Study On The Electrical Contact Failure Prediction For Contacts

The thesis makes a review of the new development of the theory of the electrical contact in the world with especial attentions to the importance and significance of the electrical contact failure prediction for contacts.The thesis puts forward a method of evaluating the reliability of electromagnetic relays by a fuzzy model of two sates pattern recognition, whose variables are contact resistance, open circuit voltage, contacts making time, and contacts breaking time, on the base of the analysis of failure mechanism and patterns for the electromagnetic relays, and constructs the failure prediction model for electromagnetic relays.Contact resistance is one of the most important variables reflecting the reliability of the relay, the thesis does some further study on the contact resistance. A general law of contact resistance variety is suggested on the base of many literatures about electrical contact after investigating the variety of contact resistance for contacts.First time we study the variety and the prediction model for the contact resistance by the method of dynamic analysis of data, namely, the method of time array analysis. The thesis compares the effects of several prediction models developed by different ways, such as moving average, exponential smooth, BP artificial neural networks. At last a prediction model for the variety of contact resistance is put forward according to the knowledge of the probability and statistics by calculate the standard error and the average of the samples.The thesis introduces a set of instrument used for testing the contact resistance. Using this instrument, I make some experiment on the JQX-13F 5A model mid-power miniature relay HH52P. A great deal of data, about the contact resistance of relay, is measured at the designed intervals of operation time during the experiment.I deal with the results of the experiment by the contact resistance prediction model suggested by this thesis. The feasibility of the contact resistance prediction model is verified further by analyzing the effect of prediction.