陳愛(ài)國(guó),王成勇,葉家瑋
1)廣東工業(yè)大學(xué)機(jī)電工程學(xué)院,廣州510090;2)廣州航海學(xué)院船舶工程系,廣州510725;3)華南理工大學(xué)土木與交通學(xué)院,廣州510640
為解決海上艦船或平臺(tái)的波浪運(yùn)動(dòng)檢測(cè)難以找到合適參照系的問(wèn)題,本課題組前期研究基于加速度計(jì)陣列的船舶波浪運(yùn)動(dòng)檢測(cè)原理[1],建立靜坐標(biāo)系OXYZ、平移坐標(biāo)系 QXpYpZp和連體坐標(biāo)系QXsYsZs,通過(guò)理論推導(dǎo)提出一種加速度計(jì)陣列,如圖1.在Q點(diǎn)采用一個(gè)3軸加速度計(jì),在A、B和C點(diǎn)各采用一個(gè)2軸加速度計(jì),通過(guò)加速度計(jì)直接檢測(cè)出aA_YS、aA_ZS、aB_XS、aB_ZS、aC_XS、aC_YS、aQ_XS、aQ_YS和aQ_ZS,已知各測(cè)量點(diǎn)之間的距離rAQ、rBQ和rCQ,由式(1)可計(jì)算出船舶橫搖、縱搖和艏搖的角加速度
圖1 加速度計(jì)配置圖Fig.1 Configuration of acceleration indicators
由式(2)可計(jì)算出船舶橫搖、縱搖和艏搖的角速度
設(shè)船體上任意一點(diǎn)M,相對(duì)基點(diǎn)Q的位移矢量為r,r在QXsYsZs坐標(biāo)系中的3個(gè)分量為rXS、rYS和rZS,根據(jù)式(3)可計(jì)得任意坐標(biāo)點(diǎn)M處在連體坐標(biāo)中的三向加速度,再通過(guò)坐標(biāo)變換可計(jì)得任意點(diǎn)M在靜坐標(biāo)系下的加速度為[1]
由于艦船波浪運(yùn)動(dòng)補(bǔ)償一般采用位移補(bǔ)償或速度補(bǔ)償策略,船舶任意坐標(biāo)點(diǎn)處的速度和位移通過(guò)對(duì)加速度進(jìn)行數(shù)值積分得到,積分方式包括時(shí)域積分和頻域積分.但僅進(jìn)行簡(jiǎn)單的積分累積誤差巨大.
為解決基于加速度計(jì)陣列的船舶波浪運(yùn)動(dòng)檢測(cè)中加速度積分累積誤差過(guò)大這一工程難題,真正建立用于工程實(shí)踐的船舶波浪運(yùn)動(dòng)數(shù)據(jù)采集系統(tǒng),文獻(xiàn)[2-4]設(shè)計(jì)和構(gòu)建液壓驅(qū)動(dòng)的船舶波浪運(yùn)動(dòng)模擬平臺(tái),以采集時(shí)間間隔st=0.01 s,采樣頻率sf=1/st=100 Hz,對(duì)加速度計(jì)B采集到2 744個(gè)數(shù)據(jù),采集的原始實(shí)驗(yàn)數(shù)據(jù)如圖2.通過(guò)對(duì)所檢測(cè)加速度數(shù)據(jù)進(jìn)行頻譜分析和積分處理,分析加速度數(shù)據(jù)積分中直流分量、低頻分量、高頻分量和積分常數(shù)項(xiàng)對(duì)積分誤差的影響,介紹其誤差消除方法;并通過(guò)對(duì)實(shí)測(cè)加速度數(shù)據(jù)的時(shí)域積分、頻域?yàn)V波后時(shí)域積分和全頻域積分處理方法的研究,提出如圖3的有效處理流程.研究表明,采用全頻域積分處理方法是解決其中積分誤差過(guò)大問(wèn)題的最佳選擇[2-4].
圖2 加速度計(jì)B測(cè)量的加速度數(shù)據(jù)Fig.2 Acceleration data measured by the accelerometer B
圖3 實(shí)測(cè)加速度頻域積分處理框圖Fig.3 Flow chart of acceleration data integral processing in frequency domain
但研究也表明,在采樣時(shí)段首尾端,由于數(shù)據(jù)長(zhǎng)度的有限性和譜泄露,仍存在偏大的積分幅值誤差,如圖4和圖5[5].
