Research On Feature Recognition Of Bearing Acoustic Emission Signal

Rolling bearing is the most basic and core component in the structure of rotating machinery.It is also the most vulnerable part to damage and failure.Therefore,ensuring the normal operation of rolling bearing is an important part of the safety work of rotating machinery.Compared with other non-destructive testing technologies,acoustic emission testing technology can detect early weak faults of rolling bearings in time,and monitor the occurrence and development of faults in time.It has high safety and reliability.In the era of big data and artificial intelligence,traditional artificial signal analysis and processing technology has been unable to meet the needs of intelligent integrated fault diagnosis of large-scale machinery and equipment.Thus,the research on intelligent recognition and fault diagnosis of acoustic emission signals of rolling bearings has a very important practical significance.In this paper,the principle of acoustic emission signal generation and the physical significance of acoustic emission signal parameters are analyzed,and the principle of acoustic emission signal generation and the physical significance of acoustic emission signal parameters are analyzed.Based on the original acoustic emission signal data of rolling bearings,the characteristics and performance of the recurrent neural network model are deeply studied,and the bidirectional long short-term memory networks model method is constructed.The essential features and deep mapping relationship of original data of rolling bearings under different faults are fully excavated,which avoid relying on complex data processing technology and signal analysis technology to extract and select fault features.End-to-end adaptive extraction and intelligent diagnosis of acoustic emission signal characteristics of rolling bearings faults are completed.At the same time,the recognition performance of the bidirectional long short-term memory networks model for fault acoustic emission signal under variable operating conditions is explored.Starting from the characteristics of acoustic emission signal distribution under different working conditions,a combination of transfer learning and bidirectional long short-term memory networks is introduced,which breaks the requirement that the training set and testing set of traditional machine learning methods must satisfy the independent and identical distribution,Getting rid of over-reliance on prior fault data,which can identify the acoustic emission signals of various fault types under various types of working conditions.Besides,it greatly shortens the time of fault identification and reduces the cost of calculation,which makes it possible to monitor rolling bearing fault online based on acoustic emission detection technology in time.The online monitoring software of rolling bearing fault based on acoustic emission detection technology is designed.From the point of view of software development technology and actual monitoring function requirements,the whole software system has four major functional modules: acoustic emission signal acquisition module,data import and processing module,model function selection and training module,real-time monitoring evaluation and alarm module.Finally,the software system and the function of each module are tested several times and continuously improved and optimized.The experimental results show that the software system can monitor the running state of rolling bearings in real time and stably for a long time,and complete the identification and diagnosis tasks of the corresponding running state.