Pumping unit fault analysis method based on wavelet transform time-frequency diagram and CNN

Abstract

According to the characteristics of the signal of the pumping unit, the processing result of the wavelet packet threshold de-drying algorithm is not good, so this paper proposes a new method to improve the algorithm. Experiments show that the improved algorithm improves the signal-to-noise ratio by 2.2 dB compared to the traditional algorithm, and the root mean square error decreases. Then the signal of the pumping unit is (CWT), the time-frequency diagram is obtained, and it is displayed in the form of grayscale image. Then the time-frequency diagram is input as the feature map, and the CNN classifier model is established to realize the intelligent fault of the pumping unit. diagnosis. The results show that the method can effectively identify the fault type of the pumping unit and has a high recognition rate.