Comparative analysis on performances of adjustable-gain single-neuron PID controllers based on general fuzzy logic and normal cloud model

The solutions to parameter setting of PID controllers have always been an essential problem of control system design. The single-neuron PID controller can achieve the parameters´ self-adaption to the real operation conditions by adjusting the gain value. Two computational intelligent algorithms deriving their theory sources from the uncertain reasoning are introduced to realize the on-line adjustments of gain, which are general fuzzy logic and a normal cloud model with universality. The designs on these two regulators are given containing the forms of the membership functions under the fuzzy logic & cloud model, control rules for 1-dimensional & 2-dimensional input modes, and the inference models, etc. The numerical simulations are implemented and the comparative analysis on the dynamic performance is presented based on the existent step responses and the adjustable parameters´ curves with adaptive changes. By contrast, it concludes that a 2-dimensional normal cloud model leads to the desired overshoot, and its 1-dimensional model with shorter program running time adapts to occasions of high real-time demands, and fuzzy logic regulators can better meet the control requirements of the shorter setting time.