The fault diagnosis of gas path system in integral bipropellant propulsion is investigated. Aiming at the problems that the failure rate of gas path system is relatively high and some key information can't be obtained through the sensors, a method of building Modified State Class Graph (MSCG) based on Partially Observed Time Petri Nets (POTPN) and State Class Graph (SCG) is presented. Firstly, the transitions are divided into observable and unobservable transitions according to the observability of the actuators, and the time intervals are modified based on the different relationships of the transitions (synchronous and asynchronous).The MSCG is then built with time Petri nets.Based on the proposed fault diagnosis method, all the effective paths that satisfy the observable transition triggering time and sequence information are found out, and it is determined whether the effective paths have fault transitions.Finally, taking the gas path system in integral bipropellant propulsion as an example, and based on the observability of the actuators, a Petri net is built for making fault diagnosis to the unobservable part of the system, which verifies the effectiveness of the algorithm.