本发明属于机器学习技术领域，公开了一种基于可迁移共特征空间挖掘的故障诊断方法，故障诊断方法的过程包括：对故障诊断的振动序列数据进行移动窗口截取，构建故障诊断训练集和测试集；构建可迁移共特征空间挖掘卷积自编码器；结合领域弱监督损失、领域自适应损失和重建损失训练卷积自编码器；在卷积自编码器训练所的特征表示基础上，构建共特征提取与比较卷积网络；应用三种小样本学习场景对特征提取与比较网络进行训练，获得可迁移共特征故障诊断模型，利用所述故障诊断模型进行故障诊断。

The invention belongs to the technical field of machine learning, and discloses a fault diagnosis method based on migrating common feature space mining. The process of the fault diagnosis method includes: intercepting vibration sequence data of fault diagnosis by moving window to construct a fault diagnosis training set and a test set; Constructing a convolution self-encoder for migrating common feature space mining; Combining domain weak supervision loss, domain adaptive loss and reconstruction loss to train convolutional self-encoder; Based on the feature representation of convolution self-encoder training institute, a common feature extraction and comparison convolution network is constructed. Three small sample learning scenarios are used to train the feature extraction and comparison network, and a transportable common feature fault diagnosis model is obtained, and the fault diagnosis model is used for fault diagnosis.