樟子松木材密度的近红外光谱特征提取最优 算法及预测模型研究

打开文本图片集

中图分类号：S781.33 文献标志码：A 文章编号：1673-923X（2025）10-0183-12

Research on the optimal algorithm for near-infrared spectral feature extraction and prediction model of Pinus sylvestris wood density

QINZijian，LIYaoxiang，ZHANGZheyu，CHENChengwu，LIYiwei （SchoolofchanicalndElectricalEngieering，NortheastForestrUersityHarbino，Heilongjangia）

Abstract：【Objective】WooddensityisapivotalindicatorforevaluatingwoodqualityTherefore，throughtheaccuratepredictionof Pinussylvestisoodsitytlicyodnieleangiatlatifo rationalutilizationofwoodresourcesandthecultivationofforest res.【Method】ThisstudyfocusesonPinussylvestriswoodnd utilizesNIRnon-destructivetestingtechnoogy.Itemploysfeatureextractionalgorithmsincludinguinformativevariableeliination （UVE），compeitieadaptiveeweghtdamling（CAR），successeprjectiosalgitasellasgridarch supportvectoracheegesso（Grid-V），geticgorit-uportvectoracheegressn（G-V）dprtilewa optimization-supportvectormachneregresson （PSO-SVM）.Thesealgoritmsaresequentiallyappledforfeaturebandextractionnd SVMmodeling.Consequntlyanidealnear-ifraedpredictiomodelforPinussystisoddensityisevelopd【Result】inety sevensamplesofinussylestisoodweredvidedtotraiingandestigsetsina1：3atio.Grid-G-Vad modelswereestablishedusingthefullspectrumofnear-infrareddata.Tosimplifythenea-ifraredmodels，threefeatureetraction algorithms-uniformativevarableelimination（UVE）ompetitivedptiveeweigtedamplng（CARS），ndsuccesiveptio algorithm（SPA）wresequentiallpledtoelectspectralbands.SubsequentlyGrid-SV，GA-SVM，andPO-SVagorits wereemployedtomodeltheelectedwavelengths.UtimatelyitwasfoundthateGA-SVmodelutilzingfeaturebandstracted byUVE，exhibited the highestaccuracyand bestperformanceforpredicting Pinus sylvestris wood density，withan R2 value of 0.910 8 andRMSEPof0.Oo59forthepredictionset.【Conclusion】OptimizedtheSVMmodelforPinussylvestriswooddensityusing nearinfraredspectroscopyfeatureextractionalgoriths，therebyachievingrapidandprecisepredictionofooddensity.Thisapproachot onlyminimizes wastageof woodresourcesbutalsomaximizestheirutiltandvalue.Mreover，itprovidesatheoreticalfoundation andtechnical guidanceforenhancingandoptimizingthecultivationofsuperiortreespeciesandforestmanagementpractices.These advancements propel the wood industry towards greater efficiency and sustainability.

Keywords：narinfraredspectroscopy;woodensityatureextraction;supportvectormachneregresion;Pinusslvetris

木材是人类生活中最广泛的天然化合物及原材料之一。（剩余17732字）