Regional Voltage Stability Prediction Based on Decision Tree Algorithm

Because of the imperfection of the theory concerning the regional dynamic and static voltage stability evaluation problem and insufficient reliability of the history data retrieved from the FTU, a novel pattern-recognition based algorithm is proposed. The proposed algorithm utilizes advantage of the decision tree algorithm and a decision tree algorithm model was built for the purpose of evaluation of voltage stability status of the regional power grid. Three feature has been extracted for the recognition of the instability pattern. The training set and the test set data are both obtained from simulation data and outputs. By using the proposed method on IEEE 39-bus system and the predefined critical unstable situations, we obtained the decision tree algorithm model on the IEEE 39-bus system stability prediction model. After the simulation, the output data are compared with the original stability status. The experiment comparison shows that the decision tree algorithm model has achieved the acceptable accuracy for the assessment of voltage stability prediction of the regional power grid. The validity of the proposed assessment scheme as well as the adaptability and the accuracy of the classification model are verified by simulation results of IEEE 39-bus system.