The Cad For Rare-earth Permanent Magnet Synchronous Machine And Research On Optimization Method

In recent years, the shortages of the energy have become an international problem for common concern. With the development of permanent magnet materials, the permanent magnet (PM) machine shouldbe applied more and more widely because of its excellent efficiency, such as: high power density, low runcost, good reliabilityand efficient of conserving energy etc.In engineer, the initial scheme would be given by the engineer who has richexperiences when the electrical machine is designed. With the retirement of them, the mode can't be operative. The work process of Artificial Neural Networks (ANN) is very similarly the human, so the current design mode can be perfect simulated by ANN.The design methods for the rare-earth permanent-magnet synchronous motor (RPMSM) are studied and discussed in this thesis. This paper gives the classical design methods, and some optimization methods, such as, Genetic Algorithm (GA), Expert System (ES). The paper also explains the principles of optimization algorithms and describes their different characteristics. Based on the studying the rare-earth permanent-magnet synchronous motor, a software had been developed, in which the object-oriented programming method and modularization design technique are used. The soft can run on the platform of Windows 9x/Windows NT. Secondly, a multi-layer neural network with adjustable parameters and an improved back-propagation learning algorithm (BP) based on LM rules are applied in the paper. The permanent-magnet synchronous motor samples are trained with the analogy method in ANN, by which the generalization capability ofparameters is realized.The simulation results show that the electrical machine design based on this method is a feasible and effective way.