Decision tree based validation of load model parameters

Load modeling is an important but complicated task for power system analysis and simulation. In previous research, much work has been done on load model parameter identification. However, the effectiveness of the identified load model parameters still need further validation. In practical situation, validation of load model parameters is commonly based on Post-Disturbance Simulation method, which has some inadequacies actually. Learning from the idea of load model parameter identification, a decision tree based method is proposed to validate the load model parameters in this paper. The structure and parameters of load model are selected first. Then, the parameters are discretized and represented as different classes so that the iteration of shapelet searching is carried on until all the samples are classified correctly. Finally, the decision trees are built using C4.5. Extensive simulation results in the CEPRI 36 bus system have demonstrated its feasibility and validity.