关键词: 动力学反解算法, 设备动态诊断, 神经网络

Abstract: The basic strategy of algorithm for inverse structure dynamics problems based on neural network are discussed, the fast-learning algorithm, samples selection method by orthogonalization and adaptive modification of inverse-mapping results of BP neural network are presented, which can overcome some difficulties encountered in BP neural network used in engineering practice. Two engineering application examples on equipments dynamic diagnosis are given and results are promising.

Key words: Algorithm for inverse structure dynamics problem, Equipments dynamic diagnosis, Neural network