摘要
针对支持向量机参数选择难的问题,提出了一种免疫支持向量回归机进行故障测距。该算法运用免疫算法来优化支持向量回归机的参数,减少人为选择参数的盲目性,提高了SVR的推广预测能力。与标准支持向量机算法相比,文章的参数选择具有更明确的理论指导,加速了参数的寻优过程。大量仿真表明:该算法测距精度高,适应性强,训练样本少,测距结果不受故障过渡电阻、对端系统阻抗变化等的影响。
Because it is difficult to select the parameters of support vector regression（SVR）,so an immune support vector regression（ISVR） for fault location is presented in the paper.It uses immune algorithm to optimize the parameters of SVR in order to globally optimize parameters and reduce the blindness of parameters selection by man and improve generation performance of SVR.Compared to the standard SVR,the parameters selection of this algorithm has more specific rule to follow and accelerates the optimization process.A large number of simulations show that the algorithm for fault location is of higher precision,stronger adaptability,fewer samples;and it is free from influence of factors such as fault transition resistance,opposite side impedance changes,and it eliminates false root in the conventional one-terminal algorithm and divergence in iteration.
出处
《华中电力》
2011年第3期94-98,共5页
Central China Electric Power
关键词
免疫算法
支持向量回归机
故障测距
immune algorithm
SVR（support vector regression）
fault location
