Neruo-pid controller
AssigneeHoneywell Inc.
InventorsAhmet F. Konar, Tariq Samad, Steven A. Harp
Life of this patent
Estimating the term…
Abstract
PID controllers form a large proportion of controllers in use in many controlled systems today. This application describes how to use a neural network which receives PID inputs to be a controller and operate as a PID controller to save on retraining and provide other efficiencies in control. Also shown is the user selectability between PID conventional controllers and Neural Network controllers.
Claims13
A nonlinear controller for a closed-loop system, having PID parameter values as inputs (variable values K.sub.I, K.sub.C, K.sub.D) and a process response related and setpoint related signal value as input, wherein said nonlinear controller contains a processor mechanism for generating an output signal value u for controlling said closed-loop system from said inputs and wherein said processor mechanism is a neutral network trained on inputs e, K.sub.I, K.sub.D, K.sub.C, which are generated by a model of the system to be controlled.
A nonlinear controller for a closed-loop system, having PID parameter values as inputs (variable values K.sub.I, K.sub.C, K.sub.D) and a process response related and setpoint related signal value as input, wherein said nonlinear controller contains a processor mechanism for generating an output signal value u for controlling said closed-loop system from said inputs and wherein said mechanism is a processor that is representative of the optimum equation for the controller output u and wherein said optimum equation is represented by a process containing equations for processing input signals which have been derived by parsing them out of neural network systems trained on equivalent processes.
A controller for a closed loop system having a conventional PID controller and a NeuroPID controller which has been trained on the system, wherein the user may select to employ either the output of the NeuroPID controller combined with the conventional PID controller, or the conventional PID controller output alone said user selection being made in a set of ways including, during operation of the controller by a user input switch means, and by automatic means responsive to preselected user criteria.
A nonlinear controller for a closed-loop system, having PID parameter values as inputs (variable values K.sub.I, K.sub.C, K.sub.D) and a process response related and setpoint related signal value as input, wherein said nonlinear controller contains a processor mechanism for generating an output signal value u for controlling said closed-loop system from said inputs and wherein said mechanism is a processor that is representative of the optimum equation for the controller output u and wherein said optimum equation is represented by a process containing look-up tables for processing input signals which have been derived by parsing them out of neural network systems trained on equivalent processes.
The controller as set forth in claim 13 wherein said process response related input is two input signals;
- one representing the desired response (y.sub.r) and one representing the process response (y).
Dependencies are inferred from the claim text.