Dominant Parameter Identification for Virtual Load Model Based on Incremental Learning

With the boom of machine learning in power system parameter identification, the increment of measured data and the lack of effective processing methods have become the bottleneck for further development. This paper introduces an incremental learning method: combine the new samples with representative samples extracted from old data, then update the parameters of neural network with them all, and generate a new representative sample library for next training batch. We apply incremental learning to the field of virtual load model dominant parameter identification. Examined with 4x500 dataset, this method not only ensures the identification accuracy, but also improves the training efficiency.