Closed loop neural network automatic tuner

Claim 1

A neural network automatic tuner for tuning a PID controller which in turn controls a system in a closed loop control system configuration, comprising: a neural network autotuner, wherein said neural network autotuner receives a y output, representing an existing value of output of a system under control, and produces a tuner output to the PID controller of optimized PID controller parameters made up of a combination of a set of parameters including K I , K D and K C , wherein said set of parameters are consistent with a PID controller-based process model selected from a set of process models similar to the system under control and are also consistent with a predefined optimization criteria, or are consistent with a combination of the set of process models and the predefined optimization criteria, and wherein said PID controller parameters are applied to said PID controller to generate a new near-optimal controller output u and a new near-optimal system output y of the system, and wherein in order to initiate tuning of the set of parameters by the neural network autotuner said neural network autotuner provides a setpoint change output which adjusts a setpoint input to the system operating in closed loop configuration.