圖4 實(shí)測(cè)加速度頻域積分速度Fig.4 integral velocity of the measured acceleration in frequency domain
本研究采用艦船波浪運(yùn)動(dòng)預(yù)報(bào)技術(shù)針對(duì)實(shí)測(cè)加速度或積分?jǐn)?shù)據(jù)進(jìn)行波浪運(yùn)動(dòng)預(yù)報(bào),修正積分末端誤差,來(lái)獲得更好的積分效果.應(yīng)用艦船波浪運(yùn)動(dòng)預(yù)報(bào)技術(shù)還能預(yù)測(cè)未來(lái)的波浪運(yùn)動(dòng)情況,對(duì)于解決艦船過(guò)駁波浪補(bǔ)償機(jī)構(gòu)的大慣量延遲問(wèn)題具有重要意義[6-9].
圖5 實(shí)測(cè)加速度頻域積分位移Fig.5 Integral displacement of the measured acceleration in frequency domain
對(duì)于艦船波浪運(yùn)動(dòng)檢測(cè)實(shí)時(shí)系統(tǒng),通常采用頻域積分技術(shù).而針對(duì)頻域特點(diǎn),采用頻譜預(yù)報(bào)技術(shù)便是首要選擇,這樣預(yù)報(bào)與積分技術(shù)可以很好結(jié)合,節(jié)省計(jì)算處理時(shí)間.
設(shè){X*(n),n=1,2,…,N}為觀測(cè)到的艦船波浪運(yùn)動(dòng)積分?jǐn)?shù)據(jù)序列,此數(shù)據(jù)系列必然包含其周期性變化成分和隨機(jī)變化成分,因此,可設(shè)
其中,{P(n),n=1,2,…,N}為周期性變化的序列項(xiàng),它表明艦船波浪運(yùn)動(dòng)按周期變化的趨勢(shì);{X(n),n=1,2,…,N}為隨機(jī)序列項(xiàng)[10-20].
針對(duì)本艦船波浪運(yùn)動(dòng)模擬平臺(tái)檢測(cè)的加速度積分速度和位移,P(n)僅由1個(gè)諧波函數(shù)組成,即
其中,ω1是諧波的角頻率;To為采樣時(shí)間;nTo為采樣時(shí)刻,n=1,2,…,N;α為復(fù)數(shù),其模表示諧波振幅,相角α代表諧波的初始相角.因此,式(4)可改寫(xiě)成式(6)形式
并簡(jiǎn)記為
其中,
鑒于加速度頻域積分速度首尾端數(shù)據(jù)的幅值偏小較多,對(duì)圖3中實(shí)測(cè)加速度頻域的積分速度數(shù)據(jù)首尾各切去1個(gè)周期,把剩下的中間數(shù)據(jù)按式(6)用最小二乘法計(jì)算,使目標(biāo)函數(shù)最小.因設(shè)定的艦船波浪運(yùn)動(dòng)模擬平臺(tái)運(yùn)動(dòng)周期約為3 s,采樣時(shí)間為0.01 s,因此用于計(jì)算目標(biāo)函數(shù)的數(shù)據(jù)點(diǎn)為n=301,302,…,1 748.
在積分過(guò)程中,對(duì)實(shí)測(cè)加速度按FFT變換和濾波,并已按式(10)處理過(guò)的數(shù)據(jù)(即傅立葉逆變換之前的速度在頻域內(nèi)的頻譜序列,
其中,A為加速度數(shù)據(jù);V為速度數(shù)據(jù).計(jì)算最大功率處的頻率和相位,計(jì)得最大功率為5.664 7×103cm2/s3,為第8個(gè)數(shù)據(jù)點(diǎn),此時(shí)的頻率為(8-1)×df=7×100/2 048=7×0.048 828=0.341 769 Hz,相位為0.789 3 rad.在第301~1 748個(gè)速度數(shù)據(jù)點(diǎn)范圍內(nèi)速度幅值的平均值為3.634 cm/s.
由此,可設(shè)計(jì)算采用的初始預(yù)報(bào)速度為
其中,n=1,2,…,2 048,T0=0.01 s.結(jié)果如圖6.
圖6 艦船波浪運(yùn)動(dòng)預(yù)報(bào)積分初速度Fig.6 Original integral velocity of ship wave movement forecasting
基于式(11),在以速度幅值3.634 cm/s、頻率0.341 769 Hz、相位 0.789 3 rad 為中心的小區(qū)域內(nèi),按式(9)在301~1 748區(qū)間以最小二乘法求取最佳積分速度的波浪預(yù)報(bào),如式(12),所得曲線如圖7.此時(shí)目標(biāo)函數(shù)式(9)計(jì)算的誤差平方和僅為271.439 cm2·s-2.
