摘要
为提高变压器油溶解气体分析法的故障诊断能力,提出了加动量批处理小波神经网络算法。实验结果表明,与传统故障诊断方法相比,该算法有效地避免了陷入局部极小、振荡和发散问题,提高了学习速度、稳定性和故障诊断效率,具有很强的可行性和有效性。
For improving the gases dissolved in transformer oil analysis of the fault diagnosis ability,the paper put forward a batch add momentum wavelet neural network algorithm.The experimental results show that,compared with the traditional fault diagnosis method,the proposed algorithm can effectively avoid into the local minimum,oscillation and divergent problems and improve the learning speed,stability and fault diagnosis efficiency,has the very strong feasibility and effectiveness.
出处
《科技通报》
北大核心
2012年第8期162-164,共3页
Bulletin of Science and Technology
关键词
小波变换
神经网络
故障诊断
wavelet transform
neural network
fault diagnosis
