selected ones of the parameters on a gauge, which optionally accepts user-based intuition about the parameter state to alter the displayed visual indication.

It is further apparent from the drawings and particularly from FIG. 1 that, in another one of its aspects the invention provides a method for providing predicted electrical line issues, including phase current imbalance or phase current harmonics to a high number of end users having interest in the reliability of the line. The method proceeds with providing a single board computer processor, connecting an electrical line to be analyzed for the presence or absence of phase current imbalance and phase current harmonics, to the single board computer processor. The invention then proceeds by using the single board computer processor to collect and extract metadata indicative of phase current imbalance or phase current harmonics from any high voltage frequency current or voltage present in the electrical line. The invention further proceeds with processing the metadata using either a machine learning support vector machine or a rule engine equipped with a base-line data set. The method further proceeds with providing the processed metadata to a big data cloud data service and thereafter making the results of the metadata processing available to a high number of end users communicating with the big data cloud service via the Internet, desirably using hand-held electronic devices such as cellular telephones and tablets.

Still in referring to the drawings, FIG. 1 is a system diagram of a distributed power line diagnosis system 100 in accordance with the invention. Distributed power line diagnosis system 100 includes a communication network 102, a machine learning engine 104, at least one and desirably a plurality of local firmware boards, a first one of which has been designated “Local Firmware Board 1” and numbered 106, and another one of which has been designated “Local Firmware Board N” and numbered 108. Distributed power line diagnosis system 100 further includes a computer database 110, an internet of things server 112, a data hub 114, a mobile application 116, and a computation engine 118, all as shown in FIG. 1.

Distributed power line diagnosis system 100 desirably includes firmware for receiving electrical line data over a communication network 102. The firmware is operatively associated with one of the firmware boards. The local firmware board in turn is operatively associated with computation engine 118.

A real time data processing system is operatively associated with the local firmware boards, such as local firmware board designated 106. The computation engine 118, being operatively associated with the local firmware boards via communications network 102, transmits computation data from a local firmware board 106, or 108, etc., to data hub 114 via communications network 102. Data hub 114 in turn is operatively associated with computation engine 118 via communications network 102. Computation engine 118 transmits computations, such as those performed by one or more of the local firmware boards 106, 108, etc., and stored on data hub 114, to internet of things server 112 over communication network 102. Internet of things server 112 computes further data, using data and computations received from the data hub, to identify and/or quantify and/or predict and/or analyze a power line issue. Internet of things server 112 computes and determines parameters to be shown on a user display portion of a mobile application device, preferably a cellular telephone, based on the operative association of server 112 with computation engine 118 and multi-classification machine learning engine 104. Internet of things server 112 may optionally be operatively associated with one or more servers that in turn are operatively associated with geographically widely distributed computers and may also be operatively associated with the “cloud.” The cloud is desirably operatively associated with and accessed by communications network 102.

Multi-classification machine learning engine 104 is operatively associated with internet of things server 112. A display module, not shown in FIG. 1, is operatively associated with either a processor portion of communication engine 118, or a processor portion of internet of things server 112, or a processor portion of machine learning engine 104, and includes a user interface. The processor, selected from among those of internet of things server 112, machine learning engine 104 and computation engine 118, and the user interface are all operatively associated with mobile application 116.

A computed or detected power line issue is mapped into a visual depiction on a video screen portion of the user interface, which is not illustrated in the drawings.

An alarm module, not illustrated in FIG. 1, is optionally provided to raise an alarm when a predetermined condition, in a power line which is under analysis by the distributed power line diagnosis system illustrated in FIG. 1, is reached. The condition is selected for one or more of the art-recognized power line issues, such as sag, defining an electrical line fault or an electrical line failure in an extreme case. The alarm module is operatively associated with multi-classification machine learning engine 104.

Local Firmware boards 1 to N are operatively associated with electrical power lines 1 to N, which are not illustrated in FIG. 1.

One of the power line analytics issues addressed by the invention is effective visualization of the processed results and/or the effectiveness of the alarm system. In one approach, the inventive power line analytics results are mapped into a simple “circular gauge” such as those shown in FIGS. 4 and 11, having a normalized scale of 0-100. A user can set the scale for setting up an alarm and scaling the predictive maintenance results respecting the electrical line issues. In this way, complex results of big data internet of things analytics for electrical line faults are visible to human operators when applying the techniques disclosed herein. The results of the analytics may also be visualized through a web-based application resident on the computation engine 118 or the internet of things server 112, or the machine learning engine 104, where the web-based application displays output from the analytics on a printer or a video screen connected to one of these devices. The printer or video screen are not illustrated in the drawings. Output of the analytics may also be viewed by an operator on the mobile application, which is desirably either a portable tablet or a cellular telephone. The web-based application is optionally operatively associated with a personal computer via communications network 102.

