首页|期刊导航|郑州大学学报(理学版)|基于XGBoost-FOA-LSTM的燃气分类负荷预测

基于XGBoost-FOA-LSTM的燃气分类负荷预测OA

Gas Classification Load Prediction Based on XGBoost-FOA-LSTM

中文摘要英文摘要

准确的燃气预测可以有效调整供应,减少不必要的能源消耗.通过分析发现,城市和居民的天然气负荷受多种因素影响,包括气温、供暖条件、天气状况、风力、空气质量、假期等.针对现有天然气负荷预测方法多局限于单一预测周期和单一用户类型的问题,提出一种基于 XGBoost-FOA-LSTM 的周负荷预测方法.将 XGBoost 的预测结果作为LSTM 滚动预测的新影响因素,用于预测不同用户在不同影响因素下、不同时期的负荷.验证结果表明,所提模型在城市和居民数据滚动日预测中的误差分别为 3.65%和 6.16%,其均方根误差和平均绝对误差均优于单一 LSTM 模型,且进一步证明了城市日负荷数据受供暖条件影响显著.

With accurate gas forecasting can be effectively adjusted supply and can be reduced unneces-sary energy consumption.The natural gas load of cities and residents was influenced by a variety of fac-tors,including temperature,heating conditions,weather conditions,wind,air quality,and holiday.To address the issue that existing natural gas load forecasting methods were mostly limited to a single forecas-ting period and user type,a weekly load forecasting method based on the XGBoost-FOA-LSTM model was proposed.The prediction results of XGBoost served as a new influencing factor for the rolling prediction of LSTM,to forecast the load of different users with different influencing factors and at different times.The validation results showed that the model had prediction errors of 3.65%and 6.16%in rolling daily fore-casting for city and residential data,respectively.Its root mean square error(RMSE)and mean absolute error(MAE)were both superior to those of the single LSTM model,further proving that the daily load data of cities were significantly affected by heating conditions.

宰光军;王浩;武杨磊;李怡彬;刘燕

郑州大学 网络空间安全学院 河南 郑州 450002郑州大学 网络空间安全学院 河南 郑州 450002郑州大学 网络空间安全学院 河南 郑州 450002郑州大学 网络空间安全学院 河南 郑州 450002郑州大学 数学与统计学院 河南 郑州 450001

能源科技

燃气预测XGBoostFOALSTM

gas predictionXGBoostFOALSTM

《郑州大学学报(理学版)》 2026 (5)

30-36,7

郑州大学产学研用科技创新团队培育项目(32214110-23)

10.13705/j.issn.1671-6841.2024194

评论