基于改进SVM与NSGA-Ⅲ的台区相序在线优化方法  

On-line Optimization Method for Phase Sequence in Station Area Based on Improved Support Vector Machine and Non-dominated Sorting Genetic Algorithm-Ⅲ

作　　者：唐捷 杨银 刘斯亮[2] 张勇军[2] 李钦豪[2] 羿应棋 TANG Jie;YANG Yin;LIU Siliang;ZHANG Yongjun;LI Qinhao;YI Yingqi(Guangdong Power Grid Co.,Ltd.,Guangzhou 510699,China;Research Center of Smart Energy Technology(School of Electric Power,South China University of Technology),Guangzhou 510640,China)

机构地区：[1]广东电网有限责任公司,广东省广州市510699 [2]智慧能源工程技术研究中心(华南理工大学电力学院),广东省广州市510640

出　　处：《电力系统自动化》2022年第3期50-58,共9页Automation of Electric Power Systems

基　　金：国家自然科学基金资助项目(52177085);广州市科技计划资助项目(202102021208)。

摘　　要：针对基于历史负荷建模的台区相序调整不足及调相后供电状态时长不能保证的问题,从数据驱动角度提出了基于改进支持向量机(SVM)超短期负荷预测的台区相序在线优化方法。首先,采用变分模态分解将预测对象的历史负荷分解成多个子序列,对每个子序列采用改进的SVM进行预测,预测模型中引入自适应权重机制改进最小二乘SVM的性能,以过滤数据噪声对结果的影响,提高预测准确度;然后,对智能换相开关负荷支路建立最小三相不平衡度、最少换相次数、最长相序维续时间的多目标优化模型,从第3代非支配排序遗传算法(NSGA-Ⅲ)求解的Pareto最优解集中选取一组满意解作为决策者的相序自动调整方案;最后,以中国广东电网某台区为例进行分析,并与其他方法对比验证了所提方法的有效性。In view of the insufficient adjustment of the phase sequence in station area based on historical load modeling and the inability to guarantee the duration of the power supply state after phase sequence adjustment,from the data-driven perspective,an on-line optimization method for the phase sequence in station areas based on improved support vector machine for ultra-short-term load forecasting is proposed.First,the variational mode decomposition is used to decompose the historical load of the forecasting object into multiple sub-sequences,and an improved support vector machine is used to forecast each sub-sequence.An adaptive weight mechanism is introduced in the forecasting model to improve the performance of least squares support vector machines,which can filter the impact of data noise on the results and improve the forecasting accuracy.Then,a multi-objective optimization model for the load branch of the intelligent commutation switch is established with the smallest three-phase unbalance,the smallest number of commutations,and the longest phase sequence duration.A set of satisfactory solutions is selected from the Pareto optimal solution set solved by the non-dominated sorting genetic algorithm-Ⅲas the automatic phase sequence adjustment scheme for decision-makers.Finally,a certain station area in Guangdong power grid of China is used as an example to analyze and compared with other methods to verify the effectiveness of the proposed method.

关 键 词：数据驱动 支持向量机 负荷预测 遗传算法 相序 

分 类 号：TM715[电气工程—电力系统及自动化]