An algorithm for automated fuzzy logic controller tuning

An automated method for calibrating a fuzzy logic controller (FLC) has been developed based on the cell state-space concept. The system's state space is quantized into cells creating a spatially discrete model of system behavior. Given a cost function and plant simulation model, a cell-state-space-based optimal control algorithm generates a table of desired control actions. The control table provides a discrete approximation to the global optimal control policy with respect to the cost function. An adjusted cost function algorithm is presented that uses fuzzy constraints to improve the accuracy of the control table. In addition, a nonuniform quantization scheme is described that also improves the accuracy of both the control table and the cell mappings used for analysis. The performance and robustness characteristics of an FLC designed using this automated method were explored with the minimum time control of an inverted pendulum.<
>