Research Of Temperature Control Based On Self-learning Logic Control

The electric heater is a kind of heating equipments, which is widely used in the labs and the industrial processes. Its temperature control is a typical process control. The electric heater is an object featuring in non-linearity, time-varying, large time lag and asymmetry. To minimize the effects of these features on the control performance, the adaptive control algorithm is adopted, but it's difficult to be transplanted to the small controller; besides, for a new temperature control system, the control parameters should be studied repeatedly. So, this algorithm is not a universalized one and can not meet the actual needs of replacing a new controller on line on the factory scene. So the algorithm based on self-learning logic control is proposed to realize precise temperature control of electric heater.The first thing to do is to adopt self-learning algorithm, which is similar to iterative learning control, to identify the parameters of the temperature control system online, if the parameters is unknown. Reckon the core parameter U0 of logic control. Then, apply expert control to make the temperature approach the set value quickly. After the temperature close to the target value, use logic control which decides the control volume by judging the logic relation of error and error rate of the temperature. Simultaneously, if the system temperature is in a temporary state of stability, perfects the parameter U0 based on the self-learning algorithm.The above-mentioned method of electric heater temperature control, which is based on self-learning logic control, has taken some advantages of logic control such as good robustness, high control precision and strong anti-disturbance. It also realizes the self-learning and self-tuning function for the unknown system or system with changed parameters. The self-learning logic control algorithm is proved that it's convergent by theoretical analysis. The algorithm was applied in the PLC temperature control instrument after being simplified and optimized further, and the relative real-time test results have proven its good performance in temperature control, which get an overshoot less than 10℃of the set value and a stable error in the range of±1℃.