摘要
改进的机器学习预测方法以梯度提升、支持向量机和随机森林模型作为底层模型,并通过使用弹性网络线性回归的方式整合3种底层模型来获得组合预测模型,以达到更好的预测效果。模型参数优化过程中采用交叉验证和贝叶斯优化方法,最终的预测结果使用平均百分比误差作为衡量指标。经实验表明,组合预测模型同时具有3种底层模型的优势,通过选择3种模型的最优权重以获得最佳的预测效果,具有更强的鲁棒性以及更加广泛的应用场景。
The improved machine learning prediction method uses gradient enhancement,support vector machine,and random forest model as the underlying models,and integrates the three underlying models through the use of elastic network linear regression to obtain a combined prediction model,in order to achieve better prediction results.During the optimization process of model parameters,cross validation and Bayesian optimization methods are used,and the final prediction results are measured by the average percentage error.Through experiments,it has been shown that the combination prediction model has the advantages of three underlying models at the same time.By selecting the optimal weights of the three models to achieve the best prediction effect,it has stronger robustness and a wider range of application scenarios.
作者
郑斌
王君莹
罗天
于晗
ZHENG Bin;WANG Junying;LUO Tian;YU Han(Zhejiang Zheneng Energy Service Co.,Ltd.,Hangzhou,Zhejiang 310003,China;Zhejiang Zheneng Digital Technology Co.,Ltd.,Hangzhou,Zhejiang 310003,China)
出处
《自动化应用》
2023年第19期9-11,14,共4页
Automation Application
关键词
弹性网络
短期负荷预测
特征选择
机器学习
elastic network
short-term load forecasting
feature selection
machine learning
