Electromagnet fault prediction method based on SA-PSO optimized BP neural network

Abstract

The invention discloses an electromagnet fault prediction method based on an SA-PSO optimized BP neural network, and relates to the field of valve body assembly quality. The method comprises the steps: A, analyzing fault types and fault reasons generated by electromagnets, and classifying and setting fault labels according to the fault types; B, determining the structure and parameter setting of the BP neural network according to the fault reason and the fault type, and establishing a three-layer neural network, namely a first model; C, determining the dimension of a particle swarm; D, initializing the speed and parameters of the N-dimensional particles; E, mapping the particles randomly generated in the step D into a weight and a threshold of the BP neural network; F, combining a simulated annealing algorithm and a particle swarm algorithm and alternately operating; G, when the number of iterations and the temperature in the step F meet the requirements, stopping internal cooling; H,combining the first model and the second model to obtain a reliable fault diagnosis result. The method has the advantages of sufficient search and rapid fault diagnosis.

Description

technical field [0001] The invention relates to the field of electromagnet quality, in particular to an electromagnet fault prediction method based on SA-PSO optimized BP neural network. Background technique [0002] An electromagnet is a component that uses the magnetism generated by the iron core to attract the armature to realize the opening and closing function or fix a mechanical part in one position to convert electrical energy into mechanical energy. At present, electromagnets are widely used. For example, the door opening and closing devices of buses or subways that we often see are gradually converted from hydraulic or pneumatic transmission to electromagnets to control the opening and closing of doors. Electromagnets are not only simple in structure The most important thing about easy installation is responsiveness. However, as the degree of automation becomes higher and higher, the reliability of the electromagnet plays a vital role in the operation of these devi...

AI Technical Summary

Problems solved by technology

However, using the traditional BP neural network for fault prediction without parameter optimization may lead to slow learning speed during error backpropagation, and it is easy to fall into local minimum, and the possibility of network training failure is high

Images