基于支持向量回归代理模型的STATCOM协同规划方法

Coordinated STATCOM planning method using support vector regression based surrogate model

在新型电力系统中,STATCOM容量的合理配置是电压稳定性提升的关键。本文提出了一种基于支持向量回归代理模型的STATCOM协同规划方法。首先,为了更好地评价短期电压稳定水平,引入了惩罚机制;然后,为了减少计算时间同时保证精度,以时域仿真结果作为训练数据,并据此选择误差最小的核函数构建了基于支持向量回归的代理模型,最后结合代理模型和非支配遗传算法实现了STATCOM的容量优化配置。在PSS/E和Matlab中对IEEE39节点系统的联合仿真结果验证了所提方法的可行性。

In modern power system,the reasonable configuration of STATCOM capacity is the key to improve the voltage stability.This paper has proposed a method based on the support vector regression surrogate model to collaboratively configure the capacity of STATCOM.First,in order to better evaluate the short-term voltage stability level,a penalty mechanism is introduced.Then,in order to reduce the calculation time and ensure the accuracy,the time-domain simulation results are used as training data to construct support vector regression surrogate model with the kernel function which has the smallest error.Finally,combined surrogate model with the non-dominated genetic algorithm can achieve the optimization capacity configuration of STATCOM.The co-simulation results of the IEEE39 node system in PSS/E and Matlab verify the feasibility of the proposed method.