TVFEMD寻优分解与智能算法优化的FLN土壤 含水量预测

打开文本图片集

关键词：时变滤波经验模态分解（TVFEMD）；算法优化;快速学习网（FLN）；土壤含水量；预测中图分类号：S271；TV93 文献标识码：A文章编号：0439-8114（2025）05-0147-08DOI：10.14088/j.cnki.issn0439-8114.2025.05.023

TVFEMD optimization decomposition and FLN-based soil moisture content prediction using intelligent algorithm optimizations

TIANYu1，CUIDong-wen²

1.YunnanInstituteofWater&HydropowerEngineeingInvestigation，DesignandResearch，Kunming 65O1，China; 2.Wenshan Zhuang and Miao Autonomous Prefecture Water Bureau，Wenshan 663Ooo，Yunnan，China）

Abstract：BasedotheobservedsoilmosturecontentdatafromlO，2，and40cmsoillyersatTanxingadPojiaostationsinYunnanProvince，a prediction model（TVFEMD-BSLO/AO/IVYA/EGO/PSO-FLN）wasconstructedbyimprovingthetime-varying filter empirical modedecomposition（TVFEMD）andfastlearning network（FLN）methods toenhancethetime-series predictionaccuracy of soil moisturecontent.Bycomparingtheperformanceofdiferentoptimzatioalgoritms，asuperiormodelingapproachasprovidedforsoilmoistureprediction.Theresultsshowed thattheTVFEMDdecompositionperformancewasprimarilyinfluencedbytwo key parameters：Bandwidth thresholdand B-splineorder.Optimizing these twparametersusing theIVYAalgorithmimproved the timeseriesdecompositionqualityand further enhanced themodel’sprediction performance.The TVFEMD-BLSO/AO/IVYA/EGO-FLN modeldemonstratedoutstandingpredictionperformanceonthetrainingset，withameanabsolutepercentageerror（MAPE）of 0.002 % ＼～0.077% and a coefficient of determination （ R2 ）of 0.999 7＼～1.000 0. The MAPE in the prediction set was 0.006%＼～0.459 % ， and R2 was0.996 6＼～1.000 0.Compared with the TVFEMD-PSO-FLN model，the TVFEMD-BLSO/AO/IVYA/EGO-FLN model showedsignificantimprovements inbothfitingperformanceandpredictionaccuracyOptimizingFLNhyperparametersusingBLSO， AO，IVYA，ndEGOagorithmseffectivelyimprovedmodelpeformance，withtheIVYAalgorithmexhibitingthemostnotableoptimization effect.

Key Words：time-varyingfilter empirical modedecomposition（TVFEMD）；algorithmoptimization；fastlearning network（FLN）; soil moisture content；prediction

土壤含水量是描述土壤干湿程度，反映旱情最直接、最重要的指标之一，提高土壤含水量时间序列预测精度对于旱情预警、农业生产、生态系统保护和水资源管理具有重要意义。（剩余10412字）