Study On Technology Of Induction Motor Vibration Fault Diagnosis

induction motor has been widely used as the main driving force equipment in several of production industries and power system because of its high price and good environmental adaptability. The operational status of induction motor has impact on normal industrial production directly. It is not only damage the induction motor itself, but will affect the whole system to work, and even endanger the personal safety and cause huge economic losses when induction motor failures and stop the run. Thus, the demand for early fault diagnosis and early urgent warning of induction motor has increased, we can detection and prevention of further deterioration of failure to reduce accidents caused by sudden loss of shutdown early through analysis and diagnosis of the common induction motor fault. Therefore, study fault diagnosis technology of induction motor in the early detection of motor fault and maintenance timely is of great theoretical significance and socio-economic benefits, it has become a scholar's research hotspot at home and abroad topic. This paper is based on fault characteristic frequency of induction motor vibration signal energy value, proposed the method of least squares support vector machines with dynamic neighborhood particle swarm(NPSO-LSSVM) to achieve common induction motor fault diagnosis.This paper introduces the faults and its principle character of induction motor. For complexity and diversity of induction motor fault, analyzing the features and deficiencies of induction motors of various fault monitoring and diagnostics method. Identified vibration fault diagnosis of large induction motor is based on vibration signal energy spectrum analysis method in this paper. Described faults signal processing of induction motors is based on wavelet technology. Using best wavelet packet select db3 wavelet packet algorithm decompensate the vibration signal measured by vibration sensors into high-frequency and low frequency parts, and then achieve the denoising of signal through quantifying each node of the threshold decomposition coefficients using wavelet packet. Decomposed the signal after denoising, extraction frequency bands related to the faults and reconstruction, excluding the main vibration component and interference items, extracting the signal of characteristic frequency range, calculates the energy of various frequency bands to form feature vectors. Finally, find the feature vectors that are suitable for fault diagnosis of induction motor through normalized.The paper introduced least squares support vector machines based on dynamic neighborhood particle swarm optimization algorithm for characteristics of vibration fault diagnosis of induction motor, introduced the process of dynamic neighborhood particle swarm optimization, and details on least squares support vector machine classification method, the established the fault diagnosis model eventually, and optimized parameters in the diagnostic model. Discussing the vibration fault diagnosis process of induction motor based on NPSO-LSSVM. According to the characteristics of induction motor vibration fault, we collected the vibration fault signal of large induction motor, selected the vibration sensor, and in accordance with the method described in Chapter 3 to achieve the vibration signal de-noising of induction motor, extracted the feature vector of induction motor vibration fault, then established vibration fault diagnosis model of induction motor, using NPSO-LSSVM method for vibration fault diagnosis of induction motor. Comparing the accuracy rates of this method with the other diagnostic methods proved that the validity.This paper applied wavelet technology, least squares support vector machine classification algorithm and particle swarm optimization and other methods to vibration fault diagnosis of induction motor, which improved accuracy and effectiveness of fault diagnosis of induction motors effectively.