| Summary: | 碩士 === 國立虎尾科技大學 === 電機工程研究所 === 100 === This dissertation proposes two algorithms for functional neurofuzzy systems (FNS) in predictive problems. The two algorithms are including the cluster-based tribes optimization algorithm (CTOA), and the tribal particle swarm optimization (TPSO). This dissertation consists of the two major parts. In the first part, the CTOA method is presented for the FNS model. The CTOA adopts a self-clustering algorithm (SCA) to divide a swarm into multiple tribes and uses various evolutionary strategies to update each particle. Furthermore, the CTOA also uses an adaptation mechanism to generate or remove particles and reconstruct tribal links. The adaptation mechanism can improve the qualities of the tribe and the tribe adaptation. In the second part, the TPSO method is presented to balance the local and global exploration of the search space effectively. The evolutionary strategies of TPSO have three different types of equations that are developed base on PSO according to the status of each particle to design. Finally, the proposed two algorithms for FNS model are applied in various predictive problems. Results of this dissertation demonstrate the effectiveness of the proposed algorithms.
|
|---|
