Rolling Bearing Fault Diagnosis Based On Multi Sensor

Rolling bearing is one of the most widely used component in the mechanical equipment, its running state directly determines the fate of the whole machine. However, In the traditional method of bearing detection, we often use a single sensor. Not only the detected information rarely, but also often affected by the external signal. The result about accurate diagnosis rate is relatively low, and even cause wrong or wrongly. Using the different detection method of acceleration and acoustic emission which has a certain correlation and complementarity has the significant effect for analyzing and detecting the bearing fault. Therefore, in this article, we use the information fusion technology to fuse the acceleration and acoustic emission to diagnose the fault of rolling bearing.The article finishes the research of monitoring system about the fault of rolling bearing,and proposed a method of information fusion based on Neural Networks. This will improve the accuracy of detection to diagnose the fault of rolling bearing. Firstly, by examining sensor and signal input devices, AE signal and vibration signal’s acquisition system has been put up; Secondly, based on the Bearing vibration mechanism; Thirdly, analyze the data in time domain and frequency domain and time-frequency domain, and make use of the wavelet technology to information for noise reduction, finally, using the Hilbert technique to envelope demodulation signal information, by Contrasting the bearing fault frequency,we can judge the bearing fault type. The experimental results validate the effectiveness of the proposed method for rolling bearing fault diagnosis. We would also build a BP neural network, which is expected to help build up the mapping relationship, and the result can be used to detect the type of rolling bearing.In the fault detection of rolling bearing, we carried out active research and exploration for the sensor technology, filtering, envelope demodulation and neural network. In addition, we combine with the hardware platform to make the experimental verification for the rolling bearing fault. From the experimental data, we get the following results. The accuracy rate of discrimination was 78% and 90% respectively for single acceleration and acoustic emission. But the accuracy rate of discrimination was 94.1% for information fusion, which indicates that the detection method of rolling bearing fault based on information fusion is effective.