The Study On Hybrid Intelligent Controller And Its Simulation On Industrial Plants

This thesis mainly studies and discusses on introducing hybrid intelligent control into the control system. Since every control strategy has its limitation and should be only suitable for some specified area, it would be very difficult to get satisfied result if only single control strategy being used. However, if different control strategies are combined organically, we can gain better control effects. This thesis proposes a few kinds of hybrid intelligent controllers on the basis of referring to a great deal of correlative data. Based on analysis of PID controller, two kinds of intelligent PID controllers are proposed: PID controller with variable arguments and rapid intelligent PI controller. VAPID changes the PID parameters on-line with the system error by the nonlinear functions. RIPI combines time optimal controller with PI controller, it is a kind of intelligent controller based on rules. Within the large error range, RIPI uses time optimal control, and within the small error range, it uses PI controller. By the state point, two schemes change each other. Based on analysis fuzzy control, three kinds of hybrid intelligent controllers basing on fuzzy control are proposed: fuzzy parameters PID adaptive controller, a robust self-turning scheme for PID type fuzzy logic controller ( STPIDFLC ) and three-dimensional fuzzy controller with control rules self-turning. Fuzzy parameters PID adaptive controller is the combination of fuzzy control and PID. In order to adaptive parameter's change of controlled process, it applies fuzzy rules and fuzzy inference to judge the parameters of PID controller. STPIDFLC's output scaling factors Gu is adjusted on-line by fuzzy rules according to the current trend of the controlled process. Three-dimensional fuzzy controller with control rules self-turning of which fuzzy rules adopts analytic description, comes true the fuzzy rules sell-optimizing by optimization selection adjusting control rules on-line. At the same time, a lot of simulation studies have been done to the hybrid intelligent controllers above mentioned, the simulation results prove better performance obtained than corresponding unimproved ones.In this thesis, ANFIS ?Adaptive Network-Based Fuzzy Inference System is applied to the control system. ANFIS that can produce fuzzy rules from sample datadirectly is organically combination of fuzzy inference system and neural network. So it isIIwell suitable for some complex industry control process lacking of expert experience. Because ANFIS realizes the combination of fuzzy quantity and numeric quantity ,and its both parameters and weights adjusted have clear physics meaning, it's much propitious to be understood, mastered and applied by control engineers. So adopted ANFIS has great promising industry application. The thesis discusses the structure and learning method of ANFIS, and shows it's application by the example of a car's brake control. By the comparison between the simulation results of the single fuzzy controller and that of ANFIS, it is proved that ANFIS is really an effective hybrid intelligent control method.