Anomaly detection of notebook computer based on Weibull decision metrics

This paper presents a approach for anomaly detection of electronic products using the Mahalanobis Distance (MD) and Weibull distribution. The MD value is used as a health index, which has the advantage of both summarizing the multivariate operating parameters and reducing the data set into a univariate distance index. The Weibull distribution is used to determine health decision metrics, which are useful in characterizing distributions of MD values. Furthermore, a case study of the proposed notebook computer anomaly detection method is carried out. The experimental results show that the proposed method is valuable.