Abstract: A method and apparatus for controlling a non-linear mill. A linear controller is provided having a linear gain k that is operable to receive inputs representing measured variables of the plant and predict on an output of the linear controller predicted control values for manipulatible variables that control the plant. A non-linear model of the plant is provided for storing a representation of the plant over a trained region of the operating input space and having a steady-state gain K associated therewith. The gain k of the linear model is adjusted with the gain K of the non-linear model in accordance with a predetermined relationship as the measured variables change the operating region of the input space at which the linear controller is predicting the values for the manipulatible variables.

Abstract: An optimization system is provided utilizing a Bayesian neural network calculation of a derivative wherein an output is optimized with respect to an input utilizing a stochastical method that averages over many regression models. This is done such that constraints from first principal models are incorporated in terms of prior art distributions.

Abstract: A method for controlling a plant to achieve desired operating results. Select operating parameters of the plant are measured and input to a plurality of transforms that define select actions to be taken by an operator of the plant as a function of the measured select operating parameters. Each of the transforms is associated with a portion of the measured select operating parameters and is operable to determine if a predetermined and associated condition exists in the plant, which would warrant the associated action being taken. The measured select operating parameters are processed through the associated transforms to determine for each of the transforms if the associated condition is present. An indication that the condition associated with any of the transforms is present, and for which transform, is provided to a user. A suggestion of the action to be taken for the associated indication is then provided to the user.

Abstract: A kiln thermal and combustion control. A predictive model is provided of the dynamics of selected aspects of the operation of the plant for modeling the dynamics thereof. The model has at least two discrete models associated therewith that model at least two of the selected aspects, the at least two discrete models having different dynamic responses. An optimizer receives desired values for the selected aspects of the operation of the plant modeled by the model and optimizes the inputs to the model to minimize error between the predicted and desired values. A control input device then applies the optimized input values to the plant after optimization thereof.

Abstract: A neural network system is provided that models the system in a system model (12) with the output thereof providing a predicted output. This predicted output is modified or controlled by an output control (14). Input data is processed in a data preprocess step (10) to reconcile the data for input to the system model (12). Additionally, the error resulted from the reconciliation is input to an uncertainty model to predict the uncertainty in the predicted output. This is input to a decision processor (20) which is utilized to control the output control (14). The output control (14) is controlled to either vary the predicted output or to inhibit the predicted output whenever the output of the uncertainty model (18) exceeds a predetermined decision threshold, input by a decision threshold block (22).

Abstract: A system and method for on-line training of a non-linear model for use in electronic commerce. The non-linear model is trained with training sets from a stream of process data. The system detects availability of new training data, and constructs a training set from the corresponding input data. Over time, many training sets are presented to the non-linear model. When multiple presentations are needed to effectively train the non-linear model, a buffer of training sets is filled and updated as new training data becomes available. Once the buffer is full, a new training set bumps the oldest training set from the buffer. The training sets are presented one or more times each time a new training set is constructed. An historical database may be used to construct training sets for the non-linear model. The non-linear model may be trained retrospectively by searching the historical database and constructing training sets.

Abstract: An E-Commerce system using modeling of inducements to customers. A potential user is interfaced with a commerce site to receive information therefrom during a commercial transaction. A commerce model of a commerce system is provided that predicts as an output a defined commercial result as a function of information related to the user when in the commerce transaction and also as a function of inducements that can be provided to the user during the commerce transaction. The inducement input to the commerce model is varied to vary the predicted output of the commerce model in a predetermined manner until a desired predicted output of the commerce model is achieved. This varied inducement is then provided to the user during the commercial transaction.

Abstract: A method and apparatus for controlling a non-linear mill. A linear controller is provided having a linear gain k that is operable to receive inputs representing measured variables of the plant and predict on an output of the linear controller predicted control values for manipulatible variables that control the plant. A non-linear model of the plant is provided for storing a representation of the plant over a trained region of the operating input space and having a steady-state gain K associated therewith. The gain k of the linear model is adjusted with the gain K of the non-linear model in accordance with a predetermined relationship as the measured variables change the operating region of the input space at which the linear controller is predicting the values for the manipulatible variables. The predicted manipulatible variables are then output after the step of adjusting the gain k.

