Design And Application Of A Multi-Variable Fuzzy Logic Control System

In practical industrial processes, there are many multi-inputs multi-outputs (MIMO) systems, such as multi-link robot and probe movement of scanning probe microscopy (SPM), etc. Morever, states and controls of many distributed parameter systems (DPS) are related with time and space. These systems are called or spatio-temporal systems, such as micro-cantilever in SPM and catalytic reaction rod, etc. DPS are inherent infinite-dimensional, and they are usually converted to MIMO systems with some low-order approximation modeling methods in convenient for controlling them, and will deduce large modeling error. Those systems have uncertainties, such as unknown nonlinearities, disturbances etc., therefore, it is difficult to get their precise models. Traditional controllers, such as PID, cannot get a sound performance. In contract, intelligent methods which are developing these years, such as fuzzy control, can reduce dependence on the mathematical model. However, it needs to design many membership functions (MF) and rule bases on expert experience when many fuzzy logic controllers are used to control them. This design method is complex relatively. In order to solve those problems, the major work of the paper focuses on the following aspects:(1) Multi-variable fuzzy logic system is built.On account of only one MF being adopted for all subsystems, MFs for other subsystems are adjusted by scaling factors. A multi-variable fuzzy logic control system is designed following the fuzzy matrix theory. At the same time, the control system adopts only one traditional rule base, and there is no need to design MFs for all subsystems according to expert experience. The control system can be considered as an extension of traditional fuzzy system, and the design method is very simple.(2) A multi-variable fuzzy logic controller is designed, and a mathematical model is deduced.Mathematical model of the multi-variable fuzzy logic controller (FLC) is deduced following an inference engine. Relationships between multi-variable FLC and PID controller are built. Therefore, scaling factors (parameters) of multi-variable FLC can be tuned through tuning method of PID controller, and the design method is simple.(3) Tuning method of multi-variable FLC is proposed based on PID controller.Tuning based on internal mode control is proposed for stable processes. Phase margin and gain margin tuning method is proposed for unstable processes. For MIMO systems, parameters of PID are used to design parameters of multi-variable FLC thorough decoupling. For DPS, a finite-dimensional approximate model is obtained through time/space variable separation and Galerkin method. The approximate model is used to design multi-variable FLC,and a spatio-temporal controller is synthesized by spatial basis and multi-variable FLC. Enough spatial information is considered in this method and the design method is effective through simulating.(4) Intelligent tuning methods of multi-variable FLC are researched.Scaling factors of multi-variable FLC are adaptively adjusted by gradient descent, and neural network is used to identify unknown variables. The simulations show the usefulness of it. Particle swarm optimization is used to tuning parameters of multi-variable FLC for DPS, and a sound performance is obtained from simulations. Stability of multi-variable FLC controlled DPS is analyzed through the mathematical model. The stability conditions are also given for the controlled system.(5) Probe movement of SPM in nano-manipulation platform is considered as a DPS. Multi-variable FLC is used to control the movement. Performance of multi-variable FLC is better than PID controller in experiment. Therefore, the design of multi-variable FLC are effective.