Fast incremental best estimate directed search-a significantly expedited algorithm for Takagi-Sugeno type fuzzy logic controller automatic optimization

摘要

Over the last few years, a series of cell state space based algorithms has been proposed for Takagi-Sugeno type fuzzy logic controller automatic optimization with promising results. The core algorithm is called incremental best estimate directed search (IBEDS), which is an implementation of the concept called globally directed random search. Originally, IBEDS only applies global random search on the training set, this paper presents a new approach that applies global random search to both the training set and the controller parameter set to further speed up the optimization process. The simulation results with a 4D inverted pendulum show that the new approach is much faster than the original IBEDS when the initial training set is empty and the search needs to bootstrap itself.