Research on fault diagnosis test sequence algorithm based on multi-signal flow graph model

Abstract

Fault diagnosis is a binary recognition problem that requires a minimum average cost testing process to distinguish the fault cause. To reduce the computational complexity in fault diagnosis strategy, the combination between Rollout algorithm and information gain under different search width and depth is researched in this paper. The basic analysis and modeling method of multi-signal flow graph model are systematically described. Illustrated by the example of active filter amplifier circuit, modeling the multi-signal flow graph model and establishing the correlation matrix. On the basis of that, we put forward to apply Rollout algorithm, and combine it with information gain heuristic algorithms, to carry out iterative updating to construct the near-optimal diagnosis strategy. This paper takes binary test as an example, the relationship between the diagnosis strategy and the average test cost under different search width and depth combinations is analyzed on the basis of Rollout algorithm.