• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Control method for multi-input multi-output non-Gaussian random vibration test with cross spectra consideration

    2017-12-22 06:23:25RonghuiZHENGHuaihaiCHENXudongHE
    CHINESE JOURNAL OF AERONAUTICS 2017年6期

    Ronghui ZHENG,Huaihai CHEN,Xudong HE

    State Key Laboratory of Mechanics and Control of Mechanical Structures,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China

    Control method for multi-input multi-output non-Gaussian random vibration test with cross spectra consideration

    Ronghui ZHENG,Huaihai CHEN*,Xudong HE

    State Key Laboratory of Mechanics and Control of Mechanical Structures,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China

    Cross spectra; Kurtosis control; Multi-input multi-output; Non-Gaussian; Random vibration test

    A control method for Multi-Input Multi-Output(MIMO)non-Gaussian random vibration test with cross spectra consideration is proposed in the paper.The aim of the proposed control method is to replicate the speci fied references composed of auto spectral densities,cross spectral densities and kurtoses on the test article in the laboratory.It is found that the cross spectral densities will bring intractable coupling problems and induce dif ficulty for the control of the multioutput kurtoses.Hence,a sequential phase modi fication method is put forward to solve the coupling problems in multi-input multi-output non-Gaussian random vibration test.To achieve the speci fied responses,an improved zero memory nonlinear transformation is utilized first to modify the Fourier phases of the signals with sequential phase modi fication method to obtain one frame reference response signals which satisfy the reference spectra and reference kurtoses.Then,an inverse system method is used in frequency domain to obtain the continuous stationary drive signals.At the same time,the matrix power control algorithm is utilized to control the spectra and kurtoses of the response signals further.At the end of the paper,a simulation example with a cantilever beam and a vibration shaker test are implemented and the results support the proposed method very well.

    1.Introduction

    The traditional Multi-Input Multi-Output(MIMO)random vibration test is mainly to force the multiple outputs to have the speci fied reference power spectra and the test can only be used for the case of stationary Gaussian random vibration.However,non-Gaussian random vibration environments,such as the action of atmospheric turbulence on aircraft,the acoustic excitation by reaction engine and the vibration by combustion instability,are often encountered in aerospace engineering.It is important to monitor the dynamic behavior of the aerospace structures in these non-Gaussian vibration environments.Furthermore,a structure exposure of the same spectra by the Gaussian vibrations or non-Gaussian vibrations will have different damages.1–4Hence,it is necessary to research the method for MIMO non-Gaussian vibration environmental test.

    In recent years,method for the simulation of non-Gaussian random signal has become one of the important topics in many fields,especially in the simulation of wind forces,radar clutters,sea waves and road vehicle vibrations.5–10Some methods are used such as Zero Memory NonLinear(ZMNL)transformation,Auto Regressive Moving Average(ARMA)models,filtered Poisson process,phase modification,alpha stable process,and spherically invariant random vectors process.In the random vibration environmental test,the ZMNL transformation and phase modification are most widely applied.11–13

    ZMNL transformation method is based on ZMNL monotonic functions,among which the most classical one is Hermitian polynomial.Wint′erstein developed a Hermitian moment model to transform a Gaussian process into a non-Gaussian process.14But this method has some inherent shortcomings and some modified forms are suggested by other scholars later.15Smallwood presented three kinds of the ZMNL functions,and each covers similar but slightly different ranges of skewness and kurtosis.11The ZMNL transformation method is simple and computationally efficient,but it may induce harmonic distortion and significant dynamic range loss.11,16

    Phase modification is also a commonly used method to generate non-Gaussian random vibration signals.It is noted that the Auto Spectral Density(ASD)of a random vibration signal is only related to the amplitudes of its Fourier spectrum;hence the kurtosis of the signal can be adjusted by modifying the phase angles without changing its ASD.Steinwolf gave an analytic phase modification formula to generate a non-Gaussian signal with a specified kurtosis from a Gaussian signal.17,18Smallwood utilized a non-uniform phase distribution method to realize the non-Gaussian signal with specified skewness and kurtosis and only kurtosis greater than or equal to 3 can be produced by the method.19Seong and Peterka constructed the Fourier phases by using the four parameterized phase angles.20Hsueh and Hamernik set the Fourier phase to zero within the selected band of frequencies to synthesize the non-Gaussian signal.21Generally,phase modi fication method is a good technique to generate non-Gaussian random signal,but its computational efficiency is not very well if the speci fied kurtosis is large.

    Single Input Single Output(SISO)random vibration test has been widely performed in the laboratory for tens of years.But it is recognized that SISO test is inadequate to simulate the multi-dimensional vibration environments in the real fields.22,23MIMO random vibration test has been emerging and applied along with the advancement of hardware and software.Compared to SISO test,it is much dif ficult to generate the drive signals in MIMO test.Smallwood and Paez contributed to some methods for the generation of stationary Gaussian random drive signals for MIMO test.24But the methodsare difficultto be extended to the MIMO non-Gaussian case.25,26The spectra and kurtoses of the responses should be controlled simultaneously in an MIMO non-Gaussian random vibration test.The kurtoses are used to measure the amplitude distribution characteristics of the responses in time domain and the spectra are used to represent the vibration intensity in frequency domain.Note that the signal in random vibration test is always set to be zero-mean and zero-skewness,so in this paper we only use kurtosis to describe the non-Gaussian characteristic of a random signal.

    In some circumstances,the reference spectra are only de fined as a diagonal matrix of auto spectral densities in an MIMO random vibration test.22In such cases,MIMO random vibration test becomes relatively simple,because only the auto spectral densities need to be controlled and the intractable coupling problems induced by cross spectral densities need not to be considered.26But,the cross spectral densities are very important and they determine the phase and coherence relationships among the outputs.So,the cross spectral densities should also be controlled in order to simulate the vibration environments more realistically.However,in MIMO case,the control to non-Gaussian random vibration test will become very dif ficult if the cross spectral densities are taken into account.Thus,the authors aim to solve this problem in the paper.

