High-precision fault diagnosis of rotating machinery plays an important role in industrial systems. Today, rotating machinery often has multiple sensors to monitor equipment condition, so it is important to fuse data from multiple rotating machinery sensors for fault diagnosis. Most of the current multi-sensor fusion fault diagnosis methods are single-level, which cannot fully utilize the effective information in multi-sensor data, and have certain limitations. Therefore, this article proposes a multi-level information fusion fault diagnosis method. Specifically, first, multilayer graph data is constructed by analysing the correlation between different sensors as well as samples to realize multi-sensor data level fusion. Then, convolutional neural network and graph convolutional network are used to extract different types of features from the data, and a feature fusion method based on the attention mechanism is proposed to realize feature enhancement. Finally, a decision fusion strategy based on information entropy is established to reduce the impact of misclassification results and maximize the reliability and robustness of the model output. Experimental validation of the proposed method is carried out using the publicly available dataset and pumped storage unit operational state data. The experimental results show that the proposed strategy is of positive significance to improve multi-sensor data fusion fault diagnosis, and the model has a higher diagnostic accuracy compared with other multi-sensor data fusion fault diagnosis models.

中文翻译：

---

基于多传感器信息多级融合框架的旋转机械故障诊断方法


旋转机械的高精度故障诊断在工业系统中发挥着重要作用。如今，旋转机械通常具有多个传感器来监测设备状态，因此融合多个旋转机械传感器的数据以进行故障诊断非常重要。目前的多传感器融合故障诊断方法大多是单层次的，不能充分利用多传感器数据中的有效信息，存在一定的局限性。因此，本文提出一种多层次信息融合故障诊断方法。具体来说，首先通过分析不同传感器和样本之间的相关性，构建多层图数据，实现多传感器数据层次融合。然后，利用卷积神经网络和图卷积网络从数据中提取不同类型的特征，并提出基于注意力机制的特征融合方法来实现特征增强。最后，建立基于信息熵的决策融合策略，以减少误分类结果的影响，最大限度地提高模型输出的可靠性和鲁棒性。使用公开数据集和抽水蓄能机组运行状态数据对所提出的方法进行了实验验证。实验结果表明，该策略对提高多传感器数据融合故障诊断具有积极意义，且该模型与其他多传感器数据融合故障诊断模型相比具有更高的诊断准确率。