The Research And Implementation Of The On-line Diagnosis System For The Motor Fault

In the motor operation process, fault generate as result of many influential factors, if detected not in time that will lead to serious results, so the paper do a research on fault diagnosis.Essentially, fault diagnosis is a distinguish problem about motor operation stages. The analysis and extracting of fault feature signal is pivotal, for it relates to the accuracy rate and reliability in early prediction of fault diagnosis. The wavelet packet analysis has perfect Time-frequency local characteristics and can divide signal in random frequency channel, it extracts fault features much effectively compared with Fourier Transform, especially when extracting weak fault features. Such method performs better, so this paper adopts wavelet packet analysis to extracting motor's fault features.In the field of fault diagnosis, the training effect of traditional method which based on experience risk minimum shows bad generalization ability under the condition of fault data is hard to obtain and the fault sample is limited, cause such diagnosis method hard to popularize. SVM based on structure risk minimum is adopts in the paper, for it can furthest dig out classification knowledge hidden in data under limited features information, is good for fault classification and category.The mainly studies of this paper are as follows:Firstly, analysis the principle and method of motor fault diagnosis technology, aim at several common motor faults, state the fault mechanism and feature in detail; Utilize the method of the wavelet packet analysis to extract corresponding fault features from the original data, in order to use in the next motor diagnosis; Combining the technology and method of motor fault diagnosis, draw out the application of SVM in the motor fault diagnosis field, introduce the theory about SVM and the parameter selection of the Gauss nuclear; Take Asynchronous motor fault diagnosis for example, the implement process of online fault diagnosis system based on SVM category tree arithmetic is described, and verify such diagnosis system is effective and feasible, indicates the method proposed in the paper is suitable for motor fault diagnosis field; Finally, utilize the Visual C++ and MATLAB by the way of mixing programming the C++ and MATLAB language to develop the simulation software of motor fault online diagnosis system, implement the function of motor fault on-line diagnosis system based on SVM category tree arithmetic.