    2.Generation of non-Gaussian random signal

    It is known that a Gaussian signal and a non-Gaussian signal can have the same ASD but different kurtoses.As shown in Fig.1,three random signals have the same ASD but different kurtoses.

    Zero-mean stationary Gaussian random signal can be completely determined by its standard deviation.But for zeromean zero-skewness stationary non-Gaussian signal,kurtosis must also be taken into consideration.Normalized kurtosis is de fined as the fourth statistical moment divided by the square of the second statistical moment as

    where x(t)is a random signal.With this definition,the kurtosis of a Gaussian signal is equal to 3 and the kurtosis of a non-Gaussian signal is not equal to 3.Random signal with a kurtosis greater than 3 is said to be leptokurtic or super-Gaussian and random signal with a kurtosis less than 3 is said to be platykurtic or sub-Gaussian.Because moment higher than the fourth is difficult to estimate,kurtosis is always the only parameter used to measure the non-Gaussian characteristic of a random signal in the engineering practice.

    As mentioned above,a non-Gaussian signal can be generated from a Gaussian signal by the ZMNL transformation method.Here,we suggest an improved ZMNL transformation method in order to overcome the defects from the original one.The improved ZMNL transformation method is based on a ZMNL function as

    where g(x)is the resulted non-Gaussian signal and x is the Gaussian signal.The constants a and b are selected to control the skewness and kurtosis of g(x).When a=b,skewness is equal to zero.K represents the kurtosis range of g(x).This improved method should be performed in an iterative process to search a set of proper a and b to obtain a desired kurtosis.At the same time,we back-substitute the Fourier amplitudes of the ZMNL transformed signal with its original ones,which will guarantee the Fourier spectrum of the signal not to be changed.With the improved ZMNL transformation method,the defects of the spectrum distortion and dynamic range loss by the original method can be overcome.

    The schematic of the improved ZMNL transformation method is shown in Fig.2.The Gaussian random signal is transformed to desired non-Gaussian random signal with the reference kurtosis Kfand tolerance kurtosis Ktolin an iterative process,where Eq.(2)is used in the ZMNL transformation.

    3.Description of reference spectra

    In MIMO random vibration test,not only the auto spectral densities need to be controlled,but also the cross spectral densities should be considered.The reference spectra matrix should be positive de finite or positive semi-de finite to be physically realizable.In general,in MIMO random vibration test,the reference spectra matrix is a positive de finite Hermitian matrix whose diagonal elements are real positive numbers and the corresponding off-diagonal elements are complex conjugate pairs.

    For a linear time invariant system with n excitations and n responses,the positive de finite three-dimensional reference spectral density matrix can be expressed as

    where Rjj(j=1,2,...,n)are the auto spectral densities and Rjk(j,k=1,2,...,n,j≠k)are the cross spectral densities.For convenience and brevity,the frequency notation ω will be omitted in the following expressions.The cross spectral densities can be de fined by the auto spectral densities as

    From the fact that the reference matrix should be positive de finite,the mathematical constraint conditions for positive de finite matrix R at each frequency line can be expressed as

    where Dkis the kth order principal minor determinant of R.Considering the general case that n=3,one can have

    Substituting Eqs.(4)and(5)into inequalities(7),one can obtain

    The former two inequalities in Eq.(8)are always true,and only the last inequality is required.Therefore,a reasonable set of coherences and phases is needed to make the reference spectra positive de finite.If the reference spectra are originated from the field measured data,there is no such concern because the reference spectra are always physically realizable.If we arti ficially de fine the reference spectra with coherences and phases as a function of frequency,the mathematical constraint conditions should be complied with.

    4.Control method

    The goal of MIMO non-Gaussian random vibration test is to control the kurtoses and the spectra of the response outputs to meet the references within speci fied tolerances.For a linear time invariant system with n excitations and n responses,we let

    where P is a diagonal matrix named random phase matrix,whose the jth diagonal element is eiθj(j=1,2,...,n)and the phase angle θjis uniformly distributed –π to π.L is the Cholesky decomposition of R as

    where the superscript ‘H’represents the complex conjugate transpose.

    Then one frame Gaussian random signal u can be achieved from U by IFFT,and the jth element of u can be expressed as24

    where F-1denotes IFFT and ljkis the element of L.The next step is to use the improved ZMNL transformation method as described in Section 2 and to modify the phase angles θj(j=1,2,...,n)in sequence to achieve one frame reference response signals unwith desired kurtoses.

    For simplicity and without loss of generality,we can describe the process in detail with n=2.When n=2,from Eqs.(9)and(11),one can have

    We first modify θ1to make the kurtosis of the first output(Line 1 in Eq.(13))meet its reference kurtosis,and then we modify θ2to make the kurtosis of the second output(Line 2 in Eq.(13))meet its reference.Note that the kurtosis of the first output is only affected by θ1and the kurtosis of the second output is affected by θ1and θ2,so θ1and θ2must be modi fied in sequence in order to avoid the cross affection between θ1and θ2.We call this method as Sequential Phase Modi fication(SPM)method.When n>2,we can continue to adjust the kurtoses from line 3 to n in sequence in Eq.(13).It is worthy to note that one cannot use the time domain randomization to u to obtain the continuous stationary non-Gaussian random signals25,27,otherwise the cross spectral structure will be destroyed.

    Up to now,we have obtained the outputs whose spectra and kurtoses are met to the references.In order to achieve the drive inputs,an inverse system method in the frequency domain is used.The Fourier spectra of reference response signals uncan be written as

    Table 1 Parameters of cantilever beam.

    where F denotes FFT.We de fine the frequency response function matrices of the MIMO system as G and its inverse as A.

