Abstract
A new methodology is introduced for designing and tuning PID-type Fuzzy Logic Coutrollers (PID-type FLCs) to control processes of which the mathematical models are unknown. The PID-type FLCs is developed with two inputs of equally spaced triangular membership functions (ESMFs), a crisp type output membership function, a linear rule base and product-sum-gravity inference method. It can be showh has equivalent to a traditional linear PID controller which is widely used in practical industrial control processes. We can obtain the initial scaling factors (GE、GR、GU and GV) of PID-type FLCs from the three well-tuned parameters (Kp3 Kl3 KD) of the linear PID controller. Theoretical analysis and the explicit mathematical forms of reasoning surface obtained by using two of the most important t-norms are addressed. The input/output relations of PID-type FLCs using these two t-norm sum-gravity inference methods are also studied A simple comparative tuning procedure is introduced to optimize the scaling factors and the membership functions by less trial and error, and the applicability of PID-type FLCs is greatly enhanced. A simulation study is conducted to reveal the performan ce improvement of the proposed PID-type FLCs, and then it is applied to a practical temperature control system which is a kind of heat exchange systems widely used in industrial processes. Such processes are not easy to control in practice because of their nonlinearity and time delay in dynamics. The result shows that PID-type FLCs is much simpler in design than traditional FLCs, and the performance is also better than linear PID controller.