基于CEEMD与改进的ELM旋转整流器故障诊断

Fault diagnosis of synchronous generator rotating rectifier based on CEEMD and improved ELM

针对目前应用于航空发电机旋转整流器故障诊断中的人工智能算法存在诊断速度慢、参数选取困难等问题,提出一种基于互补式集合经验模态分解(CEEMD)与樽海鞘优化的极限学习机(SSA-ELM)故障诊断方法。在有限元软件Maxwell与Simplorer中搭建三级式电机模型,采集励磁电流信号,利用CEEMD将励磁电流信号分解为一系列模态分量,构建故障特征参量,再通过樽海鞘群算法(SSA)优化极限学习机的训练参数ω和b,并对故障进行诊断,最后通过实验平台验证所提方法。结果证明了三级式同步电机有限元模型的有效性,所提方法相校于现有方法,具有更高的故障诊断准确率与分类速度。

Aiming at the problems of slow diagnostic speed and difficulty in pardmeter selection in artifical intelligence algorithms currently used in fault diagnosis of aviation generator rotating rectifiers,an extreme learning machine(SSA-ELM)and complementary ensemble empirical mode decomposition(CEEMD)based fault diagnosis approach is investigated.In the finite element software Maxwell and Simplorer,the three-stage motor model is built,the excitation current signal is collected,the excitation current signal is decomposed into a series of modal components by CEEMD,and the fault characteristic parameters are constructed.Then the training parameters and parameters of the extreme learning machine are optimized by the SSA algorithm,and the fault is diagnosed.Finally,it is verified by the experimental platform.The results show the effectiveness of the three-stage synchronous motor finite element model.Compared with the existing methods,the rotating rectifier diode fault diagnosis method based on CEEMD and SSA-ELM has higher fault diagnosis accuracy and classification speed.