Feature Parameter Distribution Based Diagnostics and Prognostics of Aviatic Electronics
Hou Man-yi · Shuxue de shijian yu renshi · 2012
Prognostics based on feature parameter trend evolution is an effective method,a new fault judgement integrated with feature parameter distribution is introduced to implement multi-modes diagnostics and prognostics.A more precise forecast model is built based on procedure neural network,and trend forecast is performed based on both the model and usage record of feature parameters.The way researched above is applied in a control-transfer component to set an research example.Researchs show that results aquired by this method are more according with the actual situation in which there are more than one fault modes that are of different severity,and PNN model is an effective trend forecast method as it's of higher accuracy in both approximating and extended forecasting.