Abstract: A method for providing independent static and dynamic models in a prediction, control and optimization environment utilizes an independent static model (20) and an independent dynamic model (22). The static model (20) is a rigorous predictive model that is trained over a wide range of data, whereas the dynamic model (22) is trained over a narrow range of data. The gain K of the static model (20) is utilized to scale the gain k of the dynamic model (22). The forced dynamic portion of the model (22) referred to as the bi variables are scaled by the ratio of the gains K and k. The bi have a direct effect on the gain of a dynamic model (22). This is facilitated by a coefficient modification block (40). Thereafter, the difference between the new value input to the static model (20) and the prior steady-state value is utilized as an input to the dynamic model (22). The predicted dynamic output is then summed with the previous steady-state value to provide a predicted value Y.

Abstract: A kiln thermal and combustion control. A predictive model is provided of the dynamics of selected aspects of the operation of the plant for modeling the dynamics thereof. The model has at least two discrete models associated therewith that model at least two of the selected aspects, the at least two discrete models having different dynamic responses. An optimizer receives desired values for the selected aspects of the operation of the plant modeled by the model and optimizes the inputs to the model to minimize error between the predicted and desired values. A control input device then applies the optimized input values to the plant after optimization thereof.

Abstract: An on-line optimizer is provided wherein a boiler (720) is optimized by measuring a select plurality of inputs to the boiler (720) and mapping them through a predetermined relationship that defines a single value representing a spacial relationship in the boiler that is a function of the select inputs. This single value is then optimized with the use of a plant optimizer (818) which provides an optimized value. This optimized value is then processed thought the inverse relationship of the single modified value to provide modified inputs to the plant that can be applied to the plant.

Abstract: A plant (72) is operable to receive control inputs c(t) and provide an output y(t). The plant (72) has associated therewith state variables s(t) that are not variable. A control network (74) is provided that accurately models the plant (72). The output of the control network (74) provides a predicted output which is combined with a desired output to generate an error. This error is back propagated through an inverse control network (76), which is the inverse of the control network (74) to generate a control error signal that is input to a distributed control system (73) to vary the control inputs to the plant (72) in order to change the output y(t) to meet the desired output. The control network (74) is comprised of a first network NET 1 that is operable to store a representation of the dependency of the control variables on the state variables. The predicted result is subtracted from the actual state variable input and stored as a residual in a residual layer (102).

Abstract: A neural network system is provided that models the system in a system model (12) with the output thereof providing a predicted output. This predicted output is modified or controlled by an output control (14). Input data is processed in a data preprocess step (10) to reconcile the data for input to the system model (12). Additionally, the error resulted from the reconciliation is input to an uncertainty model to predict the uncertainty in the predicted output. This is input to a decision processor (20) which is utilized to control the output control (14). The output control (14) is controlled to either vary the predicted output or to inhibit the predicted output whenever the output of the uncertainty model (18) exceeds a predetermined decision threshold, input by a decision threshold block (22).

Abstract: An on-line optimizer is comprised of a nonlinear dynamic model (702) which is operable to provide an estimation of the output of a plant. This receives manipulated variables (MV), disturbance variables (DV), and computed disturbance variables (CDB). The estimated output of the model is then compared to the actual output measured by virtual on-line analyzer (VOA) (616). This is compared is a difference block 618 to generate a bias which is then filtered by a filter(620). The output thereof is then provided to an output block (672) in a steady state optimizer (700) to offset the desired setpoints. These set points are input to a steady state nonlinear model which is operable to optimize the inputs to the plants for use for writing new set points in accordance with a predetermined cost function. This cost function is utilized to optimize the new inputs with the use of the steady state model in accordance with various constraints and target values.

