为识别光伏组件故障类型,提高光伏系统发电效率,提出了一种基于改进CNN-SVM模型的光伏组件红外图像故障诊断方法.首先以光伏组件红外图像为输入样本构建改进CNN模型,采用全局平均池化层代替传统CNN模型的全连接层,在进行图像特征提取的同时降低模型参数量;利用数据增强和批归一化技术提高模型泛化能力,降低模型过拟合.其次采用非线性支持向量机SVM代替传统CNN模型中的Softmax分类器,以提高光伏组件红外图像故障识别准确率.最后采用Infrared Solar Modules数据集对所提模型进行了实例验证.结果表明:与传统CNN模型相比,改进CNN-SVM模型故障诊断准确率高,对各故障类型的识别能力强.
Fault Diagnosis Method of Photovoltaic Modules Infrared Image Based on Improved CNN-SVM
In order to identify the fault types of photovoltaic modules and improve the power generation efficiency of photovoltaic system,we propose an infrared image fault diagnosis method of photovoltaic modules based on improved CNN-SVM model.Firstly,we used the infrared images of photovoltaic modules as input samples to construct an im-proved CNN model,and replaced the fully connected layer of the traditional CNN model with the global average pooling layer,which reduced the number of model parameters while extracting image features.Besides,we used data enhance-ment and batch normalization technique to improve the generalization ability of the model and reduce the over fitting of the model.Secondly,we used the nonlinear support vector machine SVM to replace the Softmax classifier in the tradi-tional CNN model to improve the accuracy of infrared image fault recognition of photovoltaic modules.Finally,we used the Infrared Solar Modules data set to verify the proposed model.The results show that compared with the traditional CNN model,the improved CNN-SVM model has high fault diagnosis accuracy and strong recognition ability of various fault types.
万方数据
维普
