基于蚁群支持向量机的发电机进相能力建模评估  

Modeling of Generator Leading Phase Ability Based on Ant Colony Optimization and Support Vector Machine

作　　者：孙萌萌 郭芳芳 吕国栋 李宽 高嵩 SUN Mengmeng;GUO Fangfang;LYU Guodong;LI Kuan;GAO Song(Shandong Zhongshi Yitong Group Co.,Ltd.,Jinan 250003,China;State Grid Shandong Electric Power Research Institute,Jinan 250003,China)

机构地区：[1]山东中实易通集团有限公司,山东济南250003 [2]国网山东省电力公司电力科学研究院,山东济南250003

出　　处：《山东电力技术》2020年第8期18-22,共5页Shandong Electric Power

基　　金：国家电网公司科技项目“电力系统广域协调运行控制技术”(5100-201955448A-0-0-00)。

摘　　要：通过发电机进相运行可有效调节系统运行过电压情况,但发电机进相运行是一个非线性的过程,现有方法难以对发电机进相能力进行精准建模,因此,提出采用蚁群算法优化的支持向量机来对发电机进相能力进行建模评估,可进一步提高支持向量机算法的运行效率和泛化能力。以发电机进相运行时的有功和无功作为支持向量机中数据点集的输入向量,以发电机功角和系统/定子电压作为输出向量进行建模。最后以额定功率为10 MW和300 MW的2台发电机为模型进行试验,通过对比分析试验结果可知,本文提出的蚁群支持向量机具有较强的泛化能力,可较为准确地评估发电机进相能力,有一定的实际工程应用价值。System operating overvoltage can be controlled through the generator leading phase operation effectively.However,generator leading phase operation is a nonlinear process and the current methods are difficult to accurately model the phase advance capability of the generator.Modeling of generator leading phase capacity based on ant colony algorithm optimized support vector machine is carried out.Ant colony optimization can improve the efficiency of the operation and generalization ability of support vector machine effectively.This paper selected the active and reactive power as the input vectors of the SVM data points and set the generators power-angle and system or the stator voltage as the output vectors of the SVM data points.Two generators are selected as instance simulation model whose rated power are 10 MW and 300 MW.The ACO-SVM algorithm has stronger generalization ability on comparative analysis and shows applicable value in engineering.

关 键 词：发电机进相 蚁群算法 支持向量机 泛化能力 

分 类 号：TM732[电气工程—电力系统及自动化]