    If G is ill-conditional at some frequencies,the Moore-Penrose pseudo inverse should be used.28,29Then,by the relationship between the outputs and inputs in the frequency domain,the drive spectra matrix can be obtained as

    Afterwards,one frame drive signals in time domain can be obtained as

    Table 2 Reference spectra for simulation test.

    Table 3 Reference kurtoses for simulation test.

    Repeat the above steps,and the continuously generated dnare windowed and overlapped to compose the continuous stationary drive signals d.It is obvious that the proposed method is different from time domain randomization technique.The time domain randomization technique uses one frame pseudo random signal to generate continuous stationary drive signals while the proposed method uses continuously generated dnto compose the continuous stationary drive signals by the windowing and overlapping.The function of the windowing and overlapping is to remove the discontinuities at the frame boundaries and to generate real drive signals.Here the Half-Sine window or Potter window with overlap factor of 2 is used to make d stationary and they have a good side lobe decaying.24It should be noted that the windowing and overlapping to a signal will decrease its kurtosis,but this in fluence is linear and does not affect the kurtosis control.30

    Because there are many factors which will affect the control process during the test31,one can hardly achieve reasonable responses by one-time inputs.It is essential to correct the drive signals for many times in order to obtain the responses with satisfactory spectra and kurtoses.The control algorithm is used to implement the correction process.Here the matrix power control algorithm is utilized for spectra correction,which has a good stability and does not need scale process.31The two main formulas of matrix power control algorithm are

    where Δlis the spectra error,Lris the Cholesky decomposition of the reference spectral density matrix R by Eq.(10)and Lcis the Cholesky decomposition of the spectral density matrix Scof the present responses.Lnewwill be used to substitute L in Eq.(9)to generate new U and Loldis L used in Eq.(9)last time.ε is the spectrum matrix power which is a constant between(0,1].Similarly,a control algorithm for the kurtosis correction is put forward.The two formulas are

    Table 4 Reference spectra for test.

    Table 5 Reference kurtoses for test.

    where Δkis the kurtoses error,Kris the reference kurtosis matrix which is a diagonal matrix with the reference kurtoses as the diagonal elements,and Kcis the kurtosis matrix of the present responses.Knewwill be used to generate one frame new reference response signals by the improved ZMNL transformation method in Eq.(11)and Koldis used last time.η is the kurtosis matrix power which is a constant between(0,1].

    Finally,the block diagram for the control method of MIMO non-Gaussian random vibration test is shown in Fig.3.

    5.Numerical example

    To verify the effectiveness of the proposed control method given in Fig.3,a simulation test was carried out by an aluminum cantilever beam.The parameters of the beam are listed in Table 1 and the locations of the excitation(input)and control(output)points are shown in Fig.4.Accordingly,the amplitude-frequency diagrams of Frequency Response Functions(FRFs)of the system are exhibited in Fig.5.The control frequency band for simulation was from 20 to 2000 Hz with 400 spectral lines.The reference spectral densities and kurtoses were set as described in Table 2 and Table 3 respectively.Only values at the break points are given and values at other frequency points are calculated by the linear logarithmic interpolation with the given values.

    The uncontrolled response spectra of two control points are shown in Fig.6.From Fig.6,we can see that the uncontrolled response spectra exceed the reference spectra at some frequencies because of the relatively large conditions of the FRF matrices at these frequency points,which make large errors during the inverse calculation of the FRF matrix.Fig.7 shows the controlled response spectra.It can be seen that the response spectra have been controlled within the±3 dB alarm limits after recurrent correction of the drive signals three times.Fig.8 shows the controlled response kurtoses.It is clearly to be seen that the kurtoses of the response signals have been stably controlled toward or around the reference values.The segment time histories of the drive signals and response signals are shown in Figs.9 and 10 respectively.

    6.Test

    6.1.Parameter setting

    To verify the feasibility of the proposed control method for MIMO non-Gaussian random vibration test,a two-input two-output test was carried out.A personal computer with the programmed control software and an Agilent VXI were used to control the x and y directions of a three-axis vibration shaker table.The test system is shown in Fig.11 and the amplitude-frequency diagrams of frequency response functions of the system are exhibited in Fig.12.The control frequency band was from 20 to 2000 Hz with 400 spectral lines.The reference spectral densities were de fined in Table 4 and reference kurtoses for two directions were given in Table 5.For the reference spectra in Table 4,only values at the break points were given and values at other frequency points were calculated by the linear logarithmic interpolation with the given values.The tolerances for the test were set as follows.

    (1)Auto spectral densities:set±3 dB as alarm limits and±6 dB as abort limits in all bandwidths.The relative error of the Root Mean Square(RMS)value of response acceleration at each controlled point should be within±10%according to its reference.

    (2)Cross spectral densities:

    (A)Coherence:set the tolerance to be±0.1 in the rangeand others are not set in this paper.

    (B)Phase:set the tolerance to be±10°in the range 0.5any phase is acceptable.

    (3)Kurtoses:set the tolerance to be±1 in the range K>3 and±0.5 in the range 0

    6.2.Test results

    As shown in Fig.13,the uncontrolled response spectra deviate from the references largely.The uncontrolled auto spectra are not completely within the±3 dB alarm limits and even some spectral lines are beyond the±6 dB abort limits at the frequency points about 1265 and 1675 Hz.It can be seen from Fig.12 that 1265 and 1675 Hz are the resonance peaks of the FRFs,which lead to errors in the inverse calculation of the FRF matrices.From Fig.13,we can also see that the uncontrolled coherence and phase deviate from the references largely.Fig.14 shows the controlled spectra with four spectral corrections and it can be seen that the spectra are all within the tolerance ranges and very close to the references.The spectral RMS errors between the controlled auto spectra and the references are listed in Table 6.Before the kurtoses corrections,the kurtoses of the response signals were all about 3.Fig.15 shows the kurtosis controlling process.Two segments of the two drive signals and controlled responses are shown in Figs.16 and 17 respectively.

