A novel approach for improving quality of health state with difference degree in circuit diagnosis

Abstract

Model-based diagnosis (MBD) has been widely acknowledged to be an effective fault diagnosis paradigm for combinational circuits. Most diagnosis algorithms (DAs) return a single diagnosis, a list of diagnoses or a score for every returned diagnosis, which estimates the likelihood of each diagnosis to be correct. Roni Stern et al. recently proposed the heath state as an output of DA to provide a manageable view of which components are likely to be faulty for human operators in the circuit diagnosis area. The health state can be used widely in areas such as troubleshooting problems, among others. This paper proposes a novel approach that improves the quality of health state with difference degree (IHSD) utilizing the MBD approach with multiple observations in the circuit diagnosis area, which returns a list of possible diagnoses that explain these observations simultaneously. In addition, we also propose a multi-observation “scoring” method based on the difference degree for every returned diagnosis, where the difference degree denotes the “distance” from the multi-observations diagnosis to the relevant diagnosis that satisfies only one observation. We present evidence that shows that, compared with the state-of-the-art algorithm on health state (Stern et al., Artif Intell 248:26–45 2018), our approach improves the quality of health state.