Study On Electrical Lifetime Prediction Of Contactor’s Contacts Based On Neural Network

Contactor is an important electrical component of electrical industry. Its lifetime andreliability are directly related to the reliability of the entire control system. Contact is one of thekey parts of contactors, and its lifetime is the key to the performance and longevity of contactors.So study on electrical lifetime of existing contactor’s contacts has an important function andsignificance.Firstly, the experimental design method is determined. AgSnO2contacts are selected asexperimental materials. The electrical contact material tester is used for electrical properties testof four groups of contacts. Experimental data of arcing energy, contact resistance are obtainedwhen the contacts are working. Secondly, based on the test data, single parameter andtwo-parameter neural network prediction model is established using one group of data. Theinfluence of arcing energy and contact resistance on contact lifetime is the main research. Rightalgorithm and transfer function are found, and the parameters of input layer and intermediatelayer are determined to establish the neural network model of predicting contact life.Furthermore, best weights and threshold of the network are found using genetic algorithm.Finally, the predicted value and the actual value of the life are compared. Other groups of testresults are verified using the established life prediction model.