CAJ | 학술논문

针对目前节能调度中确定性的节能效益评估方法不能满足风电出力和负荷功率随机环境下节能管理的需要，基于Monte Carlo随机模拟的方法提出了日调度计划节能效益概率评估方法。该方法首先采用拉丁超立方采样技术进行风电最大出力和节点负荷功率的随机状态模拟，然后建立了含网络安全约束和合同电量约束的日节能调度优化模型，来模拟火电机组的启停和出力状态以及风电机组出力状态。针对模型中含有大量整数变量和网络安全约束的特点，为提高求解效率，基于Benders原理将其分解成具有迭代机制的主问题和子问题求解。其中，主问题在引入起作用的整数变量辨识法对其降低求解规模后，采用CPLEX中的MILP求解器求解；子问题则解耦成24个小的子问题后，采用线性规划法求解。基于大量样本的反复抽样以及机组随机状态的模拟，最终实现了日调度计划节能效益概率评估。最后，以某省级电网为例验证了所提节能效益概率评估方法的有效性。
침대목전절능조도중학정성적절능효익평고방법불능만족풍전출력화부하공솔수궤배경하절능관리적수요，기우Monte Carlo수궤모의적방법제출료일조도계화절능효익개솔평고방법。해방법수선채용랍정초립방채양기술진행풍전최대출력화절점부하공솔적수궤상태모의，연후건립료함망락안전약속화합동전량약속적일절능조도우화모형，래모의화전궤조적계정화출력상태이급풍전궤조출력상태。침대모형중함유대량정수변량화망락안전약속적특점，위제고구해효솔，기우Benders원리장기분해성구유질대궤제적주문제화자문제구해。기중，주문제재인입기작용적정수변량변식법대기강저구해규모후，채용CPLEX중적MILP구해기구해；자문제칙해우성24개소적자문제후，채용선성규화법구해。기우대량양본적반복추양이급궤조수궤상태적모의，최종실현료일조도계화절능효익개솔평고。최후，이모성급전망위례험증료소제절능효익개솔평고방법적유효성。
ABSTRACT:In allusion to the problem that the deterministic energy-saving benefit assessment method in current energy-saving generation scheduling cannot meet the requirement of energy-saving management of wind power output and load under probabilistic environment, based on Monte Carlo simulation a probabilistic assessment method for energy-saving benefit of daily generation scheduling is proposed. In the proposed method firstly the Latin hypercube sampling technique is utilized to simulate the random state of maximum wind power output and nodal load power; then a daily energy-saving generation scheduling optimization model, which contains the constraint of grid security and the constraint of contract electricity quantity, is established to simulate start-up/shut-down and output state of fossil power generation units as well as the output state of wind power generating units. In view of the feature of the established model that there are grid security constraints and a lot of integer variables, based on Benders principle the established model is decomposed into master problem with iteration mechanism and sub-problem, thus the solution efficiency can be improved. After leading in the identification method for acting integer variables to reduce the scale of the solution, the master problem is solved by mixed-integer linear program (MILP) solver in CPLEX optimizer; and the sub-problem is decoupled into 24 smaller sub-problems and solved by linear programming method. Based on repeated sampling of a great number of samples and the simulation of random state of generating units, the probabilistic assessment of energy-saving benefit of daily generation scheduling is ultimately implemented. Finally, taking a certain provincial power grid for example, the effectiveness of the proposed probabilistic assessment method for energy-saving benefit is validated.