    Table 6 Spectral RMS errors.

    7.Conclusions

    In this paper,a control method for MIMO non-Gaussian random vibration test with cross spectra consideration is proposed.The control method is composed of four parts which are the inverse system method,the improved ZMNL method,the sequential phase modi fication method and the matrix power control algorithm.

    Because of the coupling effects of the cross spectra,it is very difficult to generate the drive signals for the MIMO non-Gaussian random vibration test.The key idea of the inverse system method is to generate the reference response signals from the reference spectra and kurtoses in the time domain first and then to obtain the drive signals by inverse system in the frequency domain.The improved ZMNL method is set forth to adjust a random signal to have a desired kurtosis without dynamic range loss.The sequential phase modification method is introduced to eliminate the cross effects among the phase selections.The matrix power control algorithm is applied to the spectra and kurtoses iteration corrections.

    At last,the proposed control method is verified by a simulation example with a cantilever beam and a shaker test and the results are satisfactory.

    Acknowledgements

    This work was supported by the Priority Academic Program Development of Jiangsu Higher Education Institutions and the Postgraduate Research&Practice Innovation Program of Jiangsu Province(No.KYCX17_0234).

    1.Sarkani S,Kihl DP,Beach JE.Fatigue of welded joints under narrowband non-Gaussian loadings.Probabilist Eng Mech 1994;9(3):179–90.

    2.Benasciutti D,Tovo R.Fatigue life assessment in non-Gaussian random loadings.J Fatigue 2006;28(7):733–46.

    3.Kihm F,Rizzi SA,Ferguson NS,Halfpenny A.Understanding how kurtosis is transferred from input acceleration to stress response and its in fluence on fatigue life.11th international conference on recent advances in structural dynamics;2013 July 1–3;Pisa,Italy.Southampton:University of Southampton Press;2013.p.6.

    4.Kihm F,Ferguson NS,Antoni J.Fatigue life from kurtosis controlled excitations.Proc Eng 2015;133:698–713.

    5.Iii WHC.Comments on kurtosis of military vehicle vibration data.J IES 1991;34(6):38–41.

    6.Gioffre`M,Gusella V,Grigoriu M.Simulation of non-Gaussian if eld applied to wind pressure fluctuations.Probabilist Eng Mech 2000;15(4):339–45.

    7.Tatarskii VV,Tatarskii VI.Non-Gaussian statistical model of the ocean surface for wave-scattering theories.Waves Random Media 1996;6(4):419–35.

    8.Rangaswamy M.Spherically invariant random processes for modeling non-Gaussian radar clutter.1993 conference record of the 27th Asilomar conference on signals,systems and computers.Pacific Grove,USA;Piscataway:IEEE Press;1993.p.1106–10.

    9.Rouillard V.On the non-Gaussian nature of random vehicle vibrations.Lecture Notes Eng Comp Sci 2007;2166(1):1219–24.

    10.Grigoriu M.Applied non-Gaussian processes:Examples,theory,simulation,linear random vibration,and MATLAB solutions.Upper Saddle River:Prentice Hall;1995.p.1–232.

    11.Smallwood DO.Generating non-Gaussian vibration for testing purposes.Sound Vib 2005;39(10):18–24.

    12.Steinwolf A.Vibration testing by non-Gaussian random excitations with specified kurtosis.Part I:Discussion and methods.J Test Eval 2014;42(3):659–71.

    13.Steinwolf A.Vibration testing by non-Gaussian random excitations with specified kurtosis.Part II:Numerical and experimental results.J Test Eval 2014;42(3):672–86.

    14.Wint′erstein SR.Nonlinear vibration models for extremes and fatigue.J Eng Mech 1988;114(10):1772–90.

    15.Yang Q,Tian Y.Comparison of non-Gaussian peak factor formulae in wind engineering applications.8th Asia-Pacific conference on wind engineering;2013 Dec 10–14;Chennai,India;Singapore:Research Publishing;2013.p.885–94

    16.Baren PV.The missing knob on your random vibration controller.Sound Vib 2005;39(10):10–6.

    17.Steinwolf A.Approximation and simulation of probability distributions with a variable kurtosis value.Comput Stat Data An 1996;21(2):163–80.

    18.Steinwolf A.Random vibration testing with kurtosis control by IFFT phase manipulation.Mech Syst Signal Pr 2012;28:561–73.

    19.Smallwood D.Vibration with non-Gaussian noise.J IEST 2009;52(2):13–30.

    20.Seong SH,Peterka JA.Experiments on Fourier phases for synthesis of non-Gaussian spikes in turbulence time series.J Wind Eng Indust Aerodyn 2001;89(5):421–43.

    21.Hsueh KD,Hamernik RP.A generalized approach to random noise synthesis:Theory and computer simulation.J Acoust Soc Am 1990;87(3):1207–17.

    22.United States Department of Defense.Test method standard for environmental engineering considerations and laboratory tests.Washington,D.C.:United States Department of Defense;2014.Standard No:MIL-STD-810G_CHG-1.

    23.Underwood MA,Keller T.Recent system developments for multiactuator vibration control.Sound Vib 2001;35(10):16–23.

    24.Smallwood DO,Paez TL.A frequency domain method for the generation of partially coherent normal stationary time domain signals.Shock Vib 1993;1(1):45–53.

    25.Chen HH,Wang PY,Sun JY.Generation of multi-input multioutput non-Gaussian driving signal based on inverse system method.Acta Aeronautica etAstronautica Sinica 2016;37(5):1544–51[Chinese].

