Research Of Fault Pattern Recognition For Rolling Bearing

In recent years, with the development of science technology and national economy, a large number of manufacturing equipments are developed towards the direction of large-scale, precision, complicated and automation. Rolling bearing is one of the most useful mechanical components in manufacturing equipments, whose operating condition relates to the whole equipments' operation and production. On the one hand, technical advancement could raise the production efficiency and bring considerable profits and benefits; On the other hand, the costs of production equipment have been increasing greatly. Once the equipment break down or fails, it will cause huge economic losses and casualties. Therefore, it is very significant for rolling bearing to be studied on operating condition pattern recognition.This paper is supported by national key technological research program of china:2006BAK01B01, and main jobs of the paper are as follows:(1) Introduced the research background and objective of the paper, described the related research and application at home and abroad, and cited the main tasks and innovations of the paper.(2) Introduced and researched some signal processing methods, feature selection and extraction techniques, including the FFT, cyclic statistic theory, empirical mode decomposition and feature extraction methods based on singular value decomposition and principal component analysis.(3) Researched and improved two important models in this paper, including the partial mean model and end effects model, which belonged to the theory of empirical mode decomposition.(4) Introduced the pattern recognition model based on the Second-order cyclic statistics and singular value decomposition, and proved this model with vibration data of rolling bearing from CWRU. The results suggested that this model have high recognition accuracy.(5) Introduced the pattern recognition model based on the empirical mode decomposition and principal component analysis. The same test was done, results suggested that this model could also have high recognition accuracy.