大规模风电集群接入电网的多目标双层无功自律优化策略

Multi-Objective Bi-Level Reactive Power Self-Adjusting Optimization Strategy for Large-Scale Wind Power Clusters Connected to the Power Grid

中文关键词:多目标无功优化  自律优化  电压控制  粒子群算法

keywords:multi-objective reactive power optimization  self-adjusting optimization  voltage control  particle swarm optimization （PSO）

中文摘要:针对新能源高渗透率接入背景下，电网运行所面临的严峻挑战，构建一种含大规模风电并网的多目标双层无功自律优化策略。依托自律优化可协同提升系统电压，系统化管控风电集群的无功输出。上层构建以电网的网损和电压偏移最小化为目标的优化模型，并求解模型获得帕累托非支配解集，自主改善电压稳定性并降损，实现电网安全节能运行；下层构建立足风电集群内部有功损耗最小化的自律优化架构，以机组无功功率为决策变量，协同优化风电单元无功输出，提升降损效能。上层为小时级模型，制定风电集群24 h无功计划并传递至下层；下层为15 min级滚动时序模型，实现最低有功损耗的优化结果。基于改进的IEEE 39节点架构进行仿真实验，以验证所建模型及求解算法的准确性。

Abstract:Aiming at the severe challenges faced by power grid operation under the background of high penetration of renewable energy， a multi-objective bi-level reactive power self-adjusting optimization strategy for large-scale wind power clusters connected to the power grid is constructed. Self-adjusting optimization can coordinately improve system voltage and systematically manage the reactive power output of wind power clusters. The upper layer constructs an optimization model aiming at minimizing power grid loss and voltage deviation， and solves the model to obtain the Pareto non-dominated solution set， which autonomously improves voltage stability and reduces loss to achieve safe and energy-saving operation of the power grid. The lower layer constructs a self-adjusting optimization architecture focusing on minimizing active power loss inside wind power clusters， taking unit reactive power as decision variables to coordinately optimize the reactive power output of wind units and enhance loss reduction efficiency. The upper layer is an hourly model that formulates a 24-hour reactive power schedule for wind power clusters and transmits it to the lower layer. The lower layer is a 15-minute rolling time-series model to achieve the optimization result of minimum active power loss. Simulation experiments based on the improved IEEE 39-node system verify the accuracy of the proposed model and solution algorithm.