基于随机森林算法优化神经网络的中期负荷预测  

Medium-Term Load Forecasting Based on Optimised Neural Network of Random Forest Algorithm

作　　者：闫泓全 孙楚词 丛鑫泽 Yan Hongquan;Sun Chuzhi;Cong Xinze(Liaoning University of Engineering and Technology,Huludao Liaoning 125000,China)

机构地区：[1]辽宁工程技术大学,辽宁葫芦岛125000

出　　处：《现代工业经济和信息化》2025年第3期120-122,共3页Modern Industrial Economy and Informationization

摘　　要：中期电力负荷预测在电网规划和运营中发挥着关键作用。为了提升预测模型的性能,提出了一种将随机森林(RF)算法与反向传播(BP)神经网络相结合的预测方法。该方法首先采用随机森林进行特征选择,识别出对负荷预测最有影响力的特征,以减少模型的输入维度和提高运算效率。然后,构建了BP神经网络模型,利用其强大的非线性映射能力来学习复杂的负荷模式。通过融合随机森林的特征选择能力和BP网络的学习能力,该方法能够有效提高中期负荷预测的准确性和鲁棒性。实验结果表明,与传统的BP神经网络相比,本方法在多个性能指标上均表现出显著的改进,证实了其在中期电力负荷预测中的有效性。Medium-term load forecasting plays a key role in power grid planning and operation.In order to improve the performance of the forecasting model,a forecasting method combining the random forest(RF)algorithm and back propagation(BP)neural network is proposed.The method firstly adopts Random Forest for feature selection to identify the most influential features for load forecasting,in order to reduce the input dimension of the model and improve the computational efficiency.Then,a BP neural network model is constructed to learn complex load patterns by using its powerful nonlinear mapping capability.By integrating the feature selection capability of random forest and the learning capability of BP network,this method can effectively improve the accuracy and robustness of medium-term load forecasting.The experimental results show that,compared with the traditional BP neural network,this method exhibits significant improvements in several performance indexes,which confirms its effectiveness in medium-term load forecasting.

关 键 词：中期负荷预测 随机森林 反向传播神经网络 特征选择 预测性能 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TM715[自动化与计算机技术—控制科学与工程]