Research on the fault identification method of oil pumping unit based on residual network

Oil extraction by pumping units is still the mainstream production method in China's oil fields, and the indicator diagram is an important basis for judging the working condition of pumping units. Analysis of indicator diagrams is the most common and effective method for diagnosing pumping unit faults. However, the traditional analysis of indicator diagrams mainly relies on manual experience, the workload is large and the recognition accuracy is easily affected by human factors. This paper proposes a model based on the ResNet-34 residual network to identify the indicator diagrams, which adds a residual block structure to the traditional convolutional neural network to establish a direct connection between the upper layer input and the lower layer output and achieves the recognition and classification of six power diagrams through parameter adjustment. The training accuracy reached 91.0%, which is about 7.7 percentage points higher than that of the AlexNet network, and the experiment shows that the model has high diagnostic accuracy.