A reciprocating compressor wear state recognition method, system, device and medium

Abstract

The application discloses a kind of reciprocating compressor wear state identification method, system, equipment and medium, the method is fused by the similarity between the historical sound signal and historical vibration signal of reciprocating compressor, and the sound vibration fusion signal obtained is used for the training of auditory scene analysis algorithm model after denoising reconstruction processing, to generate analog signal to expand signal set;The generated analog signal is detected by self-organizing mapping network model, the availability of analog signal is evaluated, and the auditory scene analysis algorithm model and data generation process are iteratively optimized based on the evaluation results to construct signal dataset;The neural network model is trained using the signal dataset, obtaining the state recognition model for the key position of reciprocating compressor, the sound vibration fusion signal to be identified is used as the input of the state recognition model, and the wear state of the key position of reciprocating compressor is output;Effectively improve the accuracy of compressor wear state identification.

Images