Reliability prognostics for electronics via built-in diagnostic tools

This paper proposes a practical model to monitor the degradation of electronic equipment and further to predict the remaining useful life based on the self-diagnostic data. The de gradation precursor, characterized by voltage or current signals, is modeled as a Non-stationary Gaussian process with tim e-varying mean and variance. Statistical testing is then used to characterize the trend patterns for the mean and the variance, from which different types of degradation paths will be extrapolated. Regression tools and time series models can be adopted to forecast the system remaining useful life. A case study drawn from the semiconductor testing equipment is used to demonstrate the applicability and the performance of the proposed method.