    26.Zheng R,Chen H,He X.Control method for multiple-input multiple-output non-Gaussian random vibration test.Packag Technol Sci 2017;30(7):331–45.

    27.Smallwood DO.Multiple shaker random vibration control—An update.Albuquerque:Sandia NationalLaboratories;1999.Report No.:SAND 98–2044C.

    28.Cui S,Chen HH,He XD,Zheng W.Multi-input multi-output random vibration control using Tikhonov filter.Chin J Aeronaut 2016;29(6):1649–63.

    29.Cui S,Chen HH,He XD.Time-domain approach for multi-exciter random environment test.J Sound Vib 2017;398:52–69.

    30.Jiang Y,Chen X,Tao JY.Study on the generation of super-Gaussian and true-random drive signals using time domain randomization.J Vibr Eng 2005;18(4):491–4.

    31.Cui XL,Chen HH,He XD,Jiang SY.Matrix power control algorithm for multi-input multi-output random vibration test.Chin J Aeronaut 2011;24(6):741–8.

    28 October 2016;revised 20 June 2017;accepted 11 August 2017

    Available online 16 October 2017

    ?2017 Chinese Society of Aeronautics and Astronautics.Production and hosting by Elsevier Ltd.This is an open access a rticle under the CCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

    *Corresponding author.

    E-mail addresses:rhzheng@nuaa.edu.cn(R.ZHENG),chhnuaa@nuaa.edu.cn(H.CHEN),hexudong@nuaa.edu.cn(X.HE).

    Peer review under responsibility of Editorial Committee of CJA.

