This study is committed to fault simulation, prediction and tolerance control in mechatronic systems under digital twins. The goal is to achieve reliable control when the system behavior speeds up and faults are difficult to observe directly. To solve this problem, we propose a unified method that integrates state tables, fault-related event extraction and uncertainty control adjustment in the same framework. Thus fault analysis and control optimization form a more direct connection, and no separate diagnosis stage is carried out. Four experimental results in terms of data show that the proposed method is always better than the baseline model.There is a maximum relative gain on the electromechanical fault simulation dataset, the F1 score rises from 86.11% to 87.89%, which is 1.78 percentage points higher than the strongest baseline. There is the best absolute result on the electromechanical system digital twin dataset, and the model reaches 91.The accuracy rate is 34%. The results show that the proposed framework has a role in handling the joint transactions of fault evolution, state estimation and control.
