基于SARIMA-BP组合模型的福州市气温预测

打开文本图片集

中图分类号：TP39；TP183 文献标识码：A 文章编号：2096-4706（2026）03-0146-05

Abstract：Accurately predicting air temperature changes isof great significance for people's travel and thearangement ofvariousactivities.ThisresearchobainstheairtemperaturedataofFuzhouCityfromJanuary1，0OtoDecember31，2022 via Webrawlers.Given the excessive misingdata insomeyears andthesimilarityofdataacrossdiffrentyears，atotalof 65 dailyair temperaturedataofFuzhouCityin2O areselectedforpredictionandanalysis.FirstlytheSARIMAmodelandheBP neuralnetworkmodelareconstructedseparatelytopredictthedailyaverageairtmperatureofFuzhouCityandtheesultssow thatthe BP neuralnetworkmodelhas higheracuracy incomparison withtheSARIMAmodel.Then，theSARIMA-BPcombined model is constructed to predict the average airtemperature ofFuzhou City for the next14 days，yielding the model values of RMSE=1.34 and MAE=0.86 ，both of which are lower than those of the single models.This demonstrates that the SARIMABPcombined modelcanfullextracttheairtemperatureseriesiformationofFuzhou Cityefectivelyintegratetheadvantages andcharacteristics of the SARIMA modeland theBPneural network model，and thus improve theaccuracyandreliabilityof air temperature prediction.

Keywords：SARIMA model; BP neural network model; SARIMA-BPcombined model; temperature prediction

0 引言

随着全球气候变化的加剧，气温成为人们关注的热点话题。（剩余7812字）