【摘要】 为了进一步提高模拟电路故障诊断的正确率，提出一种基于多重注意力机制的一维卷积神经网络故障诊断方法。该方法针对数据的维度调整网络参数，采用自适应力矩估计算法为不同的参数设计独立的自适应性学习率从而训练网络，并引入多重注意力机制以增强网络提取特征的能力，从而提高诊断正确率。实验结果表明，在对Sallen-Key滤波器电路诊断测试时正确率达到100%，在对四运放双二阶滤波电路进行故障诊断时，该方法仍具有99.74%的正确率，相比不添加注意力机制的方法高出1.8%，展现出较强的诊断能力。更多还原

【Abstract】 A one-dimensional convolutional neural network fault diagnosis method based on multiple attention mechanism is proposed to further improve the accuracy of fault diagnosis of analog circuits. In this method, the network parameters are adjusted according to the dimensions of the data, the adaptive moment estimation algorithm is used to design independent adaptive learning rates for different parameters to train the network, and a multiple attention mechanism is introduced to enhance the ability of the network to extract features, so as to improve the diagnostic accuracy rate. The experimental results show that the diagnosis accuracy rate of the Sallen-Key filter circuit reaches 100%, and the method still has an accuracy rate of 99.74%when troubleshooting the quad op amp dual second-order filter circuit, which is 1.8% higher than the method without adding attention mechanism. The result shows that the method has strong diagnostic ability.更多还原