基于數(shù)據(jù)驅(qū)動的WSN故障檢測框架

打開文本圖片集

中圖分類號：TP391 文獻(xiàn)標(biāo)志碼：A 文章編號：1001-3695（2025）06-029-1815-07

doi：10.19734/j.issn.1001-3695.2024.10.0426

Data-driven WSN fault detection based framework

Xu Hana， Tong Yinghuaa，bt （a.DeptofComper，.TeStateKeybooofbenItellgentIfoaionPresing&AcoQingiNlUesit， Xining810008，China）

Abstract：Wirelessensor network nodes typicallydeployincomplex environments，where failuresare inevitable.Toaddress thischallenge，thispaperproposedadata-drivenframeworkforreal-timefaultdetectioninWSN.Thesystemfilteredandcollectedinitialoperationaldatafromwirelesssensornetworks，usingthisdatatoconstructdatasetsfortrainingfaultdetection models，withthegoalof enhancing modelaccuracy.Theapproach segmented thedata intotime windows，alowing the fault diagnosis model to identifyandcorectsensorfaults withinthemostrecent window.The modelthenreusedtherefined dataset toretrainitself，preparingforthedetectionoffaultsintheupcoming window.Throughcontinuouscyclesof raininganddetection，thesystemensuredthereal-timeupdateof thedetectionmodeltoimproveitsperformance.Experimentalresultsdemonstratethat，whencomparedtoclassicalandstate-of-the-artmachineleamingdetectionmodels，thisapproachachievesbetter precision，accuracy，and eficiencyacrossvarious typesoffaultdatasets.Consequently，theframework offerssuperiordetection capabilities and adapts more effectively to the dynamic nature of WSN environments.

Key words：wireless sensor network；fault tolerance；faultdetection；machinelearing；datadriven；data interpolation

0 引言

在無線傳感器網(wǎng)絡(luò)（wirelesssensornetwork，WSN）中，傳感器故障會導(dǎo)致傳感器讀數(shù)不可靠和不準(zhǔn)確[1]。（剩余15464字）