Decision Tree-based Real-time Emergency Control Strategy for Power System

摘要

In order to realize the online evaluation and control of transient stability after power system fault, a decision tree method and emergency control scheme based on Fisher linear discriminant were proposed. This method firstly processed the offline simulation data based on Fisher liner discrimination (FLD), reduced the data dimension and obtained the distance from the data point to the classification hyperplane, and obtained a more adaptive classification decision tree by training the distance data. Fit the distance to obtain a stability index that can guide emergency control. On this basis, the optimal load shedding and generation shedding scheme were calculated according to the 0-1 linear programming, and the loss was reduced to the minimum under the premise of ensuring the stability of the power system. Finally, the simulation analysis was carried out in the WECC system and New England system. The results showed that the decision tree based on FLD is classified accurately and interpretatively. The online emergency control strategy based on stability margin can effectively deal with the transient instability after fault, and restore the system to stable operation.