基于CNN-BiLSTM-ATT混合模型的高校高考录取分数预测研究

打开文本图片集

中图分类号：TP183 文献标识码：A 文章编号：2096-4706（2026）04-0024-08

Abstract：ThepredictioncollegeentranceexaminationadmsionscoresinCollgesand Universities isgreatsignificance tocandidates，parentsndducatioalinstitios.Howeverthispredictionorkishalengingduetoteinfueneultiple factorsuchasaatiodiultyollmntstategsollgesndUiversisndesaleddate.foets paperroposesahybridmodelbasedonCNN-BiLSTM-ATTtopredict theadmisioscoresCollgesandUniversities.Temodel firstusesConvolutioalNeuralNetworktoextracthelocalfeaturesadmissionscoresinollgesndUniversitis，ten leas thelong-tedependencyinteteseriesthoughidirectioalLong-ShortTMmoryBLT）adfinalltoduces theAtentionMechanism（AT）toenhancethefocusonkeyyeardata toimprovepredictionperformance.Theexperimentalresults showthathe CNN-BiLSTM-ATTmodelhashighacuracyand generalizationability inpredicting theadmissonscoresColleges andUniversities.Comparedwithotermodels，itoreefectivelycaptures techangingtrendadmissionsoresandacevesbetter evaluation indices，providing a valuable reference for college entrance examination volunteer application.

Keywords： CNN-BiLSTM-ATT; collge entrance examination admission score; prediction; Neural Network

0 引言

高考作为中国规模最大的统一选拔考试，其结果每年受到广大考生家庭的高度关注。（剩余10981字）