• <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 ?

    An automatic identification algorithm for freeway bottleneckbased on loop detector data

    2014-09-06 10:49:51WangChaonanLiWenquanTongXiaolongChenChen
    關(guān)鍵詞:占有率自動(dòng)識別瓶頸

    Wang Chaonan Li Wenquan Tong Xiaolong Chen Chen

    (School of Transportation, Southeast University, Nanjing 210096, China)

    ?

    An automatic identification algorithm for freeway bottleneckbased on loop detector data

    Wang Chaonan Li Wenquan Tong Xiaolong Chen Chen

    (School of Transportation, Southeast University, Nanjing 210096, China)

    A bottleneck automatic identification algorithm based on loop detector data is proposed. The proposed algorithm selects the critical flow rate as the trigger variable of the algorithm, which is calculated by the road conditions, the level of service and the proportion of trucks. The process of identification includes two parts. One is to identify the upstream of the bottleneck by comparing the distance between the current occupancy rate and the mean value of the occupancy rate and the variance of the occupancy rate. The other process is to identify the downstream of the bottleneck by calculating the difference of the upstream occupancy rate with that of the downstream. In addition, the algorithm evaluation standards, which are based on the time interval of the data, the detection rate and the false alarm rate, are discussed. The proposed algorithm is applied to detect the bottleneck locations in the Shanghai Inner Ring Viaduct Dabaishu-Guangzhong road section. The proposed method has a good performance in improving the accuracy and efficiency of bottleneck identification.

    bottleneck; loop detector data; occupancy rate; flow rate

    Typical bottleneck analysis methods involve congestion graphics, vehicle arrival curves and the traffic flow theory. The methods above are based on the loop detector data. They assume that traffic patterns can be determined by the functional relationship of traffic flow and occupancy. The data analysis methods include decision trees, statistical analysis and filtering. With the development of data collection and information processing technology, the fuzzy theory, expert evaluation, pattern recognition, and artificial intelligence techniques have become important methods in congestion automatic identification algorithms. ACI algorithms can be divided into discrete and continuous types[1-2]. The discrete method is based on the traffic parameter threshold. It is assumed that congestion occurs when a certain traffic parameter is greater than the threshold value. In 1997, with the vehicle arrival curves, Cassidy et al.[3]discussed the length of queues and waiting time in congestion. Based on the research findings of congestion, Bertini et al.[4]proposed an automatic bottleneck recognition algorithm in 2005, identifying and classifying bottlenecks with historical data, and taking speed as the indicator of a bottleneck. However, these algorithms have not analyzed the statistical features of a bottleneck in detail, which decrease the precision of identification algorithms. Besides, most congestion identification algorithms were about congestion points. The research target of bottleneck identification is to discover the congestion influence scope, including the upstream and downstream of bottleneck locations. Therefore, a traffic flow analysis with more integrity is necessary.

    In this research, based on loop detector data, the critical flow rate is calculated as the trigger variable, which is calculated by the road conditions, the level of service and the proportion of trucks. Occupancy is calculated as the identification parameters. In addition, algorithm evaluation standards are discussed.

    1 Description of the Algorithm

    In this algorithm, the critical flow rate is defined as the trigger variable, and it is calculated based on the road conditions, the level of service and the proportion of trucks. When the flow rate is greater than the critical value, the identification process can start[5-6].

    1.1 Trigger variable of the algorithm

    According to the traffic flow theory, when traffic demand exceeds road capacity, resulting in congestion, the flow rate decreases and congestion forms. Therefore, the flow rate can be used as the trigger variable of the algorithm. The critical flow rate is determined by the designed capacity and the level of service. User perceptions are variable at different time and in different environments. AASHTO design standards suggest that a good target level of service in an urban area is D, but in a rural area a good target level of service is C. Generally, when the level of service is C, average speeds begin to decline with increasing flow[7]. We define the situation, in which the level of service is C, as the trigger point. However, it is difficult for the bottleneck automatic identification system to calculate theV/Cratio. The trigger value should be obtained immediately. Therefore, the accumulative flow rate in 5 min as the trigger variable is chosen. Tab.1 lists the criticalV/Cratio based on the level of service and the design speed. The capacity should be multiplied by theV/Cratio when the level of service is C to obtain the trigger value.

