資源描述:
《基于自適應(yīng)Kalman濾波的SAW測溫數(shù)據(jù)糾錯方法.pdf》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫。
1、第37卷第12期2016年12月儀器儀表學報ChineseJournalofScientificInstrumentV01.37No.12Dec.2016基于自適應(yīng)Kalman濾波的SAW測溫數(shù)據(jù)糾錯方法薛明喜,楊揚,張晨睿,韓韜(上海交通大學電子信息與電氣工程學院上海200240)摘要:在無源無線SAW測溫系統(tǒng)實際應(yīng)用中,閱讀器接收到的信號往往受到其所處環(huán)境電磁波的干擾。這些干擾將會使閱讀器得到錯誤的測量數(shù)據(jù)。溫度變化趨勢和測量噪聲時變的特點也給系統(tǒng)建模以及噪聲估計帶來了困難。針對實際應(yīng)用中存在的問題,在Kalman濾波的基礎(chǔ)之上,提出了一種新的自適應(yīng)算法。
2、該算法采用多項式預測的方法建立溫度測量的時變系統(tǒng)模型,根據(jù)當前及歷史測量值,自行調(diào)整預測模型參數(shù),避免因模型不準確造成Kalman濾波效果嚴重下降的問題;通過對測量數(shù)據(jù)小波變換的方法,實時估計測量數(shù)據(jù)噪聲方差,克服未知觀測噪聲的條件下精度下降的問題;當測量數(shù)據(jù)受到干擾時,測量值與糾錯值之間的差值不滿足高斯分布,通過對差值統(tǒng)計特性的分析,對測量數(shù)據(jù)進行錯誤數(shù)據(jù)判別與剔除,有效地抑制干擾對溫度測量的影響。將這種自適應(yīng)Kalman濾波算法應(yīng)用到無源無線SAW測溫系統(tǒng)中,無源無線SAW溫度傳感器測溫實驗的結(jié)果驗證了該算法能有效地糾正粗大誤差,提高測量系統(tǒng)的精度。關(guān)鍵詞
3、:Kalman濾波;多項式預測;小波變換;粗大誤差;殘差中圖分類號:TN911.7TH811.1文獻標識碼:A國家標準學科分類代碼:460.4020ErrorcorrectionmethodforSAWtemperaturemeasurementdatabasedonadaptiveKalmanfilterXueMingxi,YangYang,ZhangChenmi,HanTao(SchoolofElectronicInformationandElectricalEngineering,ShanghaiJiaoTongUniversity,Shanghai20
4、0240,China)Abstract:InthepracticalapplicationofpassivewirelessSAWtemperaturemeasurementsystem,thesignalreceivedbythereaderisofteninterferedbytheelectromagneticwavesintheenvironmentwherethereaderisin.Theseinterferenceswillmakethereadergetwrongmeasurementdata.Thetemperaturechangingtre
5、ndandthetimevaryingcharacteristicofthemeasurementnoisealsobringdifficultytothesystemmodelingandnoiseestimation.Aimingattheseproblemsexistinginpracticalapplication,onthebasisofKalmanfilter,thispaperproposesanewadaptivealgorithm,wbichisfault—tolerantIooutliers.Thealgorithmadoptspolyno
6、mialpredictionmethodtoestablishthetime—variantsystemmodeloftemperaturemeasurement.Accordingtocurrentandhistoricalmeasurementdata.thenewalgorithmcanself-adjusttheforecastingmodelparametersandavoidtheproblemthattheperformanceofKalmanfilterseverelydegradesbecauseofinaccuratemodel.Thewa
7、velettransformofthemeasurementdataisusedtoestimatethenoisevarianceofthemeasurementdatainrealtimeandovercometheproblemthattheaccuracydegradesduetounknownmeasurementnoise.Whenthemeasurementdataareinterferedtheoutliersoccur,thedifferencebetweenthemeasuredvalueandcorrectedvaluedoesnotob
8、eytheGaussiandistri