To overcome the obstacles existed in die fault detection of ground tests for liquid-propellant rocket engines such as the lack of real-time, in-time, reliable and robust ability, an adaptive correlation algorithm and an envelope algorithm for the realtime fault detection in the startup and normal process of engines were researched and developed in the paper. Results verified with a large amount of test data and real ground tests show that, the methods proposed can not only detect the faults occurred effectively and in time, and also they are reliable and robust in the real-time and strong interferential situation of ground tests. So they provide important theoretical and practical value for the research of health monitoring systems of large launching vehicles.