ON-LINE SPEED CONTROL OF PERMANENT-MAGNET SYNCHRONOUS MOTOR USING SELF-CONSTRUCTING RECCURENT FUZZY NEURAL NETWORK

• Main Authors: Yi-Kai Huang, 黃翊愷
• Other Authors: Hung-Ching Lu
• Format: Others
• Language: en_US
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/48754607760816862486

碩士 === 大同大學 === 電機工程學系(所) === 94 === Combining the merits of the self-constructing fuzzy neural network (SCFNN) and the recurrent neural network (RNN), this thesis is proposed to a self-constructing recurrent fuzzy neural network (SCRFNN). Two learning phases are adopted in the proposed network. One is the structure learning phase which is to the partition of input space. The other is the parameter learning which is based on the supervised gradient-decent method using a delta law. The SCRFNN is applied to control the speed of a permanent--magnet synchronous motor to track periodic reference trajectories In addition, we use fuzzy-neural network to determination of a suitable error term and to train the parameter of the SCRFNN on-line. Finally, the simulation results show that the control effort and chattering of the SCRFNN are smaller than those of SCFNN.