Pumping unit fault audio diagnosis method based on big data analysis

The invention relates to an oil pumping unit fault audio diagnosis method based on big data analysis. The method sequentially comprises the following steps that A, a background analysis model of oil pumping unit faults is established; b, acquiring real-time audio data of the oil pumping unit; c, sending the background analysis model and the real-time audio data into a real-time identification module for comparison and diagnosis; and D, outputting a diagnosis result. The step A specifically comprises the following steps: A1, collecting effective audios of the oil pumping unit in batches, wherein the effective audios comprise normal audios and various fault audios; a2, transmitting the audio data to a background and preprocessing the audio data; a3, marking the normal audio and the fault audio respectively; a4, acoustic features of various audios are extracted; a5, storing in a sample library and training; and A6, outputting parameters required by the real-time identification module. According to the method, te