    Tab.1 Critical V/Cratio based on LOS and FFS

    The influence of large vehicle on traffic flow is not considered when calculating the free flow speed. Therefore, the trigger value obtained from Tab.1 should be multiplied byfHV.

    1.2 Identification process

    There are two elementary values which need to be defined in the bottleneck automatic identification algorithm. One is the minimum value of the difference between the upstream and downstream occupancy rates. The other one is the minimum value of the upstream occupancy rate. To define the difference between the upstream and downstream occupancy, it must be ensured that the upstream is in a state of congestion, and the downstream is free flow[8-10].

    Before defining the critical value, we should analyze the historical data to obtain the difference between the upstream and downstream occupancy rates. Experiments show that there are some overlaps in the difference between them. As this algorithm also requires the upper occupancy rate greater than a certain critical value, it should be ensured that the critical value of the difference between the upstream and downstream occupancy rates is smaller than the occupancy rate when a bottleneck occurs.

    積累數(shù)學(xué)基本活動(dòng)經(jīng)驗(yàn)是一個(gè)長期的過程,需要我們在平時(shí)的教學(xué)中不斷為學(xué)生提供活動(dòng)的機(jī)會(huì),精心設(shè)計(jì)組織好每一個(gè)數(shù)學(xué)活動(dòng),使數(shù)學(xué)學(xué)習(xí)成為一個(gè)生動(dòng)活潑、富有創(chuàng)造意義的過程,促進(jìn)學(xué)生思維的發(fā)展。

    As the maximum upstream occupancy rate is affected by the road conditions, we use the distance between the current occupancy rate and the mean value of the occupancy rate fromts-1tots-6instead of the maximum upstream occupancy rate. The distance between the current occupancy rate and the mean value of the occupancy rate is equal to the current occupancy rate subtracting the mean value of occupancy rates fromts-1tots-6.

    Doi=O(i,ts)-E(Ots-1,…,Ots-6)

    (1)

    whereDoiis the distance between the current occupancy rate and the mean value of the occupancy rate fromts-1tots-6;O(i,ts) is the occupancy rate atts;E(Ots-1,…,Ots-6) is the mean value of the occupancy rate fromts-1tots-6.

    IfDoiis greater than the variance of the occupancy rate fromts-1tots-6, we assume that unitiis the upstream of the bottleneck.

    Doi=S(Ots-1,…,Ots-6)+α

    (2)

    whereS(Ots-1,…,Ots-6) is the variance of the occupancy rate fromts-1tots-6;αis the adjustment parameter.

    Based on the definition of a bottleneck, the downstream of the bottleneck is free flow, which means that there is a value difference between the upstream and downstream occupancy. Experiments show that regional differences have an impact on the critical value of the difference between the upstream and downstream occupancy. Data analysis is a common method to determine the critical value of difference, but the analysis process is very complicated. In order to make the parameter in this algorithm universal, we use an adjustment parameter. To speed up the process of bottleneck identification, we also use the mean value of occupancy rate fromts-1tots-6.

    The difference between the upstream and downstream occupancy rates equals the current occupancy rate at pointI, which subtracts the downstream occupancy rate. It should be noted that the value ofnin the following formula requires several further tests.

    P(ts)=O(i,ts)-O(i+n,ts)

    (3)

    whereP(ts) is the difference between the upstream occupancy rate and that of the downstream;O(i,ts) is the downstream occupancy rate at pointi;O(i+n,ts) is the occupancy rate,n=1, 2,…

    The critical value of difference after adjustment equals the difference between the upstream occupancy and the downstream occupancy, divided by the mean value of the occupancy rate fromts-1tots-6.