In one preferred practice of the invention, a sensor is used to monitor power quality issues in a factory or other facility having multiple electrically-powered machines. Parameters that are monitored and/or measured include current, for example current ranging between 0 and 100 amps, voltage, ranging for example between 0 and 600 volts, power factor ranging for example from 0 to 1.0, and either lag or lead of active power respecting reactive power, harmonic distortion, measured in percentage, swell, measured in percent of normal, surge, measured in volts, amps, or watts, sag, measured in percent of normal, active power, measured in kilowatts, reactive power, measured in kilowatts, and frequency, typically ranging from between 50 and 60 hertz. Data from the machine wearable sensors depicted in FIG. 1 is fed to a cloud server via the data hub 114. Data hub 114 collects and analyzes and stores the sensor data using big data technology.

The data collected at the sensors operatively associated with the electrical power lines are often extremely large and complex collections of data. These data are collected onto a big data server such as the internet of things server 112, which connects to the cloud via a communications network 102. The big data server such as internet of things server 112 communicates with servers that are operatively associated with and most desirably are a part of the internet of things sensors depicted schematically in FIG. 1.

One of the issues involved in power analytics of the type implemented and effectuated in the course of practice of this invention is effective visualization of the results of the analytics and applying those results to an alarm system to alert users. In one approach implemented by this invention, results of the analytics are mapped onto a circular gauge having a normalized scale from 0 to 100. With the gauge, an exemplary one of which is illustrated in FIG. 4, the user can set a scale for setting an alarm with respect to the particular parameter for which the gauge is receiving data. The user may then scale up a predictive maintenance figure into the gauge to establish an area of fault or even failure as defined by the user, such as the shaded area between the numbers “40” and “100” in FIG. 4. Accordingly, the invention facilitates display of the complex results of the big data internet of things analytics associated with electrical line faults, which results are produced by the analytics and application of the computational techniques as disclosed herein.

In the course of practice of the power line diagnostics of the invention, data for a particular power line under analysis is received from at least one of the machine wearable sensors as depicted schematically in FIG. 1. Each sensor is associated with one or more electrical power lines under analysis. The electrical power lines are not illustrated in FIG. 1. The association between the sensor and the electrical power line may be by direct physical connection or by communicative association, namely wireless connection of the sensor to the electrical power line.

Communication network 102 illustrated in FIG. 1 is desirably a WiFi, Internet-based communication network. However, it is also within the scope of the invention for communication network 102 to be a “2G” second generation wireless telephone technology network, or a “3G” third generation wireless telephone technology network, or a “4G” wireless telephone technology and mobile communication network, or even a “5G” fifth generation wireless telephone and mobile communication technology network. Communication network 102 may utilize packet oriented mobile data services that are available on both 2G and 3G cellular communications systems. If communication network 102 is implemented as a WiFi Internet-based communication system, it is within the scope of the invention to utilize the Edge proprietary communication service.

Machine learning engine 104 includes a machine learning algorithm; suitable algorithms are known to those of skill in the art.

The results of the analytics of the distributed power line diagnosis system 100 may be depicted on a circular gauge such as that illustrated in FIG. 4, which represents a user interface on which results of the analytics are displayed. The gauge illustrated in FIG. 4 desirably includes color schemes such as red, yellow, and green, where red indicates an alarming state of an electrical power line under analysis, yellow may indicate an impending issue with the electrical power line under analysis, and green indicates that there are no issues with the electrical power line under analysis. In implementation of the power line diagnostics, the machine learning engine 104 and the computer for which the computer database is numbered 110, provide alarms when a power line fault or failure is detected by the power line diagnostics of the invention.

