Online Prediction Method For Dynamics And Reliability Of Aerospace Relay

Aerospace relay plays an important role in the field of aerospace and defense system,Reliable prediction of aerospace relay is related to aerospace safety. Based on thecharacteristics of aerospace relay itself, a variety of methods are chosen to predict its lifeand reliability, the research work is as follows:(1) Life prediction model of aerospace relay based on degradation model. Based on theselected five parameters of aerospace relay, we establish the prediction algorithm ofaerospace relay life using the degradation regression model. The research finds that, theresult of the model can reflect the trend of performance parameters of the relays accurately,But it can not reflect the feature of non-stationary, that is to say, its fluctuation componentcan not be processed well.(2) Life prediction model of aerospace relay based on wavelet packet and Chaos.Through improving the transform method of wavelet packet and chaos model, on the basisof this, we can first deal with the parameters data of the relay with multi-scaledecomposition to get high and low frequency timing component of data, and then based onchaotic model we predict high and low frequency timing component respectively, as aresult, chaotic prediction results for high and low frequency timing sequence are obtained,after that, based on reconstruction method of wavelet packet we can synthesize predictedresults of various chaotic models to get the final forecasting result. The research finds that:the relays life curve predicted by it is very accurate, and it can fully reflect the fluctuationsin the performance characteristics of the relay, but it also has its limitation, it can only givethe relay life curve, it can not distinguish the failure state of the working relay from figure.(3) Life prediction model of aerospace relay based on Gaussian mixture model.Through improving Gaussian mixture model, the life and reliability numerical predictionmodel for aerospace relay based on this model can be established. The research finds that:the improved Gaussian mixture model can achieve accurate quantification of things, and itcan deal with non-stationary time series very well, therefore, this model can accuratelypredict the life and reliability of the relay numerically, it can fetch up the defect of curvedprediction algorithm such as the prediction model of wavelet packet–chaotic. Summary all, the reliability curve predicted by wavelet packet-chaotic predictionmodel has a good accuracy, it belong to the category of curve type forecasting model. Thereliability prediction based on Gaussian mixture model not only agrees with the accrualsituation well, but also it is based on numerical prediction model, thus the model has astrong practicability. Under the aerospace environment, the accurate numerical predictionmodel of life and reliability for aerospace relay has a stronger universal applicabilitycomparing with curve type forecasting model. The combination of life prediction model ofrelay based on Wavelet Packet-Chaos with predictive model based on Gaussian mixturemodel is a feasible way to achieve a more accurate prediction.