Abstract: A method for building a model of a system includes first extracting data from a historical database (310). Once the data is extracted, a dataset is then created, which dataset involves the steps of preprocessing the data. This dataset is then utilized to build a model. The model is defined as a plurality of transforms which can be utilized to run an on-line model. This on-line model is interfaced with the historical database such that the variable names associated therewith can be downloaded to the historical database. This historical database can then be interfaced with a control system to either directly operate the plant or to provide an operator an interface to various predicted data about the plant. The building operation will create the transform list and then a configuration step is performed in order to configure the model to interface with the historical database. When the dataset was extracted, it is unknown whether the variables names are still valid.

Abstract: A distributed control system (14) receives on the input thereof the control inputs and then outputs control signals to a plant (10) for the operation thereof. The measured variables of the plant and the control inputs are input to a predictive model (34) that operates in conjunction with an inverse model (36) to generate predicted control inputs. The predicted control inputs are processed through a filter (46) to apply hard constraints and sensitivity modifiers, the values of which are received from a control parameter block (22). During operation, the sensitivity of output variables on various input variables is determined. This information can be displayed and then the user allowed to select which of the input variables constitute the most sensitive input variables. These can then be utilized with a control network (470) to modify the predicted values of the input variables. Additionally, a neural network (406) can be trained on only the selected input variables that are determined to be the most sensitive.

Abstract: A neural network system is provided that models the system in a system model (12) with the output thereof providing a predicted output. This predicted output is modified or controlled by an output control (14). Input data is processed in a data preprocess step (10) to reconcile the data for input to the system model (12). Additionally, the error resulted from the reconciliation is input to an uncertainty model to predict the uncertainty in the predicted output. This is input to a decision processor (20) which is utilized to control the output control (14). The output control (14) is controlled to either vary the predicted output or to inhibit the predicted output whenever the output of the uncertainty model (18) exceeds a predetermined decision threshold, input by a decision threshold block (22).

Abstract: A method for modeling a steady-state network in the absence of steady-state historical data. A steady-state neural network can be tied by impressing the dynamics of the system onto the input data during the training operation by first determining the dynamics in a local region of the input space, this providing a set of dynamic training data. This dynamic training data is then utilized to train a dynamic model, gain thereof then set equal to unity such that the dynamic model is now valid over the entire input space. This is a linear model, and the historical data over the entire input space is then processed through this model prior to input to the neural network during training thereof to remove the dynamic component from the data, leaving the steady-state component for the purpose of training. This provides a valid model in the presence of historical data that has a large content of dynamic behavior.

Abstract: An automatic data flow architecture builder is disclosed that is operable to take raw data which is stored in a raw data buffer (60) and transform it for storage in a transformed data buffer (78). A plurality of transform blocks (68), (70) and (72) are provided for this transform operation, these disposed in a predetermined data flow. When the user inputs a new transform to be disposed within the data flow of the transforms, rules are applied via a rule-base processing system (74) to apply a set of predetermined rules in a rule database (76) to the transform. These rules determine where the transform is to be inserted. This provides an automatic construction operation of a data flow architecture.

Abstract: A method for providing independent static and dynamic models in a prediction, control and optimization environment utilizes an independent static model and an independent dynamic model. The static model is a rigorous predictive model that is trained over a wide range of data, whereas the dynamic model is trained over a narrow range of data. The gain K of the static model is utilized to scale the gain k of the dynamic model. The forced dynamic portion of the model referred to as the b.sub.i variables are scaled by the ratio of the gains K and k. The b.sub.i have a direct effect on the gain of a dynamic model. This is facilitated by a coefficient modification block. Thereafter, the difference between the new value input to the static model and the prior steady-state value is utilized as an input to the dynamic model. The predicted dynamic output is then summed with the previous steady-state value to provide a predicted value Y. Additionally, the path that is traversed between steady-state value changes.