Switching control systems and their design automation via genetic algorithms

The objective of this work is to provide a simple and effective nonlinear controller. Our strategy involves switching the underlying strategies in order to maintain a robust control. If a disturbance moves the system outside the region of stability or the domain of attraction, it will be guided back onto the desired course by the application of a different control strategy. In the context of switching control, the common types of controller present in the literature are based either on fuzzy logic or sliding mode. Both of them are easy to implement and provide efficient control for non-linear systems, their actions being based on the observed input/output behaviour of the system. In the field of fuzzy logic control (FLC) using error feedback variables there are two main problems. The first is the poor transient response (jerking) encountered by the conventional 2-dimensional rule-base fuzzy PI controller. Secondly, conventional 3-D rule-base fuzzy PID control design is both computationally intensive and suffers from prolonged design times caused by a large dimensional rule-base. The size of the rulebase will increase exponentially with the increase of the number of fuzzy sets used for each input decision variable. Hence, a reduced rule-base is needed for the 3-term fuzzy controller. In this thesis a direct implementation method is developed that allows the size of the rule-base to be reduced exponentially without losing the features of the PID structure. This direct implementation method, when applied to the reduced rule-base fuzzy PI controller, gives a good transient response with no jerking. Although sliding mode control (SMC) is theoretically excellent in terms of robustness to parameter uncertainty and invariance to unknown disturbances, occurrence of control chattering is inevitable and also undesirable to many process applications. This is due to high frequency switching in the SMC controller and this may, in turn, excite high frequency unmodelled plant dynamics. To reduce this unwanted control chattering we propose 1) a soft-switching scheme using a hyperbolic-tangent function and 2) an extension of the sliding hyperplane to a "sliding region". Due to the difficulty in devising the equivalent control for a system with unknown/time-varying nonlinearities, a PID control structure is used as an alternative. Hence, a sliding mode controller based upon the PID control structure is developed. A fuzzy sliding mode controller is also presented with the aim of improving the unwanted chattering. The smooth control action of the FLC is used to either adjust the switching gains of the