STUDIES ON FUZZY LOGIC & NEUROLOGIC CONTROL SYSTEMS

Abstract

The process controllers based on ANN's and fuzzy logic is relatively recent. The application of the controller in tough highly non-linear process such as pH control are very scarce and still remain a source of trouble in many industries. Two major classes of disturbances affect pH control systems. The first are variations in the process flow rates, which in industry are frequently attacked by feed-forward control. The second and more difficult class are changes in feed concentration. These can results in shifts of the titration curves and change in its shape. The aim of this work is to study by simulation a control system, for highly non-linear pH control, based on artificial neural network model and fuzzy knowledge based controller(FKBC). For variation in load as well as set-point the manipulated variable F2 (flow rate of NaOH) is to be regulated to control the pH in the system. The objective of this investigation is to develop and study three controllers namely ANN, FKBC, and PID to maintain a certain pH in the CSTR system in the presence of disturbances. An artificial neural network is used for steady-state and dynamic modeling of the pH-process. A back-propagation algorithm is used to train the ANN. The ANN model formed predict the test data excellently and was used as a base model for the process. A fuzzy logic controller is proposed for non-linear process control. The process is divided into three regions i.e. pH - low, pH - medium, pH - high. Then pH is used as an auxiliary variable to detect the process operating space so that the fuzzy controller can give a good performance. An ANN controller based on Back-Propagation algorithm was developed to control the above process. This controller was based on reverse modeling where the pH was an input and the manipulated variable was the output. The controller was formed to control the process satisfactorily but its performance was less satisfactory then the FKBC. The results of the above controller was compared with a conventional PID controller to establish the effectiveness of the advance controllers over the conventional controller. It was observed that though the PID controller was tuned to optimal condition but it produced high overshoot and high rise time.