    Pr(ts)=P(ts)/E(ots-1,…,ots-6)

    (4)

    wherePr(ts) is the critical value of difference after adjustment;E(ots-1,…,ots-6) represents the mean value of the occupancy rate fromts-1tots-6.

    The parameteriin Eq.(3) is defined as the section of pavement units when identifying the location of the bottleneck. The value ofiequals the longitudinal space of loop detectors, which is 20 or 50 m on the expressway.

    The parametertis defined as the data aggregation level. The original loop detector data is collected every 20 s, which is highly volatile. Different data aggregation levels can affect the promptness and accuracy of the algorithm. So the original loop detector data should be facilitated aggregation before being used in the algorithm. Details will be discussed in the following section.

    Fig.1 shows the process of bottleneck automatic identification.

    Fig.1 The process of bottleneck automatic identification

    As shown in Fig.1, the first step of the bottleneck identification algorithm is to divide the road section into several units and to determine data aggregation level. Then, the critical rate of flow is calculated. After preprocessing,f(i,t) is compared with the critical rate. Iff(i,t) is greater, the process of bottleneck identification can start. The process of identification includes two parts. One is to identify the upstream of the bottleneck by comparing the distance between the current occupancy rate and the mean value of occupancy rateDoiwith the variance of the occupancy rate. IfDoiis greater than the variance of the occupancy rate fromts-1tots-6, assume that unitiis the upstream of bottleneck. The other is to identify the downstream of bottleneck by calculatingPr(ts). IfPr(ts) is greater than 0.2, assume that uniti+nis the downstream of bottleneck.

    1.3 Evaluation model

    The accuracy of this automatic recognition algorithm of the bottleneck should be determined by three aspects. The first aspect is the percentage of the bottleneck points that can be recognized at all the bottleneck points; the second is the percentage of “fake bottleneck points” in the bottleneck points that are recognized; and the third is the time interval of the date which can also influence the sensitivity of the algorithm. In practice, regional differences make it difficult to balance the failure of recognizing the real bottleneck points and the mistaken recognition of “fake bottleneck points”. So the optimal choice of parameters depends on the user’s choice of a composite score function that takes the relative costs of missed bottlenecks and false alarms into account.

    The percentage of the bottleneck points recognized equals the number of the bottleneck points which have been recognized divided by the number of the total bottleneck points. While the percentage of “fake bottleneck points” equals the number of the fake bottleneck points which have been recognized, divided by the number of the total bottleneck points which have also been recognized.

    (5)

    whereTis the percentage of the bottleneck points recognized;tbis the number of the bottleneck points which have been recognized;nbrepresents the number of the whole bottleneck points.

    (6)

    whereFmeans the percentage of “fake bottleneck points”;fbis the number of the fake bottleneck points which have been mistaken;ibis the number of the total bottleneck points which have been recognized.

    The accuracy of this automatic recognition algorithm of bottleneck can be calculated as

    S=β(αTT-αFF)

    (7)

    whereSis the accuracy of this automatic recognition algorithm of the bottleneck;βis the correction coefficient determined by the time interval;αTis the penalty weight of the success rate;αFis the penalty weight of the false-alarm rate.

    2 Actual Analysis

    One-week (from Sept 21 to Sept 27, 2012) loop detector data on the Shanghai Inner Ring Viaduct (5:00—11:00, 14:00—20:00) were obtained from the Shanghai Transportation Operation Department. The research scope starts from Dabaishu and runs to Guangzhong Road.

    The first step of the bottleneck identification algorithm is to divide the road sections into several units and determine the data aggregation level. In this research, the number of road units is determined by the spacing of loop detectors. If the loop detector data is aggregated into 1-min data, the algorithm should be restricted by some additional conditions. For example, a sustained bottleneck filter is added to smooth the results of the algorithm. This filter discards false positives that are isolated in the time dimension from other detections at the same location.

