基于支持向量机回归的电力系统负荷建模

Load Modeling Based on SVM in Power System

支持向量机(SVM)是当前一种比较流行的学习机,具有良好的理论背景,从结构风险最小化原则出发以快速寻找到全局最优的特点。针对当前负荷建模的不足,提出了运用SVM回归来进行电力系统的非机理负荷建模,并给出了负荷建模的具体步骤。与人工神经网络(ANN)对同一个线路负荷进行建模结果比较表明,基于SVM回归的建模效果优于ANN的建模结果,证明了运用SVM的回归进行电力系统负荷建模的可行性,也为电力系统的负荷建模提供了新的思路和方法。

Currently, the support vector machine （SVM） is used widely for its good theoretical background and optimizing characteristic based on structural risk minimization principle. For the deficiency of current load modeling, a non-mechanism load modeling method based on SVM regression is put forward and the specific steps of load modeling are also provided. Compared with the modeling results of the same line load using artificial neural network （ANN）, the load modeling results using SVM regression is better. Thus, the feasibility of the proposed load modeling method based on SVM is proven, which is also the new idea and method for load modeling in power system.