基于极限学习机的有源配电网多场景静态电压安全分析

Static Voltage Safety Analysis Based on ELM for Active Distribution Power Grid

随着配电网中分布式可再生能源(distributed renewable generation,DRG)单相接入及其出力波动带来的不确定性增加,DRG接入情景下的配电网过电压、三相不平衡等静态电压安全分析面临新的挑战。为此,提出采用极限学习机模型挖掘配电网的三相潮流计算输入输出之间复杂的映射关系,训练后的网络能够大幅提升不同拓扑结构及不同DRG输出场景下三相潮流的计算效率。基于此提出了一种考虑多场景的有源配电网静态电压安全分析方法,该方法能够快速地对配电网中不同接入节点的安全性做出分析判别。最后采用接入DRG的IEEE 13节点、33节点及118节点系统进行仿真计算。仿真结果表明,所提方法较传统的三相潮流计算方法具有更高的计算速度,且不存在收敛性能上的问题,较BP神经网络方法也具有更高的效率与准确性,验证了所提方法的有效性与实用性。

With the increase of uncertainty caused by the single phase access of distributed renewable generation(DRG)in the distribution power grid,the static safety analysis for overvoltage and imbalance of the distribution power grid with DRG faces new challenges.This paper proposes to use the extreme learning machine to analyze the complex relationship between the input and output of the three-phase power flow calculations in distribution network.The trained network can greatly improve the efficiency of power flow calculation in different scenarios caused by DRG fluctuations with this topology.On this basis,a voltage safety analysis method considering multiple scenarios of distribution power grid is proposed,which can quickly analyze and judge the safety of grid nodes.Finally,this paper uses IEEE 13-node,IEEE 33-node and IEEE 118-node systems with DRG for simulation calculation.The experimental results show that the method proposed in this paper is more effective than the traditional three-phase power flow calculation method without any convergence problem,and has higher efficiency and accuracy than BP neural network.The effectiveness of the proposed method is verified.