论文部分内容阅读
针对电力负荷建模过程中数据可能存在异常值及异常值对模型性能的影响,提出一种基于自适应加权最小二乘支持向量机(AWLS-SVM)回归方法的短期电力负荷预测模型.利用改进的正态分布加权规则自适应地为每个建模样本分配不同的权值,并结合粒子群遗传算法对模型参数进行优化选择,以进一步提高模型的预测精度和泛化能力.以北方某城市电网季度负荷数据为例,对模型的性能进行检验.计算结果表明,AWLS-SVM模型在预测精度和泛化能力方面均优于最小二乘支持向量机(LS-SVM)模型及加权最小二乘支持向量机(WLS-SVM)模型.
Aiming at the possible influence of outliers and outliers on the performance of the model during the process of power load modeling, a short-term power load forecasting model based on adaptive weighted least squares support vector machine (AWLS-SVM) regression method is proposed. The normal distribution weighting rule adaptively allocates different weights to each model sample and optimizes the model parameters by using particle swarm genetic algorithm in order to further improve the prediction accuracy and generalization ability of the model.Taking a city in the north The results show that the AWLS-SVM model is superior to the LS-SVM model and the weighted least squares method in prediction accuracy and generalization ability, Support Vector Machine (WLS-SVM) model.