圖7 艦船波浪運(yùn)動(dòng)預(yù)報(bào)積分速度Fig.7 Integral velocity of ship wave movement forecasting
由圖7可見(jiàn),波浪預(yù)報(bào)取得較好效果,但在末端有較大的相位誤差,這主要是由于計(jì)算目標(biāo)函數(shù)時(shí),舍去末端數(shù)據(jù)造成的.而且由于積分?jǐn)?shù)據(jù)本身的速度幅值較實(shí)測(cè)位移求導(dǎo)速度小,導(dǎo)致預(yù)報(bào)后的速度數(shù)據(jù)幅值也偏小,為此做如下改進(jìn):
改進(jìn)1 為重視末端數(shù)據(jù),計(jì)算區(qū)間由301~1 748擴(kuò)展為301~2 048,而且對(duì)式(9)可改進(jìn)為誤差數(shù)據(jù)的權(quán)重平方和,越靠后的數(shù)據(jù)越有較大的權(quán)重,如式(13).
權(quán)重矩陣H設(shè)置為
經(jīng)編程求解得到此時(shí)的預(yù)報(bào)速度函數(shù)為
曲線放大如圖8,顯然末端相位有所改善.
圖8 艦船波浪運(yùn)動(dòng)改進(jìn)預(yù)報(bào)積分速度Fig.8 Improved integral velocity of ship wave movement forecasting
改進(jìn)2 在改進(jìn)1的基礎(chǔ)上,把預(yù)報(bào)速度的幅值用相鄰積分速度中波高減去波谷的最大值的一半代替.經(jīng)編程求得峰谷的最大值為4.027 18.
由此得到改進(jìn)后的曲線如圖9.可見(jiàn),通過(guò)頻譜波浪預(yù)報(bào)技術(shù)極大修正積分速度的誤差,達(dá)到較好的積分效果.
圖9 艦船波浪運(yùn)動(dòng)再次改進(jìn)的預(yù)報(bào)積分速度Fig.9 Further improved integral velocity of ship wave movement forecasting
以上進(jìn)行加速度積分為速度和進(jìn)行頻譜預(yù)報(bào)采用了實(shí)測(cè)加速度數(shù)據(jù)中的第1~2 048個(gè)數(shù)據(jù),為檢驗(yàn)加速度積分速度頻譜預(yù)報(bào)在預(yù)報(bào)上的準(zhǔn)確性,對(duì)實(shí)測(cè)加速度數(shù)據(jù)的第697~2 744個(gè)數(shù)據(jù)進(jìn)行頻域積分處理,對(duì)相應(yīng)數(shù)據(jù)點(diǎn)的實(shí)測(cè)位移數(shù)據(jù)進(jìn)行求導(dǎo),并畫(huà)出速度預(yù)報(bào)曲線進(jìn)行比較,如圖10.圖中改進(jìn)預(yù)報(bào)速度2曲線13.51 s之后的數(shù)據(jù)即為預(yù)報(bào)的數(shù)據(jù).可見(jiàn),改進(jìn)后的積分預(yù)報(bào)速度在圖中的7 s內(nèi)很好地預(yù)測(cè)了積分速度,但與積分速度一樣,與實(shí)測(cè)位移的求導(dǎo)速度有很小的相位誤差,這主要是由于數(shù)據(jù)采樣時(shí),有少數(shù)數(shù)據(jù)的采樣時(shí)間多于10 ms,而在頻域積分時(shí)一律按10 ms處理造成的.
圖10 艦船波浪運(yùn)動(dòng)預(yù)報(bào)積分速度的預(yù)報(bào)檢驗(yàn)Fig.10 Test of forecasting integral velocity of ship wave movement
采用同樣的預(yù)報(bào)方法對(duì)頻域積分位移進(jìn)行處理,其功率譜如圖11.在第8個(gè)數(shù)據(jù)點(diǎn)處有最大功率1.228 24×103cm2/s3,此時(shí)頻率為 0.341 769 Hz,相位為-0.781 484 rad.在第301~1 748個(gè)數(shù)據(jù)點(diǎn)范圍內(nèi)的積分位移幅值的均值為1.782 8 cm.