    It can be seen from Fig.1 that when loop detector data is aggregated into 3-min data sets, the upstream occupancy remains greater than the downstream occupancy but the difference in values fluctuates greatly. To obtain higher recognition rate, the difference between the upstream and downstream occupancy should be set to be small, which will increase the possibility of false alarm bottlenecks. Then we facilitate further aggregation into 5-min and 10-min data sets. As Figs.2(b) and (c) show, the occupancy rates in bottlenecks (Oi) remain stable.

    (a)

    (b)

    (c)

    Based on the above analysis, we conclude that the smaller time interval leads to higher sensitivity. Experiments show that 5-min aggregation data fits the algorithm best.

    In Tab.1, criticalV/Cratio is 0.83 when the level of service is C and the design speed is 80 km/h. Then,fHVis 0.9. The design capacity of the Shanghai Inner Ring Viaduct is 1 800 (pcu·h-1)/lane. Thus, the final critical flow rate in 5 min is 108 pcu.

    Then we input the 5-min data sets to Matlab. When the flow rate is greater than 108 pcu, the algorithm starts. IfDoiis greater than the variance of the occupancy rate fromts-1tots-6, assume that unitiis the upstream of the bottleneck. IfPr(ts) is greater than 0.2, assume that uniti+nis the downstream of the bottleneck.

    The session is identified by Matlab in a space-time distribution airstrip, as shown in Fig.3.

    Fig.3 Bottleneck identified with the algorithm

    Fig.3 presents the bottleneck locations identified which are marked with the triangles. In this analysis, three severe bottleneck locations were found.

    The point where No.56 loop detector is located becomes congested at 6:55, giving rise to congestion and vehicle delays between the No.55 loop detector and No.56 loop detector. The occupancy rate starts to decrease at 9:20, which means that the congestion has started to be relieved. The road section between the No.56 loop detector and No.58 loop detector is free flow between 6:00 and 6:35. Then the occupancy rate begins to increase and the point where the No.58 loop detector is located becomes congested at 7:10. The occupancy rate begins to decrease at 8:40. Finally, the road section has free flow at 9:45.

    3 Evaluation

    In section 2, we analyze the distribution of the bottleneck in time and space through identifying bottleneck locations and their activation and deactivation periods. On the other hand, we are informed of the real distribution of bottlenecks through video data. Based on this, the accuracy of the proposed algorithm is confirmed. The evaluation consists of a series of indices, including the detection rateTand the false alarm rateF.

    It can be seen from Tab.2 that the proposed algorithm has a reasonable detection performance compared to other methods. Otherwise, the percentage of fake bottleneck points is 0.1. The critical flow rate and the occupancy threshold value are decreased to increaseT. The parameter values can be optimized to decreaseFin application.

    Tab.2 Evaluation results of the proposed algorithm

    The proposed algorithm also outperforms previous cumulative curve methods in terms of precision and identification efficiency. In the cumulative curve method, each identification process is subject to interference by earlier data. In the proposed algorithm, the identifying process is only determined by the current data.

    4 Conclusion

    1) The statistical analysis of loop detector data shows that the flow rate and occupancy are more reliable parameters than speed for bottleneck identification, excluding affection by speed limits.

    2) Loop detector data should be aggregated before the identification process. Experiments show that a 5-min aggregation data fits the algorithm best.

    3) A bottleneck identification algorithm based on the flow rate and occupancy is proposed. The proposed algorithm includes the trigger variable, the identification process and the evaluation model. The algorithm reduces the influence of road conditions and data error by parameter optimization. The results show that the proposed algorithm has a good performance in improving the accuracy of bottleneck identification.

    4) Although the results are encouraging, a number of extensions to the algorithm need to be studied. Further research should be carried out to validate this algorithm on other types of roads. Note also that the original loop detectors data should be aggregated before analysis and that we need toseek a better method of data aggregation to increase the speed of the algorithm.

    [1]Chung K, Rudjanakanoknad J, Cassidy M J. Relation between traffic density and capacity drop at three freeway bottlenecks[J].TransportationResearchPartB:Methodological, 2007, 41(1): 82-95.

