摘要
针对光伏并网三相电压型逆变器开关管的开路故障,提出深度级联模型(deep cascade mode,DCM)‒主成分分析(principal component analysis,PCA)与遗传算法(genetic algorithm,GA)优化的BP神经网络结合的故障诊断方法。首先对逆变器的开路故障进行分析和仿真,确定三相电流作为故障信号,选择22类故障状态作为诊断对象,通过以稀疏表示分类(sparse representation based classififier,SRC)为基本操作单元的深度级联模型提取故障特征,DCM根据层次学习特性将故障特征分层,再由SRC部分得到不同故障的编码系数,并采用t分布—随机近邻嵌入(t⁃distributed stochastic neighbor embedding,t⁃SNE)方法验证了DCM具有较好的特征提取能力,通过PCA降低故障特征的冗余度、保留有价值的主成分提高网络映射能力,最后将故障特征向量作为GA⁃BP神经网络的输入信号实现对故障的诊断识别。通过仿真实验得到该方法的故障诊断准确率为95.64%,与DCM⁃PCA⁃BP、FFT⁃GA⁃BP和FFT⁃BP相比准确率分别提高8.71%、20.64%、51.70%,表明该方法有更好的故障特征提取能力和故障诊断效果。
Aiming at the open-circuit fault of the photovoltaic grid-connected three-phase voltage-type inverter,.a fault diagnosis method combining deep cascade mode-principal component analysis(DCM-PCA)and genetic algorithmoptimized BP(GA-BP)neural network is proposed.Firstly,the open-circuit fault of the inverter is analyzed and simulated,the three-phase current is determined as the fault signal,and 22 types of fault states are selected as the diagnosis objects,and the fault features are extracted through the deep cascade model with sparse representation classification as the basic operation unit,the DCM fault features are stratified based on the characteristics of hierarchical learning.The t-SNE method is used to verify that DCM has good feature extraction ability.PCA is used to reduce the redundancy of fault features,retain valuable principal components to improve the network mapping ability.Finally,the fault feature vector is used as the input of the GA-BP neural network to identify the fault and output the diagnosis result.The fault diagnosis accuracy of this method is 95.64%through simulation and experiments,compared with the DCMPCA-BP,FFT-GA-BP and FFT-BP,the accuracy is increased by 8.71%,20.64%and 51.70%respectively,indicating that the proposed method has better fault feature extraction capability and better fault diagnosis performance.
作者
黄敬尧
程煜
李雅恬
HUANG Jingyao;CHENG Yu;LI Yatian(College of Electrical Engineering and New Energy,China Three Gorges University,Yichang 443000,Chain)
出处
《电力科学与技术学报》
CAS
CSCD
北大核心
2024年第1期260-271,共12页
Journal of Electric Power Science And Technology
基金
湖北省自然科学基金(2019CFB331)。
关键词
逆变器
故障诊断
神经网络
深度级联模型
故障特征
inverter
fault diagnosis
neural network
deep cascade mode
fault characteristics
