Abstract: The incalculable serious consequences may be caused by nuclear power plant accident. Therefore the parameters evaluation, fault forecast and diagnosis in nuclear power plant are of great significance. DUCG (dynamical uncertainty causality graph) is able to compactly and graphically represent uncertainty causality in any cases, simplify graphics, outspread event based on evidence observed, and calculate the updated probabilities of the queries still in concern. This paper discusses the DUCG algorithm implementation and optimization in nuclear power plant fault diagnosis, especially the software realization of DUCG simplification and expression calculation. The realization of DUCG fault diagnosis reasoning machine can be used for further theoretical studies and algorithm analysis.