Motor fault mode diagnosis method based on particle swarm optimization support vector machine

Abstract

The invention relates to a motor fault mode diagnosis method based on a particle swarm optimization support vector machine. The method is characterized in that: a wavelet analysis and establishing a particle swarm optimization (PSO) least square support vector machine (LS-SVM) is used to diagnose a motor fault type. The method takes an extraction signal of wavelet analysis as input; an LS-SVM is used for establishing a nonlinear system model and outputting a fault type. Prediction errors are reduced through output feedback and deviation correctio so that the control quantity of a nonlinear system is obtained through PSO rolling optimization, and an effective and accurate fault mode is designed under the condition that a mathematical model of the nonlinear system is unknown. Through testing, the PSO-LS-SVM algorithm and wavelet analysis are used for fault diagnosis of the motor, the result is obviously superior to the application of the SVM algorithm and a previous neural network in fault diagnosis of the motor, convergence to the optimal solution is faster, and the precision of motor fault diagnosis is improved to a great extent.

Description

technical field [0001] The invention relates to the field of state diagnosis of electric equipment, in particular to a motor failure mode diagnosis method based on particle swarm optimization support vector machine. Background technique [0002] As the main energy and power equipment of modern industry, the role of motor is self-evident. If the motor driving the production equipment fails, the production process will be interrupted, resulting in huge economic losses. Rotor misalignment and bearing friction are the most likely failure modes in motor operation, and are also one of the main causes of electrical accidents. Therefore, it is of great significance to optimize and improve the method of electrode fault mode diagnosis to improve the reliability and safety of motor operation. [0003] Electric motors are complex rotating machines with many types of faults and difficult to identify. At present, the input characteristic signal mostly uses Fourier transform to analyze ...

AI Technical Summary

Problems solved by technology

However, for the early weak signals and a large number of non-stationary signals that may be included in the actual motor vibration diagnosis signal, traditional FFT spectrum analysis is difficult to give satisfactory results

Motor fault diagnosis algorithms are mostly based on neural networks. Although neural networks have good nonlinear approximation capabilities, there are still problems such as difficult to determine the network topology, slow convergence speed, and easy to fall into local minima.

Images