Design and Analysis of GA based Neural/Fuzzy Optimum Adaptive Control

Process changes, such as flow disturbances and sensor noise, are common in the chemical and metallurgical industries. To maintain optimal performance, the controlling system has to adapt continuously to these changes. This is a difficult problem because the controller also has to perform well while it is adapting. The Adaptive Neural Controller (ANC) developed in this paper satisfies these goals. Using a neural network controller, ANC modifies the network parameters through Genetic Algorithms. Along with this a Fuzzy logic Controller is also implemented for the on-line tuning of PID controller even in the presence of noise. The performance of these approaches has been evaluated using data of different plants on a common set of performance indices. The simulations results show that identified GA based Adaptive neuro-controller along with PID controller was able to adapt to process changes while simultaneously avoiding hard constraints. The identified ANC balances the need to adapt with the need to preserve generalization, and constitutes a general tool for adapting neural controllers on-line. While the fuzzy system which is rather simple to build and implement (because of small computational efforts) considerably improves the system dynamics.