摘要
采用向量序优化算法求解大规模梯级水火电机组多目标检修优化问题时,在表征集合构建、粗糙评估模型构建、序曲线形状辨识等方面均存在技术难点,为此提出一种改进的向量序优化算法。首先,基于随机森林分类器快速从庞大的解空间中抽取指定数目可行解形成表征集合;其次,为精确贴合多目标检修优化模型的特性,借助反向传播神经网络建立具有更高计算精度的粗糙模型;最后,引入动态时间规整方法评估标准序曲线与排序分层序曲线之间的相似度,精确判断多目标检修优化问题的类型并最终求解得到Pareto足够好解集。基于某省级电网的算例结果表明:相比于常规向量序优化算法、ε-约束法,所提算法计算速度分别提高了83.07%、64.50%。所提出的改进向量序优化算法能保证快速求取大规模多目标检修优化问题的足够好解,具有计算速度高、工程适用性强等优点。
When using the vector ordinal optimization algorithm to solve multi-objective maintenance optimization problems for the large-scale cascaded hydrothermal generator sets,there are technical difficulties in the construction of representation sets,the construction of rough evaluation models,and the identification of ordinal curve shapes.Therefore,this paper proposes an improved vector ordinal optimization algorithm.Firstly,a random forest classifier is used to quickly extract a specified number of feasible solutions from a large solution space to form a representation set.Secondly,in order to accurately fit the characteristics of multi-objective maintenance optimization model,a rough model with higher computational accuracy is established using the back propagation neural network.Finally,a dynamic time warping method is introduced to evaluate the similarity between the standard ordinal curve and the hierarchical ordinal curve,accurately determine the type of multi-objective maintenance optimization problem,and solve to obtain a Pareto sufficient solution set in the end.An example based on a provincial power grid shows that the computational speed of the proposed algorithm is improved by 83.07%and 64.50%respectively,compared to the conventional vector ordinal optimization algorithms and the constraint method.Meanwhile,the proposed improved vector ordinal optimization algorithm can ensure fast enough solution of the large-scale multi-objective maintenance optimization problem,and has the advantages of high computational efficiency and strong engineering applicability.
作者
谢敏
何知纯
刘明波
何润泉
赵翔宇
代江
XIE Min;HE Zhichun;LIU Mingbo;HE Runquan;ZHAO Xiangyu;DAI Jiang(School of Electric Power,South China University of Technology,Guangzhou,Guangdong 510641,China;Guangdong Key Laboratory of Clean Energy Technology,South China University of Technology,Guangzhou,Guangdong 510641,China;Power Dispatching Control Center of Guizhou Power Grid Co.,Ltd.,Guiyang,Guizhou 550002,China)
出处
《广东电力》
2023年第4期82-95,共14页
Guangdong Electric Power
基金
广东省自然科学基金项目(2021A1515012245)
中央高校基本科研业务费重大产学研合作扶持专项(x2dlD2201280)。
关键词
机组检修优化
改进向量序优化法
梯级水电站
随机森林
动态时间规整
unit maintenance optimization
improved vector ordinal optimization method
cascade hydropower station
random forest
dynamic time warping(DTW)