In the course of practice of one of the methods of the invention, the method being directed to predicting electrical line issues, is initiated by collecting data associated with one or more electrical power lines from at least one internet of things sensor of the type illustrated schematically in FIG. 1, where the sensor includes computing functionality for performing local computation, namely at the sensor, of time series data of electrical parameters, the results of the computations are then transmitted over a communications network. These computational data are preferably collected over a finite time period and transmitted to the machine learning engine 104, which is associated with computer database 110. Both real time power line parameter data and historical power line parameter data are included and stored within computer database 110. Computation engine 118 analyzes the collected power line real time data and compares it to the historical data. Computation engine 118 performs analysis, utilizing machine learning engine 104 if necessary, and internet of things server 112 if necessary, to provide results of the analysis through a user interface such as the gauge illustrated in FIG. 4. An alarm may be actuated if one of the electrical line issue parameters is shown by the analysis to be outside of an acceptable range.

In another specific implementation of the analytics of the invention, electrical power line data is harvested from electrical lines that are numbered 1 through “N;” that electrical line data is collected onto local firmware boards numbered 1 through “N” as illustrated in FIG. 1. Local firmware boards 1 through “N” transmit computed values of the data from their locations at which the local firmware boards reside, with the computed values of the data being transmitted over communication network 102 and collected on data hub 114 for further computation and analysis by computation engine 118. The data collected and housed on data hub 114 may be from multiple locations, located all over the world. The data computed by computation engine 118 is computed according to predefined criteria defined by an operator. Data hub 114 may be provided as a single data hub or there may be a plurality of data hubs 114. In any event, the data from data hub 114 in computed form is provided to internet of things server 112. Internet of things server 112 and associated machine learning engine 104 analyze the data to determine whether there is an issue with one or more of the electrical lines supplying the data to local firmware boards 1 through “N”. In the event an electrical line issue is determined, such as excessive voltage or sag, an alarm is raised over communication network 110 and supplied to mobile application 116. The communication to mobile application 116 via communication network 102 may be by any of the aforementioned communication protocols, or by short message service or email, or a combination thereof. When a predefined condition for one of the electrical lines under analysis is met, an alarm is sounded and displayed on the mobile application 116.

It is further within the scope of the invention to predict electrical power line issues by collecting one or more electrical power line readings from one or more of the internet of things sensors through a processor and transmitting the collected data readings over a communication network such as communication network 102 while also sending the collected electrical line readings to machine learning engine 104 via communication network 102. In such case, the method proceeds with analyzing the electrical power line data for determining electrical line issues using a processor based on analysis performed through a big data engine, whereupon the electrical line issues found by the big data engine are displayed on a dynamic user interface. The dynamic user interface may be a predictive maintenance circular gauge of the type illustrated in FIG. 4. In such case, an alarm is preferably sent through the processor, indicating the electrical line issues found by the analysis.

Machine learning engine 104 includes a machine learning algorithm that is preferably imbedded within the machine learning engine. Machine learning engine 104 can receive electrical power line data directly from the internet of things sensors with the data being provided over communications network 102. Machine learning engine 104 processes the received data to recognize a pattern for, and a deviation from, parameters of interest and issues alarm and control commands for action by users of the system, pertaining to the electrical line of interest for which the deviation from the pattern was detected. The alarm and control commands are preferably sent via communications network 102.

The method of predicting an electrical issue includes collecting one or more electrical line readings from one or more internet of things sensors through a processor, transmitting the readings over a communication network, and also sending the readings to a machine learning engine. Further, the method preferably includes visualizing one or more of the electrical line issues through a processor, based on analysis through a big data engine, and indicating the one or more electrical line issues through a visible user interface. The user interface is desirably a predictive maintenance circular gauge. An alarm is desirably set, through a processor, for the one or more electrical line issues.

The machine learning engine is operatively associated with a machine learning algorithm. The machine learning engine receives electrical line data from one or more sensors. The machine learning engine processes the received data to recognize a pattern and a deviation therefrom to issue alarm and control commands pertaining to the electrical line associated with the communications network.

Further, the machine learning engine is preferably operatively associated with a multi-classification engine, such as an oblique and/or support vector machine. The support vector machine includes supervised learning models and associated learning algorithms that analyze data and recognize patterns. The supervised learning models use classification and regression analysis.

The steps taken by the multi-classification engine preferably include data transformation to achieve maximum separation among fault types. The data transformation leads to more accurate multi-classifications, for example, linear discriminant functions. Further, nonlinear hyper plane fitting is preferably performed to classify different fault types. Quadratic hyper planes are desirably used in transformed variable space to develop a measure of degree of fault based on a machine learning multi-fault classification approach. The intensity of fault is calculated as a later probability of a fault type. The degree of fault information is desirably mapped onto a circular gauge, such as shown in FIG. 11. Different fault type posterior probabilities are most desirably combined and provided on a circular gauge representation. User calibration of the circular gauge enables inclusion of user intuition about the machine state in the analytics process. The multi-classification optionally ceases when the user agrees with the circular gauge.

