基于深度卷积生成对抗网络场景生成的间歇式分布式电源优化配置

Optimal Allocation of Intermittent Distributed Generation Based on Deep Convolutions Generative Adversarial Network in Scenario Generation

风电和光伏等间歇性分布式电源(distributed generation,DG)在配电网中接入比例不断提高,对配电网规划影响显著,需对其出力的不确定性进行建模,以提升含DG的配电网规划的效益与实用性。建立了考虑出力不确定性的DG双层优化配置模型。通过改进的条件深度卷积生成对抗网络模型对DG出力的不确定性进行建模,并在模型中加入月份标签信息以生成面向规划的风光联合出力场景;基于高斯混合模型确定月份标签对应的风光出力的上下限,从而刻画DG出力的不确定性范围。最后,考虑DG出力的运行边界,建立了社会综合成本最小化的DG双层优化配置模型。IEEE 33节点算例验证表明,提出的DG优化配置方案能够提升DG的接入容量,有效降低社会综合成本,提高配电网运行的经济性。

The access proportion of intermittent distributed generation(DG),such as wind or photovoltaic power generation in the distribution network,is increasing,which exerts a significant impact on distribution network planning.It is urgent to model the uncertainty of the distributed generation to improve the efficiency and practicability of the DG-based distribution network planning.A distributed generation optimal allocation model considering the output uncertainty is established.The improved conditional deep convolutions generative adversarial network model is used to model the uncertainty of the output of the distributed power generation,and the monthly label information is added to the model to generate the planning oriented scenario of wind and solar joint output.Based on the Gaussian mixture model,the upper and lower limits of the wind and solar power output corresponding to the monthly labels are determined.Considering the operation boundary of the distributed generation,a bi-level optimal allocation model is established to minimize the total social cost.The IEEE 33-node case shows that the optimal allocation scheme in this paper can improve the access capacity of the distributed power,effectively reduce the total social cost,and improve the economy of the distribution network operation.