【摘要】 标签样本少的条件下机电设备的准确故障诊断对于提高复杂机电设备的健康管理能力具有重要意义。针对标签样本少的条件下难以建立准确故障诊断模型的问题，在半监督生成对抗网络的基础上，将注意力模块引入生成对抗网络，并利用格拉姆角场将一维数据转换为二维图像；结合双向生成对抗网络特点，提出一种基于双重注意力机制及半监督双向生成对抗网络（S-BiGAN）的机电设备故障诊断模型，并以轴承数据为例进行验证。结果表明：与CNN-SVM、SGAN等算法相比，本文提出的模型能够提高样本生成质量和故障分类特征，有效解决标签样本少的情况下故障诊断问题，极大地提高了故障诊断准确率。更多还原

【Abstract】 The accurate fault diagnosis of mechanical and electrical equipment under the condition of limited label samples is of great significance for improving the health management ability of complex mechanical and electrical equipment. In response to the problem of difficulty in establishing accurate fault diagnosis models under the condition of limited label samples, the attention module is introduced into the generative adversarial network based on semi-supervised generative adversarial network, in which the Gramian angular field(GAF) is used to convert onedimensional data into two-dimensional images. In combination with the characteristics of bidirectional generative adversarial network, a semi-supervised bidirectional generative adversarial network(S-BiGAN) based on dual attention mechanism for fault diagnosis of electromechanical equipment is proposed, and the bearing data is taken as an example for verification. The results show that, compared with algorithms such as CNN-SVM and SGAN, the proposed model can improve the quality of sample generation and fault classification features, effectively solve the fault diagnosis problem in the case of fewer label samples, and greatly improve the accuracy of fault diagnosis.更多还原