SVDフィルタを用いたカルマンフィルタとニューラルネットワークによる磁極形状最適化

概要

論文の詳細を見る
This paper presents the shape optimization of a magnetic pole by using Kalman filter that uses Signal Value Decomposition(SVD)and neural network. For the calculation of Kalman filter,the UD decomposition filter is generally used. The issue that the calculation of Kalman filter using UD decomposition filter is complicated is pointed out. Therefore, in this study, the simplification of the algorithm of Kalman filter is carried out with the use of the SVD filter, and Kalman filter is calculated with it. As the result, it is clear that the calculation of Kalman filter using the SVD filter shows almost the same result with what the UD decomposition filter shows. In addition,it is shown that the magnetic pole shape optimization combined use of Kalman filter using the SVD filter and neural network is effective.
2006-02-28