Font Size: a A A

Research On Adaptive Lifetime Prediction Method For A New Type Aerospace Relays

Posted on:2018-08-31Degree:MasterType:Thesis
Country:ChinaCandidate:J H GuoFull Text:PDF
GTID:2392330620953546Subject:Applied statistics
Abstract/Summary:PDF Full Text Request
MOSFET is a new type of aerospace relays.It has gradually replaced the traditional electromagnetic relays and widely used in satellite solid state power control module for its excellent performance.Since MOSFET is the key device of solid state power control module of satellite power supply and distribution system,its working lifetime directly affects the reliability of the satellite power supply and distribution system.In traditional prediction method,engineers often use the accelerated life testing data on the ground to establish a reliability model and predict the reliability of MOSFET for future.But the validness of predictions over years based on testing data on the ground can't be guaranteed.This paper aims to find an adaptive method that predicts accurately when real-time observations can be obtained,then compare the prediction accuracy with traditional methods.First,we fit a degenerative model for on-resistance based on testing data on the ground,estimate the relevant parameters with maximum likelihood method.Second,choose the effective observing times to add data on sky,update the likelihood equation and re-estimate the parameters.Then we give the lifetime prediction method based on the new degradation model.In this paper,we choose the effective observing times by maximizing the information of the estimation.Based on the sequence of the observing data,we propose a robust criterion to determine the effective range of the data for reliability prediction.This is an adaptive method to predict the lifetime of the MOSFET.Finally,we do the simulation study and compare predictions with the traditional method.
Keywords/Search Tags:MOSFET, Life Prediction, Maximum Likelihood Method, Monte Carlo, Self-Adaptive Sampling Method
PDF Full Text Request
Related items