嵌入轻量梯度提升机评估模型的暂态稳定预防-紧急协调控制

Transient stability prevention-emergency coordinated control embedded with assessment model of light gradient boosting machine

为了综合发挥运行方式控制和稳控切机控制对电网暂态稳定性的提升作用,提出一种嵌入轻量梯度提升机(light gradient boosting machine,LightGBM)评估模型的电力系统暂态稳定预防-紧急协调控制决策方法。为快速评估控制措施对稳定裕度的影响,首先利用混合控制样本生成方法和LightGBM算法构建预防-紧急控制对稳定裕度的评估模型,考虑到不合理的切机、切负荷控制可能破坏系统稳定性,利用LightGBM评估模型的数值灵敏度来辨识有效的控制地点、缩减决策空间。进一步将LightGBM模型嵌入暂态稳定双层优化控制模型、替代暂态稳定时域仿真校核,结合改进的非支配排序遗传算法(non-dominated sorting genetic algorithm-Ⅱ,NSGA-Ⅱ)实现协调控制策略的快速求解。通过IEEE 39节点测试算例,验证了所提方法能够实现不同严重程度的故障在发生前后的预防控制和紧急控制之间的协调配合,既提高了电网安全稳定性,又减小了优化调度成本。

To bring the improvement of the transient stability of the power grid by the operation mode control and generator tripping control into full play,a power system transient stability prevention-emergency coordinated control decision-making method embedded in the light gradient boosting machine(LightGBM)assessment model is proposed.In order to quickly evaluate the degree of improvement of the system stability margin by the control measures,firstly,the hybrid control sample generation method and the LightGBM algorithm is used to construct a assessment model of the prevention-emergency control on the system stability margin.Considering that the unreasonable generator tripping and load shedding control may deteriorate the system stability.The numerical sensitivity of the LightGBM surrogate model is used to identify effective control sites and reduce the decision space.The LightGBM model is further embedded in the transient stability two-layer optimal control model,which replaces the transient stability time-domain simulation,combines the improved non-dominated sorting genetic algorithm-Ⅱ(NSGA-Ⅱ)to realize the rapid solution of the coordinated control strategy.Through the IEEE 39-node test example,it is verified that the proposed method can realize the coordination and cooperation between the preventive control before the occurrence of faults and the emergency control after the occurrence of faults of different severity,improve the security and stability of the power grid,reduce the cost of optimal dispatching.