电池储能系统参与用户侧削峰填谷的鲁棒优化调度策略

现有的用户侧调峰领域的储能调度鲁棒优化方法,缺乏考虑非线性多目标优化模型的研究。当储能出力范围直接受负荷不确定性限制时,传统列和约束生成算法无法求解此类模型。针对以上问题,文章考虑负荷、电池储能等约束条件,建立了以净负荷方差、用户侧用电支出等为优化目标的电池储能系统调度鲁棒优化模型,并同时考虑了用户负荷波动、新能源出力波动等不确定因素。由于第二阶段的决策与优化结果会影响第一阶段决策的取值范围,需要改进列和约束生成算法以实现对该类鲁棒优化问题的求解。通过将目标函数、部分约束条件视为多个子问题考虑,对原列和约束生成算法的每次迭代中的子问题求解步骤进行扩充,使每次迭代中求解多个子问题,能够有效拓宽该算法的适用范围。最后,仿真结果验证了该方法的有效性,能够在不改变传统算法性能优势的情况下成功应用于非线性多目标鲁棒优化问题。

Robust Optimal Dispatch Strategy for Battery Energy Storage System Participating in User-Side Peak Load Shifting

The existing robust optimization methods for energy storage dispatch in the field of user-side peak load shifting lack the research on robust optimal scheduling of energy storage considering nonlinear multi-objective optimization model. When the range of energy storage output is directly limited by the load uncertainty, the conventional column and constraint generation algorithm cannot solve this kind of model. Aiming at these problems, considering the user-side load, battery energy storage and other constraints, a robust optimization model of energy storage system dispatch with net load variance and user-side power expenditure as optimization objectives is established, while the uncertainty of user load and new energy output is also taken into consideration. Since the decision-making and optimization results of the second stage will affect the value range of the decision-making in the first stage, it is necessary to improve the generation algorithm of columns and constraints to solve this kind of robust optimization problem. With the sub-problem solving steps in each iteration of the original column and constraint generation algorithm expanded by considering the objective function and some constraints as multiple sub-problems, solving multiple sub-problems in each iteration can effectively broaden the application scope of the algorithm. In the end, the simulation results are presented to verify the effectiveness of the method, which can be successfully applied to nonlinear multi-objective robust optimization problems without losing the performance advantages of the conventional algorithm.