电力市场输电阻塞管理模型

A Management Model for Transmission Congestion in Power Market

本文对公平开放市场条件下,独立电网的输电阻塞管理问题做了模型研究。首先,在局部线性化假设下,利用多元线性回归求取线路潮流分布与机组出力分配之间的近似公式。本文对带有常数项和没有常数项的两种线性回归模型分别做了回归分析和细致的假设检验,并由电力系统分析的背景知识,阐明了电网潮流分布与机组出力只有统计规律性,带有常数项的回归模型更合理。根据阻塞调整产生的影响,本文设计了"按损失成比例补偿"和"按市场规则确定费用"两种阻塞费用计算规则,并做了详细地比较讨论。根据电力市场交易规则,兼顾计算的时间效率,本文利用递归策略给出了简单易行的出力分配预案计算方法及其流程图,在机组数不多时简单的手工计算很容易求得分配预案。对阻塞调整问题,本文按电网"安全第一,兼顾经济"的原则,提出分阶段(共分四个阶段:阻塞检查、调整预案、裕度输电、拉闸限电)按步骤规划的计算流程,并对各个规划阶段分别建立了数学模型:阻塞检查为判断一组不等式;调整预案是求解以阻塞费用最小为目标的规划问题:裕度输电规划先以裕度占用率最小为目标,再在裕度占用率不增的条件下以阻塞费用最小为目标做规划:拉闸限电规划则是在保证电网最低安全水平的条件下,以总出力最大为目标做规划。化简后,各阶段的规划模型,除调整预案模型是线性约束条件下阶梯函数族的最大最小规划外,其余阶段规划模型均为线性规划。出于计算效率的考虑,结合题目特点,本文发现以Huffman树作为决策树时,阻塞管理问题的规划流程具有最高计算效率,此时通过对几条简单的规则的判定即可确定应该进行哪一个阶段的规划,从而不必一步步按部就班地进行。本文还对Huffman决策树流程的一些技术细节及改进节点定义的最优流程做了讨论。另外,本文从广义函数角度对阶梯函数的数学分析性能及优化解法做了讨论,并给出了求解以阶梯函数为目标的优化问题的求解建议及两种简单易行的启发式算法,并在附录中,给出了一些典型算法的流程图。本文方法简单有效,思路清晰。主要缺点表现在:因专业知识匮乏,没有结合现行的几种典型的电力市场运营模式的特点给出更合理的阻塞管理办法。

In this paper, on the basis of an open and competitive market, model is developed to study the management of transmission congestion in an independent power system. Firstly, under the assumption of local linearization, the approximate formula for the relation of the power flow distribution to the machine output is derived by multivariate linear regression. Both the models with and without constant-term are considered and tested. Furthermore, according to the knowledge from power system analysis, we demonstrate that the model with the constant-term is more reasonable. Considering the effect of the congestion adjustment, we design two rules for calculating congestion cost and make detailed comparison between them. According to the transaction rule of the power market and accounting for the computation efficiency as well, we use the recursion strategy to figure out a simple and feasible method for the preliminary scheme of output allocation. This article proposes a four-stage computation strategy for the congestion adjustment, which is: congestion check, pre-scheme adjustment, margin transmission and load limitation. In the first stage, a group of inequalities are judged;the second stage is to program with the objective of minimum congestion cost; in the third stage, our first target is the minimum occupancy of the margin, then the objective is the lowest congestion cost while the occupancy remains the same if no less than. In the last stage, assuring the lowest security level, the objective is maximum output. After simplification, all models are linear except the second one, which is a staircase function maxi-min programming subjected to linear constraint condition. Also, we find that the procedure is most efficient if the Huffman tree is used as the decision tree. In this case, only a few simple rules are needed to determine which stage ought to be carried on instead of doing step by step. Last but not the least, we study the programming performance of staircase function from the perspective of generalized function, and we propose two simple heuristic algorithms and other suggestions for the optimization problem whose objective is staircase function. The methods proposed in this paper are clear and efficient. However, for the lack of the professional knowledge, we do not give reasonable congestion management method for some current typical patterns of the power market.