科技创新与应用2025，Vol.15Issue(20)：1-6,6.DOI:10.19981/j.CN23-1581/G3.2025.20.001
基于贝叶斯优化的BiLSTM-Adaboost热电厂热负荷预测研究
摘要
Abstract
In this paper,a BiLSTM-Adaboost prediction model based on Bayesian optimization is proposed for the heat load prediction problem of thermal power plants.First,the effects of primary network heating parameters and meteorological factors on heat load are considered comprehensively,and the Pearson correlation coefficient method is utilized to screen the model input variables.Secondly,using the feature extraction ability of bidirectional long short-term memory network(BiLSTM)for time series data,Adaboost algorithm is introduced to integrate multiple BiLSTM models to improve the accuracy and robustness of the prediction;finally,Bayesian optimization method is adopted to optimize the hyper-parameters of the model to solve the problem of reduced prediction accuracy due to the perceived improper settings.Simulation experiments are carried out with the actual operation data of a thermal power plant in China,and the results show that the proposed Bayesian optimization BiLSTM-Adaboost model has high prediction accuracy and stability in heat load prediction compared with other network models.关键词
供热负荷预测/BiLSTM-Adaboost神经网络预测/贝叶斯优化算法/超参数寻优/预测精度Key words
heating load prediction/BiLSTM-Adaboost neural network prediction/Bayesian optimization algorithm/hyperparameter optimization/prediction accuracy分类
信息技术与安全科学引用本文复制引用
张语珊..基于贝叶斯优化的BiLSTM-Adaboost热电厂热负荷预测研究[J].科技创新与应用,2025,15(20):1-6,6.基金项目
国家能源集团科技项目(2020-653) （2020-653）
