电力系统机组组合问题的进制组合算法

摘要

Unit commitment is a typical non-linear hybrid integer programming problem. In this article, existing unit commitment optimization methods are analyzed. To solve current problems such as over use of penalty functions, poor global searching ability and slow convergence speed, a scale combination algorithm applied to unit commitment is put forward. Based on the scale concept, the algorithm first transforms the discrete start-stop matrix into a continuous one to solve the discrete variable serialization problem. Then the use of penalty function is reduced by limiting value range to meet the generating unit output size constraint and unit climbing speed constraint, and the algorithm convergence speed is improved. Finally, through the delay preferred and weakening global learning, the algorithm achieves better global searching ability and improved accuracy. Results of calculation examples show that the solution of the presented algorithm presents higher convergence speed and better global optimizing ability, draws better unit combination plan, and can solve the unit commitment problem better.%电力系统机组组合问题是典型的非线性混合整数规划问题。本文对现有机组组合优化方法进行了分析，并针对现有方法存在的过多使用惩罚函数、全局搜素能力差和收敛速度慢等问题提出了一种适用于机组组合问题的进制组合算法。首先，算法通过进制转换的思想将离散的启停矩阵转化为连续的启停矩阵，解决了离散变量的连续化问题；其次，通过限定取值范围的方式来满足机组出力大小约束和机组爬坡速度约束，减少了惩罚函数的使用，提高了算法的收敛速度；最后，通过延后择优，弱化全局学习等方式使算法具有更好的全局搜索能力，提高了算法寻优精度。经算例验证，该方法具有较好的收敛速度和更好的全局寻优能力，能够得出更优的机组组合方案，较好地解决了机组组合问题。