This paper presents a Huber-M estimation of Thevenin equivalent calculation method using variable forgetting factor and projection statistics for the accuracy of Thevenin equivalent parameter identification in time-varying outliers. The method employ the Huber function based on projection statistics to suppress the influence of outliers on parameter identification to improve the robustness of the algorithm by using local measurement data. It also can trace the change of system quickly by using a variable forgetting factor. In order to improve the stability and generalization performance of the model, the paper adopt regularization technique algorithm to solve the problem of ill-conditioned matrix inversion. The simulation results of IEEE 30 node systems verify the effectiveness and accuracy of the proposed method.
Keywords Thevenin equivalent, parameter identification, forgetting factor generalized regularization, bad data