A novel RNN based load modelling method with measurement data in active distribution system

摘要

Due to the high penetration of renewable energy and the wide application of electronic devices in the distribution systems, it becomes more complicated to formulate an accurate load model than ever before. For some cases, the formulation cannot be successfully established via the traditional modelling approaches. Therefore, how to establish an accurate load model under the new challenges is drawing a great deal of attention. In this paper, measurement based method using recurrent neural network (RNN) is proposed for constructing an accurate equivalent of active distribution systems, and an application scheme of the proposed model is presented. With the purpose of thoroughly investigating the performance of the proposed RNN based load model in an electric power system, it is applied to reproduce the dynamic behaviors of active distribution systems. In the testing scenario, the performance of the RNN model is evaluated by two types (i.e. using the same disturbance cases and different new disturbance cases).