摘要
针对风电机组轴承疲劳实验成本高导致故障数据不足的问题,提出基于粒子群算法(PSO)优化的辅助分类器生成对抗网络(ACGAN),利用PSO对ACGAN的参数进行寻优,进而利用ACGAN生成与原始样本高度相似的新样本;针对单一模型对风电机组轴承故障诊断的准确率较低的缺点,引进合作博弈理论对多个子模型的诊断结果进行融合,将各个子模型的诊断概率矩阵由合作博弈理论进行融合并输出最终的诊断结果。实验证明,优化后的ACGAN模型和基于合作博弈的模型融合能有效提高轴承故障诊断的准确率。
Auxiliary classifier generative adversarial network(ACGAN)based on particle swarm optimization(PSO)was proposed to solve the problem of insufficient fault data caused by high cost of wind turbine bearing fatigue experiments.The parameters of ACGAN were optimized by PSO,and then ACGAN was used to generate new samples that were highly similar to the original samples.In view of the low accuracy of a single model for wind turbine bearing fault diagnosis,the cooperative game theory was introduced to fuse the diagnostic results of multiple sub-models,and the diagnostic probability matrix of each sub-model was fused by the cooperative game theory and the final diagnostic results were output.Experimental results show that the optimized ACGAN model and the model fusion based on cooperative game can effectively improve the accuracy of bearing fault diagnosis.
作者
李俊卿
胡晓东
王罗
马亚鹏
何玉灵
Li Junqing;Hu Xiaodong;Wang Luo;Ma Yapeng;He Yuling(Department of Electric Power Engineering,North China Electric Power University,Baoding 071003,China;China Three Gorges Corporation,Wuhan 430010,China;Department of Mechanical Engineering,North China Electric Power University,Baoding 071003,China)
出处
《太阳能学报》
EI
CAS
CSCD
北大核心
2024年第1期234-241,共8页
Acta Energiae Solaris Sinica
基金
国家自然科学基金(52177042)。
关键词
风电机组
轴承
生成式对抗网络
故障诊断
模型融合
合作博弈
wind turbines
bearings
generative adversarial network
fault diagnosis
model fusion
cooperative game
作者简介
通信作者:李俊卿(1967-),女,博士、教授,主要从事电机及其系统分析和电气设备故障诊断方面的研究。junqing03@163.com。