圖11 頻域積分位移的功率譜Fig.11 Power spectrum of integral displacement in frequency domain
因此,可設(shè)計(jì)算采用的初始預(yù)報(bào)位移為
圖12 艦船波浪運(yùn)動(dòng)積分預(yù)報(bào)初始位移Fig.12 Original displacement velocity of ship wave movement forecasting
利用改進(jìn)的方法1,以式(13)為目標(biāo)函數(shù)利用最小二乘法在第301~2 048個(gè)數(shù)據(jù)點(diǎn)范圍內(nèi)求得的預(yù)報(bào)位移函數(shù)為
改進(jìn)后預(yù)報(bào)函數(shù)的曲線末端放大圖如圖13.可見(jiàn),末端相位得到改善,但幅值偏小,為此再應(yīng)用改進(jìn)方法2作出改進(jìn)函數(shù)如式(19),最終預(yù)報(bào)曲線如圖14.可見(jiàn),艦船波浪運(yùn)動(dòng)最終的預(yù)報(bào)積分位移曲線很好地修正了積分位移的誤差,與實(shí)測(cè)位移曲線非常逼近.
圖13 艦船波浪運(yùn)動(dòng)改進(jìn)預(yù)報(bào)積分位移Fig.13 Improved integral displacement of ship wave movement forecasting
圖14 艦船波浪運(yùn)動(dòng)再次改進(jìn)的預(yù)報(bào)積分位移Fig.14 Further improved integral displacement of ship wave movement forecasting
以上進(jìn)行位移積分和頻譜預(yù)報(bào)采用實(shí)測(cè)加速度數(shù)據(jù)中的第1~2 048個(gè)數(shù)據(jù),為檢驗(yàn)加速度積分位移頻譜預(yù)報(bào)在預(yù)報(bào)上的準(zhǔn)確性,對(duì)實(shí)測(cè)加速度數(shù)據(jù)的第697~2 744個(gè)數(shù)據(jù)進(jìn)行頻域積分處理,并畫(huà)出相應(yīng)數(shù)據(jù)點(diǎn)的實(shí)測(cè)位移曲線進(jìn)行比較,如圖15.圖中改進(jìn)預(yù)報(bào)位移2曲線13.51 s之后的數(shù)據(jù)即為預(yù)報(bào)數(shù)據(jù).由圖15可見(jiàn),改進(jìn)后的積分預(yù)報(bào)位移在圖中的7 s內(nèi)很好地預(yù)測(cè)了積分位移.
圖15 艦船波浪運(yùn)動(dòng)預(yù)報(bào)積分位移的預(yù)報(bào)檢驗(yàn)Fig.15 Test of forecasting integral displacement of ship wave movement
通過(guò)改進(jìn)頻譜波浪預(yù)報(bào)技術(shù),使之用于艦船波浪運(yùn)動(dòng)模擬平臺(tái)實(shí)測(cè)加速度頻域積分速度和位移的修正,取得良好效果,極大地糾正了積分速度和位移的誤差,且可對(duì)艦船波浪運(yùn)動(dòng)在較長(zhǎng)時(shí)間內(nèi)進(jìn)行有效預(yù)報(bào).本研究改進(jìn)了基于加速度陣列的艦船過(guò)駁波浪運(yùn)動(dòng)檢測(cè)系統(tǒng)的積分處理,使之達(dá)到更高精度;對(duì)艦船過(guò)駁波浪運(yùn)動(dòng)的成功預(yù)報(bào)有助于控制系統(tǒng)提前動(dòng)作,解決執(zhí)行機(jī)構(gòu)的大慣量滯后問(wèn)題.
/References:
[1]Chen Aiguo,Ye Jiawei,Chen Yuanming.Measurement principle of ship wave movement based on accelerometer array[J].Journal of Shenzhen University Science and Engineering,2010,27(3):374-378.(in Chinese)陳愛(ài)國(guó),葉家瑋,陳遠(yuǎn)明.基于加速度計(jì)陣列的艦船波浪運(yùn)動(dòng)檢測(cè) [J].深圳大學(xué)學(xué)報(bào)理工版,2010,27(3):374-378
[2]Chen Aiguo,Ye Jiawei,Su Shu,et al.Measuring and processing about the acceleration of experimental platform simulating ship wave movement[J].Shipbuilding of China,2011,52(3):83-90.(in Chinese)陳愛(ài)國(guó),葉家瑋,蘇 曙,等.艦船波浪運(yùn)動(dòng)模擬平臺(tái)加速度數(shù)據(jù)檢測(cè)與處理 [J].中國(guó)造船,2011,52(3):83-90.
[3]Chen Aiguo,Ye Jiawei.Experimental study on acceleration measurement and numerical integral of ships wave movement[C]//The 21st International Offshore and Polar Engineering Conference. Hawaii(USA):ISOPE,2011,148-151.
