Fault diagnosis of multi-sensor oil system of road header based on convolution neural network

To address the limitations of single-sensor data in providing comprehensive and multidimensional insights into the oil systems of tunneling machines during long-term operations, we propose a fault diagnosis method based on convolutional neural networks (CNNs). Firstly, oil monitoring methods assess the performance and condition of the oil by analysing its physical an d chemical properties, as well as the information carried by wear particles. This approach is unaffected by environmental noise and directly extracts optimal features from the oil data without the need for preliminary data conversion or manual feature extraction. Subsequently, the data collected from multiple sensors are utilized as input to the CNN, enabling the diagnosis o f faults in the oil system. Finally, we compare the fault state classification accuracy of our method with that of other machin e learning methods. The experimental results demonstrate that the diagnostic accuracy of our proposed method surpasses that of competing methods, thereby achieving efficient diagnostic performance.