基于GA-BP的船舶同步发电机定转子绕组匝间短路故障诊断研究  

GA-BP-based diagnosis of inter-turn short-circuit fault diagnosis of ship rotor synchronous generator

作　　者：孙卫鹏 徐合力[1] 高岚[1] SUN Weipeng;XU Heli;GAO Lan

机构地区：[1]武汉理工大学能源与动力工程学院,湖北武汉430063

出　　处：《中国修船》2020年第4期48-54,共7页China Shiprepair

摘　　要：为诊断与分析船舶同步发电机定转子绕组匝间短路故障,文章采用基于主成分分析法(PCA)和遗传算法(GA)优化BP神经网络(GA-BP神经网络)的故障诊断方法。首先利用Maxwell软件平台故障仿真得到的定子三相电流作为特征信号,通过小波包分解重构以及PCA降维的处理方式,生成15维的样本数据,降低了网络规模以及处理计算时间,并针对传统BP神经网络收敛速度慢以及易陷入局部极小值的特点,利用GA算法对BP神经网络权值与阈值进行优化。通过样本数据对GA-BP神经网络进行训练测试,验证了PCA和GA-BP神经网络对于船舶同步发电机定转子匝间短路故障诊断具有可行性以及准确性。In order to diagnose and analyze the turn-to-turn short-circuit fault of the stator rotor of the ship's synchronous generator,the article adopts the fault diagnosis method based on the principal component analysis(PCA)and genetic algorithm(GA)to optimize the BP neural network(GA-BP neural network).Firstly,the three-phase stator current obtained from the fault simulation of the Maxwell software platform is used as a characteristic signal.Through wavelet packet decomposition and reconstruction and PCA dimensionality reduction processing,15-dimensional sample data is generated,which reduces the network scale and processing calculation time.BP neural network has the characteristics of slow convergence speed and easy to fall into local minimum value.The GA algorithm is used to optimize the weight and threshold of BP neural network.Through the sample data,the GA-BP neural network was trained and tested,and the feasibility and accuracy of the PCA and GA-BP neural network for the ship synchronous generator stator-to-rotor short-circuit fault diagnosis were verified.

关 键 词：发电机 故障诊断 匝间短路 遗传算法 BP神经网络 

分 类 号：U665.1[交通运输工程—船舶及航道工程]