| DC relay is an emerging electronic component,especially in the field of new energy vehicles.The research of DC relay contact bounce can effectively improve the performance of the relay and better play its role in controlling the main circuit;At present,the research on high voltage DC relays mainly focuses on electrical life,electromagnetic optimization and contact welding characteristics.and the research on contact bounce characteristics is limited to technical methods such as high-speed camera and binocular vision,and it is impossible to achieve contact in a closed state.Point bounce detection.This paper is based on high-voltage DC relays,using acoustic emission technology to monitor through sensors,from the detection principle,equipment selection,to feature extraction,and life prediction system to launch theoretical and experimental research.First,from the basic theory to the post-processing system,the framework of the acoustic emission detection,processing,and analysis process and the main signal analysis methods are sorted out;then,the principle of acoustic emission signal generation is introduced.Based on the microscopic dislocation and slip phenomenon of the material,the acoustic emission characteristics of the contact bounce of DC relays are analyzed;according to the principle of acoustic emission detection,a test plan for the dynamic characteristics of DC relay contacts is designed,and using the constant stress acceleration method to complete the full life cycle dynamic characteristics test of the DC relay contact,the meaning and characterization phenomenon of each part of the waveform in the time domain are briefly explained.Second,using wavelet packet signal decomposition,FFT transformation,time-frequency analysis and other methods to extract the time/frequency domain eigenvalues,the main frequency band of the signal is determined by wavelet decomposition,use signal reconstruction to calculate the energy ratio of each frequency band of the signal,the main frequency band of the signal is obtained through FFT transformation and time-frequency analysis,and the result is mutually confirmed with the wavelet analysis result.Compare the signal under different working times,analyze the signal’s main frequency band,the energy ratio of each frequency band,and extract the corresponding feature values for fitting,and determine the degraded characteristics of the contact dynamic characteristics.The above research provides data support for subsequent life prediction.Finally;Build a BP neural network prediction model,according to the extracted data set of the characteristic parameter degradation of the DC relay contact’s full life cycle,the prediction of the residual life of the contact under single factor is completed.To improve the accuracy of life prediction and reduce errors,proposed a multi-factor forecasting model,greatly reduce the forecast error,then it was verified by random data,which proved the effectiveness of the BP neural network prediction model.At the same time,it also proves the feasibility of acoustic emission detection method,which has certain engineering guiding significanceThere are 80 pictures,8tables and 99 references in this article. |
