Research On Rotating Machinery Fault Diagnostics Based On Multiscale Geometric Analysis

With the rapid development of modern social economy and science technology,there are more and more requirements for the safe operation of production equipment.In the rotating machinery,gears and bearings are the core parts of the rotating part,and it has a great significance to find the rotating machinery failure and improve the reliability of rotating machinery in timely and accurate access to the core part of the running state information and identify the fault information.In this paper,the method of time frequency image processing for vibration signal of rotating machinery is studied,the time frequency image of the vibration signal is taken as the processing object,and the characteristic information of the running state of the equipment is extracted to classify the different fault states,the main contents are as follows:(1)The time-frequency joint analysis method can fully express the time domain and frequency domain information of the vibration signal.Therefore,this paper expounds the advantages of wavelet transform time-frequency analysis from the aspects of time and space resolution and multi frequency signal analysis,and the wavelet transform is used to obtain the time-frequency image of the vibration signal for the prepare of the feature extraction.At the same time,some methods and classification technology of time-frequency image feature extraction are summarized.(2)Elaborating the geometric analysis of Contourlet transform and its algorithm which has been improved nonsubsampled contourlet transform.The paper proposes a feature extraction method for grayscale map based on the nonsubsampled contourlet transform which can be applied to gear and bearing fault classification.The results show the effectiveness of the proposed method.(3)With the combination of local binary pattern and nonsubsampled contourlet transform to process the feature extraction of time-frequency image,the classification accuracy for method one will be improved.That Introducing local binary pattern into the analysis of time-frequency image can be very helpful to characterize the spatial structure of local texture and extract of the texture image effectively.Therefore,the classification accuracy is improved obviously when he method is applied to gear and bearing fault classification with the less training samples.(4)Based on the national instruments virtual instrument platform Lab VIEW,Program the data acquisition system and analysis system.The acquisition and storage of vibration signal is finished by using the equipment such as the sensor and acquisition card.Program the offline analysis module for the time-frequency image of rotating machinery vibration signal based on the study of feature extraction and classification method mentioned in this paper.The practicability of the system is verified by field test data.