Multivariate degradation modelling and its application to reliability testing

Abstract

<P>DEGRADATION TEST, WHICH IS MORE COMPLICATED TO MODEL THAN LIFE TEST, IS AN ALTERNATIVE TO OBTAIN SUFFICIENT DATA FOR RELIABILITY ANALYSIS WHEN LIFE TIME DATA IS INSUFFICIENT. THE INCREASE OF THE INFLUX OF NEW TECHNOLOGY, WHICH MAY MOTIVATE THE NEED OF MODELLING AND ANALYSIS OF BI-/MULTI-VARIATE DEGRADATION DATA, ADDS THE DIFFICULTY OF ESTIMATING SYSTEM RELIABILITY. </P><P>IN THIS THESIS, MODELLING AND ANALYSIS OF BIVARIATE DEGRADATION DATA FROM CONSTANT-STRESS DEGRADATION TEST AND ACCELERATED DEGRADATION TEST WERE DISCUSSED. IN ADDITION, MODELLING AND ANALYSIS OF MULTIVARIATE DEGRADATION DATA FROM CONSTANT-STRESS DEGRADATION TEST WAS ALSO EXPLORED. THE MODELS WERE DEVELOPED USING GENERALIZED LINEAR MODEL AND COPULA FUNCTIONS. WITH THE HELP OF COPULA FUNCTIONS, THE JOINT DISTRIBUTION OF THE BI-/MULTI-VARIATE COULD BE ASSESSED. FAILURE DISTRIBUTION AND RELIABILITY BOUNDS AND LIMITS FOR BIVARIATE DEGRADATION DATA WERE ALSO DISCUSSED. THE MODELS WERE APPLIED TO ANALYZE SYSTEM REL