Research And Application Of Rolling Bearing Diagnosis System Based On Multi-feature Reduction Of Vibration Signal

With the increasing level of industrialization,the application of modern machinery is more and more extensive.Rolling bearing is one of the common parts of machinery.According to statistics,a considerable part of mechanical failure is caused by the failure of rolling bearing.Rolling bearings are often in a complex and harsh production environment.The vibration signals generated during operation are non-linear and nonstationary,and can be disturbed by noise in the environment.These factors increase the difficulty of monitoring and fault diagnosis of rolling bearings.Therefore,this paper focuses on the decomposition of vibration signals,the extraction and dimensionality reduction of characteristic signals,and the identification of fault types.The main research contents are as follows:(1)Aiming at the non-linear and non-stationary characteristics of rolling bearing vibration signals,an intrinsic time-scale decomposition(ITD)vibration signal processing method is proposed.The obtained component signals are reconstructed according to the kurtosis principle to reconstruct a new special diagnosis signal sequence,and then multi-dimensional time-frequency domain features are extracted from the new reconstructed signals,and a high-dimensional feature information group is established.The feature information is used to establish a genetic algorithm optimized support vector machine(GA-SVM)fault diagnosis model,and the diagnosis model is established to effectively recognize the bearing state.(2)To eliminate the redundancy of multi-dimensional feature information,a dimension reduction method based on t-distributed stochastic neighborhood embedding(t-SNE)is proposed.The non-linear dimensionality reduction of multi-dimensional feature is carried out using t-SNE algorithm.Finally,the low-dimensional feature is used to build a fault diagnosis model based on(GA-SVM).The experimental results show that the t-SNE algorithm can effectively reduce information redundancy and improve the recognition rate of fault diagnosis.(3)A WIFI-based real-time monitoring system for bearing operation is designed and implemented.The slave computer in the system is the acquisition module,which mainly collects and transmits real-time vibration signals.The host computer is the signal analysis module,which is used to receive vibration signals and data processing.In the signal acquisition module,MPU6500 is used as vibration signal sensor,STM32 microcontroller is used as the core,and WIFI transmission technology is used to transmit vibration signal to the host computer.The received vibration signal is diagnosed by the host computer,and the running state of the bearing is monitored.In this paper,rolling bearing is taken as the research object to solve the technical difficulties of signal feature extraction,signal redundancy reduction,fault type identification and so on.It has certain application value in practical production.