Aiming at the inherent characteristics of complicated structure, multiple fault types and system uncertainty of marine main power plant, a quantitative algorithm for fault isolation and parameter estimation based on bicausal BG (bond graph) was developed. Based on sensor configuration scheme, the system bicausal BG model was built. By analyzing the model causal path, the system ARRs (analytical redundancy relations) were derived and the fault signature matrix was obtained. The LFT (linear fractional transformation) BG model was developed for fault simulation. Through coupling LFT and bicausal BG models, the fault isolation and parameter estimation of marine main power plant were realized and the effectiveness was validated by MATLAB/Simulink software. Results show that the bicausalled BG helps to simplify ARRs deviation, and the fault isolation and parameter estimation can be effectively realized and the fault status can be tracked and predicted by analyzing the element constitutive equation of bicausalled BG.