    国产精品 欧美亚洲| 最好的美女福利视频网| 多毛熟女@视频| 19禁男女啪啪无遮挡网站| 久久久国产成人免费| 人妻久久中文字幕网| 亚洲三区欧美一区| 国产精品美女特级片免费视频播放器 | 国产深夜福利视频在线观看| 欧洲精品卡2卡3卡4卡5卡区| 午夜福利影视在线免费观看| 在线观看免费视频网站a站| 丰满饥渴人妻一区二区三| 亚洲国产精品一区二区三区在线| 两个人免费观看高清视频| 国产精品美女特级片免费视频播放器 | 99热只有精品国产| 老司机午夜福利在线观看视频| a在线观看视频网站| 日韩欧美在线二视频| 99国产精品免费福利视频| 国产99久久九九免费精品| 国产激情欧美一区二区| 天堂动漫精品| 久久国产精品影院| 亚洲自拍偷在线| 桃色一区二区三区在线观看| 手机成人av网站| 亚洲 欧美一区二区三区| 欧美中文日本在线观看视频| 12—13女人毛片做爰片一| bbb黄色大片| 午夜免费鲁丝| 757午夜福利合集在线观看| 热re99久久精品国产66热6| 久久性视频一级片| www.999成人在线观看| 91成人精品电影| 90打野战视频偷拍视频| 日韩免费av在线播放| 免费高清视频大片| 一进一出好大好爽视频| 在线观看免费视频日本深夜| 欧美成人午夜精品| av超薄肉色丝袜交足视频| 久久人人精品亚洲av| 黄色毛片三级朝国网站| 一级毛片高清免费大全| 成人永久免费在线观看视频| 国产高清videossex| 亚洲成人免费电影在线观看| 99久久人妻综合| 亚洲美女黄片视频| 久久久久久久午夜电影 | 欧美日本中文国产一区发布| 日本三级黄在线观看| 亚洲第一青青草原| 一二三四在线观看免费中文在| 中文字幕人妻熟女乱码| 一区在线观看完整版| 久久影院123| 亚洲欧洲精品一区二区精品久久久| 国产精品香港三级国产av潘金莲| 久久久久国产一级毛片高清牌| 精品一区二区三区视频在线观看免费 | 妹子高潮喷水视频| www.999成人在线观看| 精品午夜福利视频在线观看一区| 精品国产超薄肉色丝袜足j| 亚洲中文av在线| 亚洲一卡2卡3卡4卡5卡精品中文| 人妻丰满熟妇av一区二区三区| 国产男靠女视频免费网站| 日日摸夜夜添夜夜添小说| 999精品在线视频| 欧美日韩黄片免| 国产一区二区三区视频了| 人人澡人人妻人| 97碰自拍视频| 国产97色在线日韩免费| avwww免费| 成人av一区二区三区在线看| 一级a爱片免费观看的视频| www日本在线高清视频| 日韩国内少妇激情av| 亚洲色图 男人天堂 中文字幕| 丝袜美足系列| 在线av久久热| 亚洲成a人片在线一区二区| 男人舔女人下体高潮全视频| 18美女黄网站色大片免费观看| 国产1区2区3区精品| 亚洲少妇的诱惑av| 琪琪午夜伦伦电影理论片6080| 久久国产精品男人的天堂亚洲| 国产蜜桃级精品一区二区三区| av网站免费在线观看视频| 18禁黄网站禁片午夜丰满| 国产成人av教育| 99精国产麻豆久久婷婷| 免费一级毛片在线播放高清视频 | 精品国产亚洲在线| 丰满迷人的少妇在线观看| 国产又色又爽无遮挡免费看| 97碰自拍视频| 一级作爱视频免费观看| 亚洲精品av麻豆狂野| 视频在线观看一区二区三区| 色综合站精品国产| 欧美日韩亚洲国产一区二区在线观看| 日韩精品青青久久久久久| 午夜免费鲁丝| 黑人猛操日本美女一级片| 国产人伦9x9x在线观看| 91字幕亚洲| 欧美另类亚洲清纯唯美| 在线观看免费高清a一片| 天天影视国产精品| 国产精品久久视频播放| 国产精品野战在线观看 | 一个人观看的视频www高清免费观看 | 欧美中文综合在线视频| netflix在线观看网站| 精品一区二区三区av网在线观看| 多毛熟女@视频| 久久亚洲精品不卡| 精品国产一区二区三区四区第35| 欧美成人性av电影在线观看| 黄网站色视频无遮挡免费观看| 夜夜躁狠狠躁天天躁| 午夜福利欧美成人| 久久人人精品亚洲av| 久久这里只有精品19| 亚洲成国产人片在线观看| 午夜精品久久久久久毛片777| 长腿黑丝高跟| 91国产中文字幕| 成人三级做爰电影| 欧美丝袜亚洲另类 | 啪啪无遮挡十八禁网站| 欧美午夜高清在线| 日韩免费高清中文字幕av| 亚洲精品美女久久av网站| 欧美黄色淫秽网站| 亚洲专区国产一区二区| 一个人免费在线观看的高清视频| 男女床上黄色一级片免费看| 女人精品久久久久毛片| 日日爽夜夜爽网站| 亚洲欧美日韩高清在线视频| 国产亚洲欧美精品永久| 亚洲国产欧美日韩在线播放| 又黄又粗又硬又大视频| 亚洲免费av在线视频| 18禁裸乳无遮挡免费网站照片 | 999久久久国产精品视频| e午夜精品久久久久久久| 男女之事视频高清在线观看| av在线天堂中文字幕 | 多毛熟女@视频| 日韩 欧美 亚洲 中文字幕| 狠狠狠狠99中文字幕| 久久精品国产99精品国产亚洲性色 | 久久久久国产精品人妻aⅴ院| 国产精品爽爽va在线观看网站 | 又黄又粗又硬又大视频| 电影成人av| 亚洲视频免费观看视频| 精品久久久久久,| 成人亚洲精品av一区二区 | 高清av免费在线| 亚洲国产欧美日韩在线播放| 99久久人妻综合| a级毛片黄视频| 久久草成人影院| x7x7x7水蜜桃| 国内毛片毛片毛片毛片毛片| 国产1区2区3区精品| 精品国产乱码久久久久久男人| 久久精品亚洲av国产电影网| 一边摸一边抽搐一进一出视频| 脱女人内裤的视频| 欧美成人午夜精品| av视频免费观看在线观看| 叶爱在线成人免费视频播放| 国产精品秋霞免费鲁丝片| 精品第一国产精品| 久热爱精品视频在线9| 精品一区二区三卡| 十八禁网站免费在线| 亚洲av美国av| 亚洲狠狠婷婷综合久久图片| 国产极品粉嫩免费观看在线| 夜夜躁狠狠躁天天躁| 一边摸一边做爽爽视频免费| av天堂在线播放| 日本精品一区二区三区蜜桃| a级毛片在线看网站| 女人精品久久久久毛片| 免费观看精品视频网站| 丰满饥渴人妻一区二区三| 日韩欧美一区视频在线观看| 色综合站精品国产| 制服诱惑二区| 中文欧美无线码| 