    [2]Banks J H. Review of empirical research on congested freeway flow[J].TransportationResearchRecord, 2002, 1802: 225-232.

    [3]Cassidy M J, Windover J R. Methodology for assessing dynamics of freeway traffic flow[J].TransportationResearchRecord, 1995(1484): 73-79.

    [4]Bertini R L, Myton A. Using PeMS data to empirically diagnose freeway bottleneck locations in Orange County, California[J].TransportationResearchRecord, 1925, 2005: 48-57.

    [5]Shoraka M, Puan O C. Review of evaluating existing capacity of weaving segments[J].InternationalJournalofCivilandStructuralEngineering, 2010, 1(3): 683-694.

    [6]Coifman B, Mishalani R, Wang C, et al. Impact of lane-change maneuvers on congested freeway segment delays: pilot study [J].TransportationResearchRecord, 2006, 1965: 152-159.

    [7]Cassidy M J, Mauch M. An observed traffic pattern in long freeway queues[J].TransportationResearchPartA:PolicyandPractice, 2001, 35(2): 143-156.

    [8]Kumar R, Wolenetz M, Agarwalla B, et al. DFuse:a framework for distributed data fusion[C]//Proceedingsofthe1stInternationalConferenceonEmbeddedNetworkedSensorSystems. Los Angeles, CA, USA, 2003:114-125.

    [9]Newell G F. A simplified car-following theory: a lower order model[J].TransportationResearchPartB:Methodological, 2002, 36(3): 195-205.

    [10]Cassidy M J. Bivariate relations in nearly stationary highway traffic[J].TransportationResearchPartB:Methodological,1998, 32(1): 49-59.

    基于線圈數(shù)據(jù)的瓶頸點(diǎn)自動(dòng)識別算法

    王超楠 李文權(quán) 童小龍 陳 晨

    (東南大學(xué)交通學(xué)院, 南京 210096)

    提出了一種基于線圈數(shù)據(jù)的瓶頸點(diǎn)自動(dòng)識別算法.算法以臨界流量作為算法的觸發(fā)變量,根據(jù)道路條件、服務(wù)水平和大型車比例計(jì)算臨界流量.算法的識別程序包括2部分:首先通過計(jì)算當(dāng)前占有率與前時(shí)刻占有率的相對差值來判定瓶頸點(diǎn)上游位置;然后通過計(jì)算上游占有率與下游占有率的相對差值確定瓶頸點(diǎn)下游的位置.此外,提出了基于數(shù)據(jù)集計(jì)周期、瓶頸點(diǎn)識別率和誤判率的算法性能評價(jià)方法.利用上海市內(nèi)環(huán)高架大柏樹-廣中路段的線圈數(shù)據(jù)進(jìn)行試驗(yàn),結(jié)果表明,瓶頸點(diǎn)自動(dòng)識別算法在準(zhǔn)確率和效率上有顯著提高.

    瓶頸點(diǎn);線圈數(shù)據(jù);占有率;流量

    U491.2

    Received 2014-02-22.

    Biographies:Wang Chaonan (1990—), female, graduate; Li Wenquan(corresponding author), male, doctor, professor, wenqli@seu.edu.cn.

    :Wang Chaonan, Li Wenquan, Tong Xiaolong, et al. An automatic identification algorithm for freeway bottleneck based on loop detector data[J].Journal of Southeast University (English Edition),2014,30(4):495-499.