In one embodiment of the apparatus aspect of the invention a distributed power line diagnosis system includes of one or more firmware computation engines receiving a plurality of electrical line data over a communication network, a real time data processing system associated with distributed databases, a local firmware board, a data hub, an internet of things server, a multi-classification machine learning engine associated with the internet of things server, a display module associated with one or more processors, optionally a user interface and also optionally an alarm module.

Power line issues are mapped for depiction on the user interface. The power line issue is determined based on computations at one or more of the local firmware boards, the data hub and the internet of things server. The alarm module raises an alarm when a pre-set condition is breached. The alarm module is operatively associated with the multi-classification machine learning engine.

The method of a distributed power line diagnosis includes receiving one or more electrical readings at a firmware board computation engine with output of the firmware board computation engine being transmitted through a communication network to a data hub computation engine. Output of the data hub computation engine is transmitted through the communication network to a big data server. One or more electrical line issues are visualized by an operator based on analysis by the big data server; the electrical line issues are visibly indicated through a dynamic user interface. An alarm may be raised for the one or more electrical line issues.

The mobile application 112 may receive and/or display i_rms (root mean square value of current), v_rms (root mean square value of voltage), pf (power factor), active power, reactive power and frequency for each line. Further mobile application 112 preferably displays alarm thresholds using a circular gauge.

A mobile application may be used to display historical data associated with a machine and/or an electrical line for a time such as, three months, and metadata for a shorter time, such as twenty-four hours.

A mobile application may be used to allow a user to select a filter window, such as the last twenty-four hours, between any two dates, last week, last month, the last three months, and so on. The selected window allows the user to view a daily maximum power/minimum power chart, average power usage by hour, by day, by month and time of the day when power peaks. Further, the user may use the mobile application to view and compare average and peak power usage between machines.

The distributed power line diagnosis system of the invention preferably utilizes a multi-layer big data gauge based on big data visualization to simplify issues and alarms associated with an electrical power line, such as, active power, reactive power, voltage, current, power factor, sag, swell, surge, harmonic distortion, etc.

The distributed power line diagnosis system preferably includes at least two layers, the first or front layer being a gauge (single or multi-parametric or multi-dimensional) and the second layer being analytical. A user sets an alarm for electrical line issues such as swell, etc. based on direct rules and/or by using a multi-classification machine learning algorithm using a base-line calibration method.

FIG. 2 is a diagrammatic representation of a data processing system for processing instructions to perform any of the methodologies herein. FIG. 2 includes a diagrammatic representation of machine in an exemplary form of a computer system 226 within which a set of instructions, for causing the machine to perform any one or more of the methodologies discussed herein, can be executed. The machine may operate as a standalone device and/or may be connected and networked to other machines.

In a networked deployment, the machine preferably operates in the capacity of a server or equally preferably as a client machine in server-client network environment or as a peer machine in a peer-to-peer, distributed network environment. The machine may be a personal computer, a tablet, a personal digital assistant, a cellular telephone, a web appliance, a network router, a switch and/or bridge, an embedded system and/or any other machine capable of executing a set of instructions (sequential and/or otherwise) that specify actions to be determined by that machine. Further, while only a single machine is illustrated, the term “machine” shall also be taken to include a collection of machines that individually or jointly execute a set or multiple sets of instructions performing one and/or more of the methodologies discussed herein.

As an exemplary embodiment, computer system 226 includes a processor 202, which may be a central processing unit or a graphics processing unit or both, a main memory 204 and a static memory 206, which communicate with each other via a bus 208. Computer system 226 further includes a display unit 210, which may be a liquid crystal display or a cathode ray tube or both. Computer system 226 also includes an alphanumeric input device 212 such as a keyboard, a cursor control device 214 in the form of a mouse, a disk drive unit 216, a signal generation device 218 such as a speaker, and a network interface device 220.

Disk drive unit 216 includes a machine-readable medium 222 on which is stored one or more sets of instructions 224 in the form of software embodying one or more of the methodologies and/or functions disclosed herein. Instructions 224 constituting machine-readable media may reside completely or at least partially within main memory 204 and/or within processor 202 during execution thereof by computer system 226 or by the combination of main memory 204 and processor 202.

