Research and application of Koopman-based local voltage control of DG in distribution network

Due to the inherent volatility and intermittency, the output of the distributed generator (DG) fluctuates dramatically. The high penetration of DGs deteriorates the condition of voltage excursion in distribution networks. The model-based methods rely on accurate network parameters, which may be difficult to be applied in the practical operation. In contrast, data-driven-based methods can extract the operating rules of distribution networks with the operational data. In this paper, a data-driven-based voltage regulation method for DGs is proposed based on the Koopman theory. Training with the historical data, a control strategy generation model is established day-ahead, which reflects the complex relationship between states and strategies. Then, the model is decomposed into several sub-models and released to each local area. In the local stage, the sub-model can determine the control strategies of DGs during real-time operation by collecting local measurements and forecasting data. The proposed Koopman-based method can effectively alleviate the computational burden while achieving near-global optimization. It is beneficial to tackle the uncertainties of voltage fluctuations caused by DGs in distribution networks.