【摘要】 针对风电机组的故障诊断问题，提出一种基于改进卷积神经网络(convolutionalneuralnetwork,CNN)和双向门控循环单元(bidirectional gated recurrent unit,BiGRU)双通道特征融合的风电机组故障诊断模型，该模型有别于传统的串联式结构，采取了并联式结构将改进的CNN和BiGRU进行结合．首先，利用批量归一化(batchnormalization,BN)层代替传统CNN中的Dropout层，CNN作为第1个通道提取特征；其次，给传统BiGRU添加1个多层感知机网络，BiGRU作为第2个通道提取特征；最后，通过识别层的特征融合层将两通道连接起来，利用支持向量机代替传统的Softmax层进行故障分类．实验结果表明，相较于其他模型，该模型的准确率更高，整体效果更好．更多还原

【Abstract】 Aiming at the problem of wind turbine fault diagnosis,a wind turbine fault diagnosis model based on dual channel feature fusion of improved convolutional neural network(CNN)and bidirectional gated recurrent unit(BiGRU)is proposed in our study. The model is different from the traditional series structure and adopts a parallel structure to combine the improved CNN and BiGRU. Firstly,the batch normalization layer is used to replace the Dropout layer in the traditional CNN,and CNN is used as the first channel to extract features. Secondly,a multi-layer perceptron is added to the traditional BiGRU,and BiGRU is used as the second channel to extract features. Finally,the two channels are connected through the feature fusion layer of the recognition layer,and then the support vector machine is used to replace the traditional Softmax layer for fault classification. The results of the experiments show that compared with other models,this model has higher accuracy and better overall effect.更多还原