Instructions 224 may further be transmitted and/or received over a network 200 via the network interface device 220. While the machine-readable medium 222 is shown in an exemplary embodiment to be a single medium, the term “machine-readable medium” as used herein includes a single medium and/or multiple media such as a centralized and/or a distributed database, and/or associated caches and servers, that store data and/or instructions. The term “machine-readable medium” as used herein includes any medium capable of storing, encoding and/or providing a set of instructions for execution by the machine, which cause the machine to perform any of the methodologies disclosed herein. Accordingly, “machine-readable medium” as used herein includes solid-state memories, optical and magnetic media, and carrier wave signals.

FIG. 3 is a process flow diagram detailing operation of a method for predicting an electrical line issue, according to the invention. This method of predicting an electrical line issue includes the steps of (i) collecting electrical line readings from one or more internet of things sensors 302, (ii) transmitting the collected electrical line readings to a machine learning engine 304, (iii) visualizing electrical line issues based on an analysis by a big data engine 306, (iv) indicating the electrical line issues on a dynamic user interface 308, and setting an alarm for the one or more electrical line issues through one or more rule based engines and a multi-classification machine learning engine 310. The method of a distributed power line diagnosis predicts sag, surge, and swell through one or more power factors, peak value, harmonics etc. based on analysis by a machine learning engine.

FIG. 4 is a diagrammatic representation of a circular gauge to depict a state of an electrical line, such as an electrical line failure.

FIG. 5 is a process flow diagram detailing steps of distributed power line diagnosis, according one aspect of the invention. These steps of distributed power line diagnosis may include (i) receiving one or more electrical readings at a firmware board computation engine as in step 502 of FIG. 5, (ii) transmitting an output of firmware board computation engine to a data hub computation engine over a computer network as in step 504 of FIG. 5, (iii) transmitting an output of the data hub computation engine to a big data server as in step 506 of FIG. 5, (iv) visualizing one or more electrical line issues based on an analysis through the big data server as in step 508 of FIG. 5, (v) indicating the one or more electrical line issues through a user interface dynamic as in step 510 of FIG. 5, and (iv) setting an alarm for the one or more electrical line issues through one of a rule based engine and a multi-classification machine learning engine as in step 512 of FIG. 5.

The distributed power line diagnosis system may be based on the internet of things and may incorporate the Internet in various ways in the course of practice of the invention. The internet of things based distributed power line diagnosis system of the invention normally includes sensors, such as machine wearable sensors. Further, the system may be used for overseeing predictive maintenance of one or more power lines, which the system including a plurality of machine-wearable sensors, each of which is associated with a power line, with each sensor transmitting captured power line data over a wireless communication network. The system preferably further includes a sensor network for receiving and transmitting the captured data over a communication network and a machine learning algorithm engine receiving and analyzing data from the sensor network. The machine learning algorithm engine processes the received data, recognizing a pattern and a deviation respecting a parameter of interest such as sag and in response thereto issues control commands pertaining to the machine. The system preferably includes one or more control modules disposed in operative communication with a local firmware board associated with the power line, where the local firmware board receives and sends one or more control commands, executes the control commands, and transmits calculated/computed data over a communication network.

Machine learning engine 104 preferably raises an alarm when an electrical line failure is detected. Machine learning engine 104 is preferably operatively associated with internet of things server 112. The machine learning engine 104 preferably issues commands based on a learning outcome from an analysis of distributed calculations.

A three stage computation is desirable and may be necessary for distributed power line diagnosis. A first computation is at the local firmware board, a second computation is at the data hub, and the last computation is at the internet of things server. A computation engine is operatively associated with the local firmware board, and/or the data hub, and/or an internet of things server over a communication network.

The learning outcome as a result of analysis by the internet of things server is dependent on recognition of one of a pattern and deviation by the machine learning engine.