[4]Chen Aiguo,Ye Jiawei,Chen Yuanming.Measurement Technique and Data Processing of Ship Wave Movement[J/OL].[2012-10-17].Science Paper Online,http://www.paper.edu.cn/index.php/default/releasepaper/content/201210-158.(in Chinese)陳愛(ài)國(guó),葉家瑋,陳遠(yuǎn)明.船舶波浪運(yùn)動(dòng)檢測(cè)技術(shù)與數(shù)據(jù)處理 [J/OL].[2012-10-17].中國(guó)科技論文在線,http://www.paper.edu.cn/index.php/default/releasepaper/content/201210-158.
[5]Chen Aiguo.Study on the Key Technique of New Wavemovement Compensative System for Sealift Ships[D].Guangzhou:South China University of Technology,2011.(in Chinese)陳愛(ài)國(guó).新型艦船過(guò)駁波浪補(bǔ)償系統(tǒng)關(guān)健性技術(shù)研究[D].廣州:華南理工大學(xué),2011.
[6]Chen Yuanming,Ye Jiawei,Wei Dong.Experimental study on a stable platform system having the function of wave motion compensation[J].Machine Tool& Hydraulics,2008,36(4):67-71.(in Chinese)陳遠(yuǎn)明,葉家瑋,魏 棟.波浪運(yùn)動(dòng)補(bǔ)償穩(wěn)定平臺(tái)系統(tǒng)的試驗(yàn)研究 [J].機(jī)床與液壓,2008,36(4):67-71.
[7]Ye Jiawei,Chen Yuanming,Wang Dongjiao,et al.Wave motion compensation scheme and its model tests for the salvage of an ancient sunken boat[J].China Ocean Engineering,2006,20(4):635-643.
[8]Chen Aiguo,Ye Jiawe.Research on four layer back propagation neural network for the computation of ship resistance[C]//Proceedings of IEEE International Conference on MechatronicsandAutomation.Changchun(China):IEEE Press,2009:2537-2541.
[9]Chen Aiguo,Ye Jiawei.Research on the genetic neural network for the computation of ship resistance[C]//Proceedings of International Conference on Computational Intelligence and Natural Computing.Wuhan(China):IEEE Press,2009:366-369.
[10]Schellin T E,Beiersdorf C,Chen X B,et al.Comparative frequency domain sea keeping analysis of a fast mono hull in regular head waves[C]//Proceedings of the 21st International Conference on Offshore Mechanics and Arctic Engineering.Oslo(Norway):ASME,2002:753-760.
[11]Storhaug G.Experimental Investigation of Wave Induced Vibrations and Their Effect on the Fatigue Loading of Ships[D].Oslo(Norway):Norwegian University of Science and Technology,2007.
[12]Magnusson A K,Reistad M.Guide to Wave Analysis and Forecasting[M].Geneva(Switzerland):Secretariat of the World Meteorological Organization,1998:34-42.
[13]Reistad M,Magnusson A K.Guide to Wave Analysis and Forecasting[M].Geneva(Switzerland):Secretariat of the World Meteorological Organization,1998:57-66.
[14]Kim K,Jurnalov C D,Ham S.Mechanisms of female urinary continence under stress:frequency spectrum analysis[J].Journal of Biomechanics,2001,34(5):687-691.
[15]Wilson R V,Carrica P M,Stern F.Unsteady RANS method for ship motions with application to roll for a surface combatant[J].Computers and Fluids,2006,35(5):501-524.
[16]Zhang Y,Ong W,Arakelyan I,et al.Polaron-to-polaron transitions in the radio-frequency spectrum of a quasi-twodimensional Fermi gas [J].Physical Review Letters,2012,108(23):235302-235307
[17]Jeong Chayjeon,Jae Hyunkim,Jae Geunyoo,et al.Frequency spectrum analysis of corona discharge source measured by ultrasound detector[C]//International Conferences on Computer Applications for Modeling,Simulation and Automobile.Jeju Island(Korea):Springer-Verlag Press,2012:294-299.
[18]Nakajima T,Shingyoji M,Anayama T,et al.Spectrum analysis of endobronchial ultrasound for prediction of lymph node metastasis in patients with lung cancer[J].Chest Journal,2012,142(4):1251-1258.
[19]Liu Can,Wu Jingquan,Li Guanghui,et al.Frequencyspectrum characteristics of force in end milling with tool wear and eccentricity[J].The International Journal of Advanced Manufacturing Technology,2012,23(3):10-24.
[20]Dai Chunyang,Zhang Xiaoling.A novel two-dimensional frequency spectrum for bistatic SAR processing[C]//IEEE CIE International Conference on Radar.Chengdu(China):IEEE Press,2011:1551-1553.