    10.3969/j.issn.1003-7985.2014.04.016

    10.3969/j.issn.1003-7985.2014.04.016

    猜你喜歡
    占有率自動(dòng)識別瓶頸
    數(shù)據(jù)參考
    自動(dòng)識別系統(tǒng)
    特別健康(2018年3期)2018-07-04 00:40:18
    微軟領(lǐng)跑PC操作系統(tǒng)市場 Win10占有率突破25%
    突破霧霾治理的瓶頸
    金屬垃圾自動(dòng)識別回收箱
    基于IEC61850的配網(wǎng)終端自動(dòng)識別技術(shù)
    電測與儀表(2016年6期)2016-04-11 12:06:38
    突破瓶頸 實(shí)現(xiàn)多贏
    滁州市中小學(xué)田徑場地現(xiàn)狀調(diào)查與分析
    9月服裝銷售疲軟
    中國服飾(2014年11期)2015-04-17 06:48:50
    如何渡過初創(chuàng)瓶頸期
    五月天丁香电影| 大陆偷拍与自拍| 午夜激情福利司机影院| 日韩精品有码人妻一区| 精品亚洲乱码少妇综合久久| 亚洲自偷自拍三级| 纵有疾风起免费观看全集完整版| 直男gayav资源| 黄色怎么调成土黄色| 交换朋友夫妻互换小说| 久久久a久久爽久久v久久| 午夜老司机福利剧场| 国产黄片视频在线免费观看| 日韩av不卡免费在线播放| 99热这里只有精品一区| 天天躁日日操中文字幕| 22中文网久久字幕| 国产欧美亚洲国产| 男女那种视频在线观看| 性色avwww在线观看| 一本色道久久久久久精品综合| 久久精品人妻少妇| 伦精品一区二区三区| 高清毛片免费看| 日韩亚洲欧美综合| 91久久精品国产一区二区三区| 日日啪夜夜撸| 成人亚洲精品一区在线观看 | 午夜免费男女啪啪视频观看| 国产亚洲一区二区精品| 久久久精品欧美日韩精品| 成人黄色视频免费在线看| 99久久精品热视频| 国产色婷婷99| 美女内射精品一级片tv| 亚洲内射少妇av| 国产一区二区在线观看日韩| 国产精品久久久久久久电影| 免费黄频网站在线观看国产| 国产成人aa在线观看| 成人漫画全彩无遮挡| 亚洲综合色惰| 国产午夜福利久久久久久| 精品久久久噜噜| 亚洲欧美一区二区三区国产| 欧美精品国产亚洲| 两个人的视频大全免费| 成人综合一区亚洲| 毛片女人毛片| 久久精品综合一区二区三区| 日日啪夜夜撸| 国产一区二区三区综合在线观看 | 国产精品国产av在线观看| 高清欧美精品videossex| 男女边吃奶边做爰视频| av网站免费在线观看视频| 欧美高清成人免费视频www| 国产精品蜜桃在线观看| 天天躁日日操中文字幕| 人妻夜夜爽99麻豆av| 不卡视频在线观看欧美| 人妻 亚洲 视频| 精品人妻一区二区三区麻豆| 韩国av在线不卡| 男人和女人高潮做爰伦理| 免费大片18禁| 国产欧美另类精品又又久久亚洲欧美| 亚洲熟女精品中文字幕| 久久精品人妻少妇| 欧美日韩综合久久久久久| 国产成人午夜福利电影在线观看| 高清毛片免费看| 久久久久国产精品人妻一区二区| 18禁在线播放成人免费| 国产一区亚洲一区在线观看| 午夜日本视频在线| 亚洲国产日韩一区二区| 欧美潮喷喷水| 久久久久久久大尺度免费视频| 久久午夜福利片| 2018国产大陆天天弄谢| 身体一侧抽搐| 亚洲欧美清纯卡通| 美女视频免费永久观看网站| 大片免费播放器 马上看| 在现免费观看毛片| 欧美+日韩+精品| 18禁动态无遮挡网站| 