In an exemplary practice of the invention, data is collected from diverse locations, for example from as many as ten thousand different factory locations, for what is often abbreviated as “3P” or “prescriptive, preventative and predictive” maintenance. The system preferably uses a combination of a distributed database, such as the distributed database commercially available under the trademark “Cassandra,” a fast, general-purpose engine for large-scale data processing, such as the general purpose engine commercially available under the trademark “Spark,” and an open-source distributed real-time computation system, such as the open-source distributed real-time computation system commercially available under the trademark “Storm”. Additionally, the system uses a distributed streaming platform, such as the distributed streaming platform commercially available under the trademark “Kafka.” This combination of software tools allows processing data in real-time in a big data architecture approach using a broker system for storing the alarms in a buffered database, and then using the distributed database for creating a system for maintenance, repair, and operation analysis. This real time big data architecture is operatively associated with the internet of things server.

In accordance with the invention so-called “3P” maintenance is implemented for an electrical power line. Big data methodologies are disclosed herein are used to analyze data obtained from various locations respecting the power line through the internet of things sensor network. Big data techniques as used in the practice of the invention are needed to handle the massive volume of both structured and unstructured data, which could not be processed using a traditional database and traditional software techniques. Therefore, the invention uses a distributed real-time computation system, such as the one available commercially under the trademark “Apache Storm,” for distributed power line diagnosis.

In the practice of the invention, the real time data processing system is operatively associated with distributed databases. The real time data processing system is a “big data” system.

Using the invention, various electrical line issues as listed above are identified and quantified based on the system's computations, which are part of the analysis associated with the machine learning engine.

In the practice of the invention an alarm is optionally desirably set by either a rule based engine or a multi-classification machine learning engine.

In the preferred implementation of the invention, the communication network may operative under Wi-Fi, 2G, 3G, 4G, GPRS, EDGE, Bluetooth, ZigBee, Piconet of BLE, Zwave; or any other suitable protocol, or a combination thereof.

In the practice of the invention, an alarm is operatively raised over the communication network, providing a notification to a user on a mobile application such as a cellular telephone, tablet or laptop computer using short message service, email, or a combination thereof.

In the apparatus aspect of the invention, the computation engine is associated with one or more of the local firmware boards and/or the data hub, and/or the internet of things server.

Power supply quality may be sometimes inconsistent and/or poor. Poor and/or inconsistent power supply leads to increased maintenance costs for elected equipment. Power quality is a major issue particularly when sensitive electronic equipment is used under varying internal loads within individual plants. Operation of variable speed drives, microprocessor based devices, and other loads, such as lighting and battery chargers also contribute to the poor quality of electric power in a circuit. Operation of these devices may inherently cause poor power factor, harmonics and power quality events, such as sag, swell and surge.

In the power quality monitoring system and method aspects of the invention, the internet of things based architecture of the invention provides round the clock power quality tracking of individual machines. The power quality monitoring system includes sensors implementing and incorporating chip technology, a wireless network and a computation engine. The power quality monitoring system and methods according to the invention reduce operational costs of individual machines by a large percentage. This reduction in costs is further achieved through use of a combination of a single silicon chip, open source networking and cloud based software. The system further includes power monitoring sensors for tracking harmonic distortion, swell, sag, surge, flickering, etc.

The invention further embraces installation and use of snap-split-core: sensors installed on three phase electrical lines going into machines. A data hub collects the data from the sensors through a wireless network. The data is preferably pushed to cloud server from which a processor receives in real time a summary of issues respecting the three phase power lines going into individual machines. The data is optionally displayed on a smart phone and/or a tablet.

The predictive maintenance circular gauge illustrated in FIG. 11 is used in various implementations to visually depict various electrical line issues such as electrical line failure, sag, surge, swell, harmonics and flickering, as such parameters are analyzed using the method, system, and apparatus of the invention. The predictive maintenance circular gauge is desirably associated with one or more colors; the colors preferably include red, yellow, and green. Red indicates a dangerous mode of operation where the electrical power line may have failed or is about to fail. Yellow indicates an intermediate state of operation for an electrical with which the predictive maintenance gauge is associated. Green indicates an ideal, smooth state of operation for the electrical line with which the predictive maintenance circular gauge is associated.

It is within the scope of the invention to raise an alarm when the color is yellow or red.

In the practice of the method and operation of the invention, the internet of things sensors are preferably enabled to compute time series data.

Although the present embodiments have been described with a reference to specific example embodiments, it will be evident that various modifications and changes may be made to these embodiments without departing from the broader spirit and scope of the various embodiments. The various devices and modules described herein may be enabled and operated using hardware circuitry, firmware, software or any combination of hardware, firmware, and software embodied in a machine readable medium. The various electrical structures and methods may be embodied using transistors, logic gates and electrical circuits such as application specific integrated circuitry or digital signal processor circuitry.

