Improved Adaptive Particle Swarm Optimization for Optimization Functions and Clustering Fuzzy Modeling System

摘要

In this paper, a new PSO algorithm with a new adaptive weight of inertia and time acceleration coefficients (TVAC) is proposed; this algorithm called IAPSO is introduced for global optimization. The objective of this proposition is to initialize the weight of inertia to a high value, giving priority to the global exploration of the research space and gradually decreasing the new inertia adaptable to the weight in order to obtain refined solutions. The test of our algorithm is performed on three standard reference functions (Schwefel's (unimodal), Ackley (multimodal) and Griewank (Multimodal)). The proposed IAPSO algorithm combined with the Fuzzy Clustering NPCM algorithm for modeling and identifying a non-linear system. The new NPCM-IAPSO grouping algorithm also solves the problems of the classical clustering algorithm (FCM, GK, PCM, EPCM, FCM-PSO, EPCM-PSO ... etc.), such as convergence towards local optimization and sensitivity to initialization. The effectiveness of the proposed NPCM-IAPSO algorithm was tested on the furnace gas Box and Jenkins, dryer system and two other nonlinear systems described by differential equations.