一種基于圖學(xué)習(xí)的試飛試驗點關(guān)聯(lián)性挖掘算法

打開文本圖片集

中圖分類號：TN99-34 文獻標識碼：A 文章編號：1004-373X（2025）17-0160-07

引用格式：，.一種基于圖學(xué)習(xí)的試飛試驗點關(guān)聯(lián)性挖掘算法[J].現(xiàn)代電子技術(shù)，2025，48（17）：160-166.

Agraph-basedflighttestpointsrelationshippredictionalgorithm

LIU Peng，DENG Xiaozheng （ChineseFlightTestEstablishment，Xi'an71oO89，China）

Abstract：Thearrangementofflighttestpointsisfundamentalinflighttestig.Howtosientificallandefectivelyarange thesetestpointsandformarationalflighttestplanplaysacrucialroleinensuringsafetyficiencyandcost-ectivenes throughouttheentireflighttestlifecycle.Amongthem，theanalysisofthecorrelationbetweentestpoints，especiallthe determinationoftheprecedencerelationship，representstheexecutionorderofthetestpointsandisakeyfactorinthe arrangement oftheflighttest plan.Therefore，aknowledgemining algorithmbasedongraph convolutional neural networksis proposed tomeetthedemandforpredictingtheprecedencerelationshipof testpoints.Theentirealgorithmmodelisdevelopedin a knowledge graphbasedonthestructuralrepresentatioof test points.Subsequently，modulessuchasgraph knowledgeelement extraction，testpoints‘deepfeatureminingbasedongraphconvolution，andtestpointpairs'logicalrelationshipegreionare designedto explorethecorelationrelationshipbetweentestpointpairsandachieveaccurate precedencerelationshipprediction. Intheflighttest，thetestpointdatasetwastested，andmultipleclassicalmodelswerecompared.Theproposedalgorithm demonstratessignificantadvantagesintermsofaccracyandstability.Theeffectivenessof theproposedalgorithmisverified.

Keywords：flight test;test pointexecutionorder;graph learning;graphconvolutional network；autoencoder;knowledge graph

0 引言

近年來，隨著我國航空裝備的飛速發(fā)展，對飛行試驗的安全和效率提出了更高的要求。（剩余10086字）