In addition, it will be appreciated that the various operations, processes and methods disclosed herein may be embodied in a machine-readable medium and/or a machine accessible medium compatible with a data processing system, namely a computer, and may be performed in any order. The medium may be, for example, a memory, a transportable medium such as a compact disk, a digital video disk, a floppy disk, or a diskette. A computer program embodying aspects of the invention may be loaded onto a retail portal. The invention, when embodied in a computer program is not limited to specific embodiments discussed above and may, for example, be implemented in an operating system, an application program, a foreground or background process, a driver, a network stack or any combination thereof. The computer program may be executed on a single computer processor or multiple computer processors.

Claims

1. A method of providing predicted electrical line issues including phase current imbalance or phase current harmonics to a high number of end users having interest in the reliability of the line, comprising:

a) providing a single board computer processor;

b) connecting an electrical line to be analyzed for the presence or absence of phase current imbalance and phase current harmonics to the single board computer processor;

c) using the single board computer processor to collect and extract metadata indicative of phase current imbalance or phase current harmonics from any high frequency current or voltage present in said electrical line;

d) processing the metadata using either a machine learning support vector machine or a rule engine, which is equipped with a baseline dataset;

e) providing the processed metadata to a big data cloud data service database; and

f) making the results of the metadata processing available to a high number of end users communicating with the big data cloud data service database.

2. Apparatus for diagnosing status of electrical line performance by receiving and analyzing a plurality of electrical line data from a plurality of electrical lines, comprising:

a) a collection of Internet of things sensors, one operatively connected to each of the electrical lines, each sensor sensing one or more electrical parameters including current, voltage, power factor, harmonic distortion, swell, surge, sag, active power, reactive power and frequency of electrical power carried by the associated electrical line;

b) a communications network;

c) a collection of local firmware boards, one for each of the internet of things sensors and being connectedly associated therewith, for receiving sensed electrical line data from the internet of things sensors;

d) a data hub operatively connected to the local firmware boards and receiving therefrom the data sensed by the internet of things sensors;

e) a computation engine operatively connected to the local firmware boards, the data hub, and an internet of things server, for classifying different fault types in the electrical line data using quadratic hyperplanes in transformed variable space at the local firmware boards, at the data hub, or at the internet of things server, and computing at the local firmware boards, at the data hub, and at the internet of things server data parameter values pre-defined by a user;

f) a machine learning engine;

g) the internet of things server being operatively connected to the machine learning engine for analyzing the sensed electrical line data by comparing the sensed data to pre-defined prior condition data indicative of acceptable operation and raising an alarm if the sensed data deviates from the prior data indicative of acceptable operation by more than a preselected amount;

h) a gauge for displaying a visual indication of the performance state of a selected one of the parameters, the gauge accepting user-based intuition about the parameter state to affect the displayed visual indication.

3. A method for determining status of electrical line performance comprising:

a) positioning a plurality of internet of things sensors in operative communication with a plurality of electrical lines, one internet of things sensor for each line;

b) sensing electrical line data with the collection of internet of things sensors, each sensor sensing one or more electrical parameters, including current, voltage, power factor, harmonic distortion, swell, surge, sag, active power, reactive power and frequency of electrical power carried by the associated electrical line;

c) providing a collection of local firmware boards, one for each of the internet of things sensors and being connectedly associated therewith, for receiving sensed electrical line data from the internet of things sensors;

d) receiving from the local firmware boards the electrical line data sensed by the internet of things sensors on a data hub operatively connected to the local firmware boards;

e) providing a cloud-based internet of things server and computation engine operatively connected to the data hub;

f) classifying different fault types in the electrical line data using quadratic hyper planes in transformed variable space at the local firmware boards, at the data hub, or at the internet of things server;

g) computing at the local firmware boards, and/or at the data hub, and/or at the internet of things server preselected data parameters using quadratic hyperplanes in transformed variable space;

h) analyzing the sensed electrical line data by comparing the sensed data to pre-defined parameter values of prior data established using quadratic hyperplanes in transformed variable space that are indicative of acceptable operation;

i) raising an alarm if the sensed data deviates from the parameter values of the prior data by more than a preselected amount;

j) displaying a visual indication of the performance state of selected ones of the parameters, on a gauge accepting user-based intuition about the parameter state to affect the displayed visual indication.