Disturbed trajectories fully reflect the effects on system dynamic behavior of both models and parameters. Based on the stability mechanism， a mechanism difference degree is proposed as an index of power system disturbed trajectories. The index increases as the difference of dynamic behaviors increases between two scenario trajectories. The two trajectories completely overlap if and only if the index is zero. Therefore， the proximity of the different dynamic behaviors can be quantified， which can be used to verify models and identify parameters through sensitivity analysis. The simulation studies show the efficiency of the index. An essential phenomenon of multi-solution in multi-parameter identification is revealed， so that the identification result depends on both the initial value and search strategy. Different excitation disturbances are required for parameter identification. This paper explores the complexity of system identification， which is far beyond existing knowledge.