An Intelligent Diagnosis Method for Emulsion Pump Faults Based on PCA and Bi-LSTM

Aiming at the problem that the fault of underground emulsion pump is difficult to diagnose under strong background noise, an intelligent fault diagnosis model of emulsion pump based on principal component analysis and bidirectional long-term memory network is proposed. Initially, the time-domain and frequency-domain features of the emulsion pump state data amalgamate to form a diverse set of features, thereby amplifying the saliency and robustness of the input features. Then principal component analysis is used to further reduce the dimension to improve the stability of features. On this basis, the accuracy of fault classification of emulsion pump is improved by using Bi-directional Long Short-Term Memory. The test data from the test bench were used for verification. The results show that the proposed method has high classification and recognition accuracy.