Fault Diagnosis Method of Electric Deep Well Pump Based on CEEMDAN-CNN

The electric deep well pump is the core equipment on the tanker, and it is the key execution link when the tanker is working. However, due to its harsh working environment and complex structure, it is prone to malfunctions such as abnormal vibration and loose support, resulting in performance degradation or even failure to work properly. This paper proposes a CEEMDAN-CNN-based fault diagnosis method for electric deep-well pumps. First, combined with wavelet threshold denoising and CEEMDAN method, the original vibration signal is de-noised and decomposed. Then, the CNN model is used to diagnose and classify the typical faults of deep-well pumps, and compared with Traditional SVM, Bayes and decision tree methods. The results show that the model training time of the CEEMDAN-CNN method proposed in this paper is reduced by nearly 60%, and the test accuracy rate is as high as 99.7%.