Genetic algorithm based Neuro Fuzzy tuning of PID controller for a nonlinear temperature water bath with feedforward control

Abstract

Temperature control is an important task in the modern industries like chemical, pharmaceutical etc. considering the real time load and set point variations in their plants. The present work gives experimental results for the control of a constant level stirred nonlinear temperature water bath with continuous inlet and outlet flow under load and set point disturbances. When the inlet temperature becomes comparable to the desired set point their effect becomes that much prominent in terms of rise time, overshoot and short term stability. In the present system, the level plays an important factor in the heat distribution. The slight variation in the inlet flow rate disturbs the system by varying its level assumed to be constant. Also the fluctuation in the mains voltage plays a major role in the short term stability of the system. In the work presented here, all these three disturbances have been considered using feedforward control. A real time PID controller taken with a fixed bias voltage has been used and tuned by Genetic Algorithm based Neuro Fuzzy Controller (GANFC) developed on the platform of Visual BASIC 6.0. The PID controller was designed and developed based on 89C51 microcontroller around MK-70 MLW bath with 10Litre capacity. A 1500W heater was operated between 0–240Vac with 256 different firing angles for the Triac. Fluke Hydra 2625A Data acquisition unit has been used for acquiring the data. The load disturbances have been created by varying inlet flow rate from 100m1/min to 500m1/min at 28°C for a set point of 40°C. Also the set point disturbance was induced by changing the set point from 35.0°C to 45.0°C. The constant level is maintained by keeping the outflow rate same as the inflow rate. Results for the modified GANFC, including feedforward control, were compared with GANFC controller without feedforward control.