本文提出了一种基于贝叶斯证据框架下加权最小二乘支持向量机(Weighted Least Squares Support Vector Machine,WLS-SVM)的短期负荷预测模型和算法.在对历史负荷数据进行完预处理基础上,分析影响负荷变化的重要因素,然后选择最佳的输入数据作为LS-SVM训练模型的输入向量.通过贝叶斯证据三层推断寻找到模型的最佳参数:第一层推断确定LS-SVM的权向量w和偏置值b,第二层推断确定模型的超参数Y,第三层推断确定核函数的超参数σ.为了提高模型的鲁棒性,赋予了每个样本误差不同的权系数,建立了具有良好泛化性能的WLS-SVM回归模型,从而进一步提高了模型预测的精度.采用上述方法对一固定预测区电网中期负荷进行了预测,结果证明了该方法具有良好的预测效果.%This paper presents a Bayesian evidence framework based on the weighted least squares support vector machine (Weighted Least Squares Support Vector Machine,WLS-SVM) middle-term load forecasting model and algorithms.After the pretreatment of the historical load data,the important factors which affecting the load changes are analyzed,and then the best input data as input vectors of LS-SVM training model are select.Bayesian inference fmds the best evidence of three parameters of the model:first layer is determined to infer LS-SVM weight vector w and the bias value b,the second layer is determined to infer the parameters γsuper model,the third layer is determined to infer hyperparameter kernel function σ.To improve the robustness of the model,WLS-SVM regression model with good generalization performance is established by giving a different weight coefficient to each sample error,it further improves the prediction accuracy of the model.Using the above method to predict a fixed area of the mid-term load is predicted,the results show that this method has good prediction.
展开▼
