Study On Fault Diagnosis Technology Of Rotating Equipment Based On Time-frequency Features

Rotating equipment is the most widely used equipment in the manufacturing industry,and its reliability has a profound impact on the production and development of the manufacturing industry.As the complexity and precision of the equipment structure continue to increase,the direct or indirect losses caused by faults are more and more unbearable.Therefore,fault diagnosis technology has important practical significance for improving equipment reliability and maintenance efficiency.In view of this,this paper studies the time-frequency feature extraction technology,the operating state anomaly detection technology,the equipment fault diagnosis technology and the adaptive fault diagnosis technology under multiple working conditions in the fault diagnosis of rotating equipment,and uses the rolling bearing and milling cutter vibration data to verify the effectiveness of the proposed method.In this paper,a signal feature extraction method based on multi-resolution and time-frequency domain analysis is proposed,and the isolation forest algorithm is used to detect the abnormality of the state signal.This method uses discrete wavelet transform to decompose the signal into signal components of different frequency bands,and extracts dominant components for signal reconstruction.Using the continuous wavelet transform method to transform the reconstructed signal into timefrequency domain,and the feature matrix in the time-frequency domain of the signal is obtained.The isolation forest algorithm is used for anomaly detection of timefrequency features abnormal signal samples are obtained.The effectiveness of the proposed method is verified by the rolling bearing vibration signal.The fault diagnosis technology of equipment vibration signal based on convolutional neural network is studied to solve the locating problem of complex equipment fault sources.In this paper,using the time-frequency feature matrix of typical fault vibration signal as the input of the convolutional neural network model.The influence of structural parameters on the diagnostic performance is analyzed and a reasonable fault diagnosis model is established.In the model training process,the data enhancement and regularization methods are used to optimize the diagnostic model to improve its robustness and generalization.The effectiveness of the proposed method is verified by the rolling bearing and milling cutter vibration signal,and further compared with the traditional method to prove the superiority of the proposed method.This paper studies the fault diagnosis technology based on transfer learning to solve the problem of small sample fault diagnosis.The convolutional neural network model is used as the basic diagnosis model,and a large number of samples under known working conditions are used as source data sets for transfer learning to complete the model pre-training,and then the model is migrated to the new working condition small sample fault diagnosis task.The transfer diagnostic model is slightly adjusted using the new sample small sample data set to realize the migration of the cross-condition diagnosis model.Finally,the diagnosis results of the migration model are verified by the rolling bearing fault signals under various load conditions.The research in this paper has theoretical significance for enriching the fault diagnosis method of complex rotating equipment.It has engineering practical value for improving equipment fault diagnosis accuracy,improving equipment reliability and maintenance efficiency.