Abstract: In order to eliminate the inherent drawbacks of qualitative fault diagnosis method,it is necessary to use quantitative information.The signed direct graph (SDG) of liquid rocket engine system lacks quantitative information.Therefore,it is improved by replacing the sign on the branches by the ratio between deviations of the effect and cause variables.The membership function of fuzzy set theory is used to integrate quantitative process knowledge into the SDG model,thus a qualitative/quantitative model is formed.After the membership function has been defined,it is easy to get the membership between the engine system and the patterns whose fault origins are known by this model,the fault of the engine system can be found according to the principle of maximum membership.This method utilizes enough available numerical resource of variables monitored in the system,it is shown by the diagnosis results that the method has a better identification and resolution capability for the fault patterns with identical qualitative features and can also identify the fault quickly.