SDG-based fault diagnosis and application based on reasoning method of granular computing

Signed directed graph (SDG) as a qualitative model is used in fault diagnosis, because it can express causal relationships among variables of large-scale complex industry systems. However, many relevant results are included in diagnostic conclusions leading to low resolution in based-SDG fault diagnosis, so to solve this problem, granule is used to formally express the elements of SDG model in this paper, after that granular base containing knowledge which reflects the causal relation of faults and symptoms is constructed. And a searching and reasoning method based on granule is used in searching of fault source, consequently fault source is obtained by searching granular base and computing the most similarity. So the resolution could be improved. A 65t/h steam boiler system is taken as an example in the paper, and its answer show the method is feasible.