极品人妻少妇av视频| 免费在线观看日本一区| 亚洲一区二区三区色噜噜 | 国产精品 国内视频| 首页视频小说图片口味搜索| 亚洲av成人一区二区三| 每晚都被弄得嗷嗷叫到高潮| 精品乱码久久久久久99久播| a级毛片在线看网站| 国产成人欧美在线观看| 久久青草综合色| 一级毛片女人18水好多| 黄色成人免费大全| 日韩 欧美 亚洲 中文字幕| 黄频高清免费视频| 最新在线观看一区二区三区| 女人爽到高潮嗷嗷叫在线视频| 悠悠久久av| 88av欧美| 欧美成狂野欧美在线观看| 中文字幕最新亚洲高清| 久久久久国产精品人妻aⅴ院| 午夜免费成人在线视频| 国产亚洲欧美在线一区二区| 99精品久久久久人妻精品| 亚洲情色 制服丝袜| 午夜激情av网站| 美女大奶头视频| 精品国产乱码久久久久久男人| 国产麻豆69| 久久国产亚洲av麻豆专区| 亚洲一区二区三区色噜噜 | 黄片大片在线免费观看| 欧美老熟妇乱子伦牲交| 亚洲成人国产一区在线观看| 18美女黄网站色大片免费观看| 老司机福利观看| 免费在线观看视频国产中文字幕亚洲| 天天躁夜夜躁狠狠躁躁| av有码第一页| 欧美日韩视频精品一区| 中文字幕最新亚洲高清| 国产精品一区二区精品视频观看| 欧美在线黄色| 交换朋友夫妻互换小说| 精品午夜福利视频在线观看一区| 韩国av一区二区三区四区| av天堂在线播放| 少妇裸体淫交视频免费看高清 | 欧美中文日本在线观看视频| 久久精品亚洲熟妇少妇任你| 欧美在线黄色| 一级片免费观看大全| svipshipincom国产片| 亚洲精品一卡2卡三卡4卡5卡| 男女下面进入的视频免费午夜 | 99久久99久久久精品蜜桃| 久久欧美精品欧美久久欧美| 欧美日韩中文字幕国产精品一区二区三区 | 亚洲成人精品中文字幕电影 | 久久久国产一区二区| 精品福利观看| 午夜免费观看网址| 久久性视频一级片| 亚洲色图综合在线观看| 999久久久精品免费观看国产| 免费在线观看完整版高清| 午夜精品久久久久久毛片777| 一级毛片女人18水好多| 国产深夜福利视频在线观看| 99精国产麻豆久久婷婷| 国产区一区二久久| 不卡av一区二区三区| avwww免费| 男人的好看免费观看在线视频 | 精品国产超薄肉色丝袜足j| 成人免费观看视频高清| 热99re8久久精品国产| 搡老熟女国产l中国老女人| 在线视频色国产色| 久久99一区二区三区| 老司机福利观看| 亚洲欧美日韩另类电影网站| 丰满的人妻完整版| 久久久久九九精品影院| 黄片播放在线免费| 亚洲成人免费电影在线观看| 亚洲成人国产一区在线观看| 欧美日韩福利视频一区二区| 搡老熟女国产l中国老女人| 男人舔女人下体高潮全视频| 在线观看午夜福利视频| 亚洲第一欧美日韩一区二区三区| 欧美人与性动交α欧美软件| 午夜福利一区二区在线看| 真人一进一出gif抽搐免费| 久久精品国产综合久久久| 男女下面进入的视频免费午夜 | 夜夜躁狠狠躁天天躁| 波多野结衣高清无吗| 国产成人精品久久二区二区91| 免费久久久久久久精品成人欧美视频| 激情在线观看视频在线高清| 美女高潮喷水抽搐中文字幕| 免费在线观看黄色视频的| 两性午夜刺激爽爽歪歪视频在线观看 | 日韩 欧美 亚洲 中文字幕| 午夜免费激情av| 亚洲av片天天在线观看| 国产一区在线观看成人免费| 久久精品亚洲av国产电影网| 丝袜美足系列| 亚洲成人精品中文字幕电影 | 中文字幕最新亚洲高清| 一级黄色大片毛片| 91精品三级在线观看| 女警被强在线播放| 高清欧美精品videossex| 一区福利在线观看| 人人妻人人澡人人看| 香蕉久久夜色| 久久精品aⅴ一区二区三区四区| www.www免费av| 丁香六月欧美| 亚洲欧美日韩无卡精品| 久久天躁狠狠躁夜夜2o2o| 国产精品野战在线观看 | 国产成人一区二区三区免费视频网站| 咕卡用的链子| 另类亚洲欧美激情| 如日韩欧美国产精品一区二区三区| 精品国产乱子伦一区二区三区| 韩国精品一区二区三区| 欧美在线一区亚洲| 久久青草综合色| av中文乱码字幕在线| 夜夜爽天天搞| 窝窝影院91人妻| 啦啦啦免费观看视频1| 如日韩欧美国产精品一区二区三区| 在线国产一区二区在线| 亚洲精品久久成人aⅴ小说| 午夜精品久久久久久毛片777| 两性午夜刺激爽爽歪歪视频在线观看 | 久久中文看片网| 亚洲成a人片在线一区二区| 亚洲人成网站在线播放欧美日韩| 亚洲av成人av| 可以免费在线观看a视频的电影网站| 亚洲欧美精品综合一区二区三区| 91精品三级在线观看| av片东京热男人的天堂| 精品少妇一区二区三区视频日本电影| 中文字幕人妻丝袜一区二区| 中文字幕色久视频| 国产精品爽爽va在线观看网站 | 亚洲全国av大片| 电影成人av| 亚洲中文日韩欧美视频| 国产主播在线观看一区二区| 人人妻人人澡人人看| 欧美日韩av久久| 亚洲精品粉嫩美女一区| 欧美另类亚洲清纯唯美| 久久欧美精品欧美久久欧美| 高清毛片免费观看视频网站 | 精品电影一区二区在线| a级毛片在线看网站| 黑人操中国人逼视频| 亚洲熟妇中文字幕五十中出 | 久久亚洲精品不卡| 久久久国产成人精品二区 | 一区在线观看完整版| 精品卡一卡二卡四卡免费| 美女午夜性视频免费| 亚洲精华国产精华精| 91大片在线观看| 黑人巨大精品欧美一区二区蜜桃| 亚洲狠狠婷婷综合久久图片| 国产精品久久电影中文字幕| 国产欧美日韩一区二区三区在线| 交换朋友夫妻互换小说| 国产成人免费无遮挡视频| 午夜福利欧美成人| 美女 人体艺术 gogo| 在线观看www视频免费| 免费观看人在逋| 日日爽夜夜爽网站| 午夜a级毛片| 成人国语在线视频| 精品国产亚洲在线| 日本a在线网址| √禁漫天堂资源中文www| 亚洲午夜理论影院| 国产一区在线观看成人免费| 亚洲人成电影免费在线| 亚洲精品成人av观看孕妇| 国产伦人伦偷精品视频| 日日摸夜夜添夜夜添小说| a级片在线免费高清观看视频| 两人在一起打扑克的视频| 国产成人欧美| 一级片免费观看大全| 91九色精品人成在线观看| 天堂动漫精品| 日韩欧美在线二视频| 亚洲欧美激情在线| 18禁美女被吸乳视频| 黑人操中国人逼视频| 亚洲,欧美精品.