Development of New Algorithm for Power System Unit Commitment Considering Fuel Consumption Constraint

考虑油耗约束的电力系统机组承诺新算法开发

基本信息

批准号：
07455118
负责人：
OYAMA Tsutomu
金额：
$ 3.52万
依托单位：
YOKOHAMA NATIONAL UNIVERSITY
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (B)
财政年份：
1995
资助国家：
日本
起止时间：
1995 至 1997
项目状态：
已结题

来源：
https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-07455118/
关键词：
Operation of Power System Unit Commitment Genetic Algorithm 並列処理

项目摘要

The thermal plant unit commitmetn is a very important problem to improve the economics of power system without degrading the reliability. Since the number of power plant is very large in the real power system, however, it is very difficult to search the optimal solution. It is especially dificult to solve the problem if there are any long term constraint such as fuel consumption constraint.In this study, the algorithm that can solve large size unit commitment problem including fuel consumption constraint is developed. In the developed algorithm, first, candidates of possible solution for each period are obtained. Using the obtained candidates, the size of the problem can be reduced. Therefore, a combinatorial problem involved in the unit commitment probelm can be solved by existing algorithms such as genetic algorithm. The most important point in the algorithm is to obtain a good and effective candidates for each period. In this study, the method based on the priority list method is used.In order to prove the effectiveness of the developed algorithm, 5-machine 10-period model and 10-machine 20-period model, and 20-machine 24-period model is used. It is clarified that the result using the developed algorithm is better than that using the simple priority list method, genetic algorithm, or Lagrangian relaxation method.

火电厂机组的投入是在不降低可靠性的情况下提高电力系统经济性的一个非常重要的问题。然而，由于实际电力系统中发电厂的数量非常多，因此搜索最优解非常困难。如果存在诸如燃料消耗约束等长期约束，则解决该问题尤其困难。在本研究中，开发了可以解决包括燃料消耗约束的大尺寸机组组合问题的算法。在所开发的算法中，首先，获得每个时期的可能解决方案的候选者。使用获得的候选者，可以缩小问题的规模。因此，单元承诺问题中涉及的组合问题可以通过现有算法如遗传算法来解决。算法中最重要的一点是获得每个时期的良好且有效的候选者。本研究采用基于优先级列表法的方法。为了证明所开发算法的有效性，分别采用5机10周期模型、10机20周期模型、20机24周期模型被使用。结果表明，使用所开发的算法的结果优于使用简单优先级列表法、遗传算法或拉格朗日松弛法的结果。