Fuzzy measure-based fuzzy rule interpolation based on PSO-based fuzzy integral-learning techniques

In this paper, we propose a fuzzy measure-based fuzzy interpolative reasoning method for sparse fuzzy rule-based systems. A particle swarm optimization algorithm (PSO)-based fuzzy integral-learning technique is employed. The proposed method is able to deal with fuzzy rule interpolation with fuzzy measure-based antecedent variables and fuzzy rule interpolation based on polygonal membership functions. The optimal fuzzy density of the antecedent variables and the fuzzy rules are automatically learnt by a PSO-based fuzzy integral-learning algorithm. The proposed fuzzy measure-based fuzzy interpolative reasoning method and the proposed PSO-based fuzzy integral-learning algorithm are applied to deal with the truck backer-upper control problem. Based on statistical analysis techniques , experimental results show that the proposed fuzzy measure-based fuzzy interpolative reasoning method with fuzzy density optimized by PSO-based fuzzy integral-learning algorithm yields statistically significantly smaller error rates in comparison to existing methods.