Autotuning a PID controller: A Fuzzy-genetic Approach

A new method for tuning the parameters of the proportional integral derivative (PID) controller is presented in this paper. The technique adopted in this proposition is based on the format of dead-beat control. Fuzzy inference mechanism has been used here for predicting the future values of the controller output while crisp consequent values of the rulebase of the Takagi–Sugeno model are optimized using a genetic algorithm. The proposition is an extension of the work in R. Bandyopadhyay, D. Patranabis (A new autotuning algorithm for PID controllers using dead-beat format, ISA Trans., accepted for publication) where the rulebase was prepared based on the knowledge of process experts. The use of genetic algorithm for optimizing the crisp values of the rulebase has considerably improved the performance of the PID autotuner. The proposed algorithm seems to be a complete and generalized PID autotuner as can be seen from the simulated and experimental results. In all the cases the method shows substantial improvement over the controller tuned with Ziegler–Nichols formula and the PID controller proposed in (loc cit).