| 高清av免费在线| 国产亚洲精品第一综合不卡| 99久久国产精品久久久| 在线观看66精品国产| 亚洲精华国产精华精| 国产乱人伦免费视频| 精品国产乱码久久久久久男人| 免费一级毛片在线播放高清视频 | 国产精品一区二区三区四区久久 | 午夜精品国产一区二区电影| 日韩大码丰满熟妇| 欧美精品啪啪一区二区三区| 国产三级黄色录像| 免费观看精品视频网站| 亚洲成人国产一区在线观看| 欧美中文综合在线视频| 一区福利在线观看| 亚洲男人的天堂狠狠| 女人高潮潮喷娇喘18禁视频| www.www免费av| 亚洲精品一卡2卡三卡4卡5卡| 午夜亚洲福利在线播放| 亚洲专区字幕在线| 淫秽高清视频在线观看| 亚洲精品av麻豆狂野| 国产一卡二卡三卡精品| 国产精品永久免费网站| 国产又爽黄色视频| 97超级碰碰碰精品色视频在线观看| 国产成人欧美| 999精品在线视频| 男女下面进入的视频免费午夜 | 女人高潮潮喷娇喘18禁视频| 久久久精品国产亚洲av高清涩受| 欧美成人午夜精品| 日韩大码丰满熟妇| 国产亚洲av高清不卡| 久久久久九九精品影院| 日本黄色视频三级网站网址| 色精品久久人妻99蜜桃| 老司机在亚洲福利影院| 一级,二级,三级黄色视频| 色老头精品视频在线观看| 日本a在线网址| 欧美精品啪啪一区二区三区| 在线看a的网站| 一本综合久久免费| 18禁美女被吸乳视频| 一边摸一边抽搐一进一出视频| 日本a在线网址| 高清欧美精品videossex| videosex国产| 午夜成年电影在线免费观看| 99国产极品粉嫩在线观看| 操出白浆在线播放| 亚洲精品中文字幕在线视频| 男女之事视频高清在线观看| 国产熟女xx| 午夜福利免费观看在线| 国产激情欧美一区二区| 91精品三级在线观看| 大码成人一级视频| 自线自在国产av| 十八禁人妻一区二区| 午夜福利影视在线免费观看| 国产成人系列免费观看| 亚洲免费av在线视频| 免费看a级黄色片| 免费在线观看亚洲国产| 99久久久亚洲精品蜜臀av| 亚洲色图 男人天堂 中文字幕| 激情在线观看视频在线高清| 亚洲人成77777在线视频| 人人妻人人添人人爽欧美一区卜| 久久精品国产亚洲av香蕉五月| 女人精品久久久久毛片| 精品电影一区二区在线| 欧美人与性动交α欧美精品济南到| 成年版毛片免费区| 最近最新中文字幕大全免费视频| 久久久久久久久中文| 国产成人精品久久二区二区91| 亚洲五月色婷婷综合| 国产av又大| 午夜免费成人在线视频| 丰满人妻熟妇乱又伦精品不卡| 亚洲激情在线av| 啦啦啦在线免费观看视频4| 男女之事视频高清在线观看| 日本黄色视频三级网站网址| 97超级碰碰碰精品色视频在线观看| 亚洲国产精品999在线| 国产高清videossex| 黄色视频不卡| 欧美激情久久久久久爽电影 | 黄色片一级片一级黄色片| 99re在线观看精品视频| 两个人看的免费小视频| 国产有黄有色有爽视频| 麻豆成人av在线观看| 超色免费av| 欧美乱色亚洲激情| 在线观看免费日韩欧美大片| 男女午夜视频在线观看| 久久午夜综合久久蜜桃| 久久久久久久午夜电影 | 在线观看午夜福利视频| 亚洲精品在线观看二区| 亚洲精品国产色婷婷电影| 免费高清视频大片| 热99国产精品久久久久久7| av超薄肉色丝袜交足视频| 国产无遮挡羞羞视频在线观看| 国产欧美日韩综合在线一区二区| 免费少妇av软件| 高潮久久久久久久久久久不卡| 亚洲性夜色夜夜综合| 亚洲精品一卡2卡三卡4卡5卡| 99在线视频只有这里精品首页| 欧美日本亚洲视频在线播放| 久久热在线av| 免费观看人在逋| 国产99白浆流出| 亚洲第一欧美日韩一区二区三区| 亚洲成人国产一区在线观看| 精品国产国语对白av| 日韩人妻精品一区2区三区| 欧美+亚洲+日韩+国产| 欧美另类亚洲清纯唯美| 琪琪午夜伦伦电影理论片6080| 老司机亚洲免费影院| 成人手机av| 最好的美女福利视频网| 久久影院123| 亚洲av日韩精品久久久久久密| 欧美激情久久久久久爽电影 | 最新美女视频免费是黄的| 亚洲九九香蕉| 国产一区二区三区视频了| 亚洲成国产人片在线观看| 亚洲久久久国产精品| 搡老熟女国产l中国老女人| 一区二区三区精品91| 婷婷精品国产亚洲av在线| 国产高清国产精品国产三级| 视频区欧美日本亚洲| 又紧又爽又黄一区二区| 久久久久久亚洲精品国产蜜桃av| 18禁国产床啪视频网站| 无限看片的www在线观看| 在线观看午夜福利视频| 色在线成人网| 精品一区二区三区av网在线观看| 少妇裸体淫交视频免费看高清 | 极品教师在线免费播放| 天堂俺去俺来也www色官网| 美女 人体艺术 gogo| 18禁国产床啪视频网站| 超碰成人久久| 免费在线观看黄色视频的| 丰满的人妻完整版| 久久精品亚洲精品国产色婷小说| 电影成人av| 日本五十路高清| 国产aⅴ精品一区二区三区波| 国产精品自产拍在线观看55亚洲| 国产亚洲av高清不卡| www国产在线视频色| 在线永久观看黄色视频| 在线观看免费高清a一片| 国产成人影院久久av| 搡老岳熟女国产| 中文字幕av电影在线播放| 精品国产乱码久久久久久男人| 国产视频一区二区在线看| 性少妇av在线| 十八禁网站免费在线| 97人妻天天添夜夜摸| 国产亚洲欧美98| 99在线人妻在线中文字幕| 一级a爱视频在线免费观看| 欧美黄色淫秽网站| 日韩三级视频一区二区三区| 久热爱精品视频在线9| 欧美中文日本在线观看视频| 国产精品九九99| 在线观看66精品国产| 性欧美人与动物交配| 久久久水蜜桃国产精品网| 欧美精品亚洲一区二区| tocl精华| av视频免费观看在线观看| 9191精品国产免费久久| 如日韩欧美国产精品一区二区三区| 日本欧美视频一区| 久久青草综合色| 久久亚洲真实| 免费看a级黄色片| 99精品在免费线老司机午夜| 日韩高清综合在线|