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 controller directs the operation of an air pollution control (APC) system having one or more controllable operating parameters and a defined operating limit representing a regulatory limit on an allowed amount of pollutant to be emitted by the APC system. An interface receives data representing a value of a regulatory credit available for emitting less of the pollutant than the regulatory limit on the allowed amount of pollutant. A control processor (i) determines a target set point for each of at least one of the one or more controllable operating parameters, which will maximize the regulatory credits earned, based on the received data and (ii) to directs control of each of the at least one controllable operating parameter based on the determined target set point for that parameter.

Abstract: Method and apparatus for training a system model with gain constraints. A method is disclosed for training a steady-state model, the model having an input and an output and a mapping layer for mapping the input to the output through a stored representation of a system. A training data set is provided having a set of input data u(t) and target output data y(t) representative of the operation of a system. The model is trained with a predetermined training algorithm which is constrained to maintain the sensitivity of the output with respect to the input substantially within user defined constraint bounds by iteratively minimizing an objective function as a function of a data objective and a constraint objective. The data objective has a data fitting learning rate and the constraint objective has constraint learning rate that are varied as a function of the values of the data objective and the constraint objective after selective iterative steps.

Abstract: A method and apparatus for optimizing air flow to a boiler of a power generating unit using advanced optimization, modeling, and control techniques. Air flow is optimized to maintain flame stability, minimize air pollution emissions, and improve efficiency.

Abstract: A method and apparatus for estimating and/or controlling mercury emissions in a steam generating unit. A model of the steam generating unit is used to predict mercury emissions. In one embodiment of the invention, the model is a neural network (NN) model. An optimizer may be used in connection with the model to determine optimal setpoint values for manipulated variables associated with operation of the steam generating unit.

Abstract: A method and apparatus for optimizing the operation of a single or multiple power generating units using advanced optimization, modeling, and control techniques. In one embodiment, a plurality of component optimization systems for optimizing power generating unit components are sequentially coordinated to allow optimized values determined by a first component optimization system to be fed forward for use as an input value to a subsequent component optimization system. A unit optimization system may be provided to determine goals and constraints for the plurality of component optimization systems in accordance with economic data. In one embodiment of the invention, a multi-unit optimization system is provided to determine goals and constraints for component optimization systems of different power generating units. Both steady state and dynamic models are used for optimization.

Abstract: Method and apparatus for training a system model with gain constraints. A method is disclosed for training a steady-state model having an input and an output and a mapping layer for mapping the input to the output, the model comprising a stored representation of a plant or process, and including a linear portion and a non-linear portion, where the non-linear portion includes a function. Input is received to the model, and predicted output computed corresponding to attribute(s) of the plant or process. The predicted output is stored, and is usable to manage the plant or process. The model is trained to optimize a specified objective function subject to one or more constraints, e.g., via a non-linear programming (NLP) optimizer, the constraints including, hard constraint(s) comprising strict limitations on the training in optimizing the objective function, and/or soft constraint(s) comprising a weighted penalty function included in the objective function.

Abstract: A kiln thermal and combustion control. A predictive model is provided of the dynamics of selected aspects of the operation of the system 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 system 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 system after optimization thereof.

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 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: Non-linear model with disturbance rejection. A method for training a non linear model for predicting an output parameter of a system is disclosed that operates in an environment having associated therewith slow varying and unmeasurable disturbances. An input layer is provided having a plurality of inputs and an output layer is provided having at least one output for providing the output parameter. A data set of historical data taken over a time line at periodic intervals is generated for use in training the model. The model is operable to map the input layer through a stored representation to the output layer. Training of the model involves training the stored representation on the historical data set to provide rejection of the disturbances in the stored representation.

Abstract: A method for providing independent static and dynamic models in a 000 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: Method and apparatus for training a system model with gain constraints. A method is disclosed for training a steady-state model, the model having an input and an output and a mapping layer for mapping the input to the output through a stored representation of a system. A training data set is provided having a set of input data u(t) and target output data y(t) representative of the operation of a system. The model is trained with a predetermined training algorithm which is constrained to maintain the sensitivity of the output with respect to the input substantially within user defined constraint bounds by iteratively minimizing an objective function as a function of a data objective and a constraint objective. The data objective has a data fitting learning rate and the constraint objective has constraint learning rate that are varied as a function of the values of the data objective and the constraint objective after selective iterative steps.

Abstract: At least one of the multiple process parameters (MPPs) is a controllable process parameter (CTPP) and one is a targeted process parameter (TPP). The process also has a defined target limit (DTV) representing a first limit on an actual average value (AAV) of the TPP. A first logical controller predicts future average values (FAVs) of the TPP based on the AAVs of the TPP over a first prior time period and the DTV. A second logical controller establishes a further target limit (FTV) representing a second limit on the AAV of the TPP based on one or more of the predicted FAVs, and also determines a target set point for each CTPP based on the AAVs of the TPP over a prior time period and the FTV. The second logical controller directs control of each CTPP in accordance with the determined target set point.

Abstract: A controller directs performance of a process having multiple process parameters (MPPs), including a controllable process parameter (CTPP), a targeted process parameter (TPP), a defined target value (DTV) representing a limit on an actual average value (AAV) of the TPP over a defined moving time period of length TPLAAV. A storage device stores historical data representing the AVs of the TPP at various times over a prior time period (PTP) having a length of at least TPLAAV. A processor predicts future average values (FAVs) of the TPP over a future time period (FTP) based on the stored historical data and the current values of the MPPs. The processor also determines a target set point for each CTPP based on the predicted FAVs, the current values of the MPPs and the DTV, and directs control of each CTPP in accordance with the determined target set point for that CTPP.

Abstract: A kiln thermal and combustion control. A predictive model is provided of the dynamics of selected aspects of the operation of the system 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 system 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 system after optimization thereof.

Abstract: Method and apparatus for training a system model with gain constraints. A method is disclosed for training a steady-state model, the model having an input and an output and a mapping layer for mapping the input to the output through a stored representation of a system. A training data set is provided having a set of input data u(t) and target output data y(t) representative of the operation of a system. The model is trained with a predetermined training algorithm which is constrained to maintain the sensitivity of the output with respect to the input substantially within user defined constraint bounds by iteratively minimizing an objective function as a function of a data objective and a constraint objective. The data objective has a data fitting learning rate and the constraint objective has constraint learning rate that are varied as a function of the values of the data objective and the constraint objective after selective iterative steps.

Abstract: Method and apparatus for training a system model with gain constraints. A method is disclosed for training a steady-state model, the model having an input and an output and a mapping layer for mapping the input to the output through a stored representation of a system. A training data set is provided having a set of input data u(t) and target output data y(t) representative of the operation of a system. The model is trained with a predetermined training algorithm which is constrained to maintain the sensitivity of the output with respect to the input substantially within user defined constraint bounds by iteratively minimizing an objective function as a function of a data objective and a constraint objective. The data objective has a data fitting learning rate and the constraint objective has constraint learning rate that are varied as a function of the values of the data objective and the constraint objective after selective iterative steps.

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 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.