A Low-Cost Observer-Based Evaluation of Slow Performance Deterioration Due to Component Ageing in DC—DC Converters

Abstract

This paper presents the design, analysis, and evaluation of a method for detection of the performance degradation due to component ageing, after-shock change in parameters, or excess in environmental conditions. The major novelty of this method consists of using a statistical based decision process rather than one-off threshold-based switch for detection of deterioration in performance. The method builds upon the previous observer-based method for abrupt (catastrophic) fault detection, and uses a Cumulative Sum algorithm for detection of slow trends in performance. Firstly, a review outlines the rates of change in various parameters due to ageing of components. The design and a comprehensive analysis of the proposed method are performed offline in MATLAB® using an average State Space model. Contrary to previously expensive observer-based methods, the implementation is proposed on a low-cost microcontroller, and demonstrated at a 142 kHz sampling and switching rate. Timing diagrams and resources for implementation are also discussed.