欧美xxxx黑人xx丫x性爽| 18禁在线无遮挡免费观看视频| 欧美性猛交╳xxx乱大交人| 成人黄色视频免费在线看| 三级经典国产精品| 成人欧美大片| 热re99久久精品国产66热6| 美女cb高潮喷水在线观看| 噜噜噜噜噜久久久久久91| 国产精品国产三级国产av玫瑰| 国产精品国产三级专区第一集| 全区人妻精品视频| 欧美老熟妇乱子伦牲交| 欧美成人午夜免费资源| 水蜜桃什么品种好| 久久午夜福利片| 视频中文字幕在线观看| 草草在线视频免费看| 男的添女的下面高潮视频| 波野结衣二区三区在线| 高清视频免费观看一区二区| 联通29元200g的流量卡| 少妇熟女欧美另类| 国产精品熟女久久久久浪| 婷婷色综合www| 国产精品久久久久久久电影| 成人综合一区亚洲| 尤物成人国产欧美一区二区三区| 中文字幕亚洲精品专区| 久久99热6这里只有精品| 午夜爱爱视频在线播放| 日韩免费高清中文字幕av| 国产成人a∨麻豆精品| 久久人人爽av亚洲精品天堂 | 交换朋友夫妻互换小说| 九草在线视频观看| 欧美日本视频| 大又大粗又爽又黄少妇毛片口| 国产一区有黄有色的免费视频| 日本黄色片子视频| 简卡轻食公司| 国产人妻一区二区三区在| 一本一本综合久久| 18禁在线无遮挡免费观看视频| 美女内射精品一级片tv| 听说在线观看完整版免费高清| 国产亚洲91精品色在线| 色吧在线观看| 久久精品国产自在天天线| 国产精品偷伦视频观看了| 人体艺术视频欧美日本| 高清视频免费观看一区二区| 久久久久国产网址| 高清毛片免费看| 免费观看的影片在线观看| 国产黄片视频在线免费观看| 亚洲av福利一区| 97超视频在线观看视频| 99久久精品热视频| 又大又黄又爽视频免费| 国产极品天堂在线| 婷婷色麻豆天堂久久| 欧美区成人在线视频| 大陆偷拍与自拍| 99视频精品全部免费 在线| 国产精品一区二区在线观看99| 成人免费观看视频高清| 人人妻人人澡人人爽人人夜夜| 亚洲欧美日韩东京热| kizo精华| 欧美+日韩+精品| 国产伦理片在线播放av一区| 久热这里只有精品99| a级一级毛片免费在线观看| 欧美97在线视频| 男女无遮挡免费网站观看| 亚洲天堂国产精品一区在线| av黄色大香蕉| 日本一本二区三区精品| 国产欧美日韩一区二区三区在线 | 久久久精品94久久精品| 中国三级夫妇交换| 99久久精品热视频| 亚洲美女视频黄频| 少妇被粗大猛烈的视频| 日本色播在线视频| 国产精品福利在线免费观看| 久久国产乱子免费精品| 免费av毛片视频| 777米奇影视久久| 免费av毛片视频| 99热网站在线观看| 亚洲综合色惰| 汤姆久久久久久久影院中文字幕| 免费看a级黄色片| 亚洲精品成人久久久久久| 久久99蜜桃精品久久| 日本色播在线视频| 精品国产乱码久久久久久小说| 日本与韩国留学比较| 亚洲激情五月婷婷啪啪| 一区二区三区精品91| 日本色播在线视频| www.色视频.com| 七月丁香在线播放| 2018国产大陆天天弄谢| 嫩草影院精品99| 免费观看无遮挡的男女| 亚洲,一卡二卡三卡| 六月丁香七月| 大香蕉97超碰在线| 在现免费观看毛片| 中文乱码字字幕精品一区二区三区| 高清欧美精品videossex| 欧美xxxx性猛交bbbb| 街头女战士在线观看网站| 国产伦在线观看视频一区| 人妻制服诱惑在线中文字幕| 亚洲欧美日韩东京热| 老司机影院毛片| 寂寞人妻少妇视频99o| 三级国产精品欧美在线观看| 两个人的视频大全免费| 亚洲精品视频女| 亚洲精品视频女| 国产精品偷伦视频观看了| 成人国产麻豆网| 国产精品秋霞免费鲁丝片| 亚洲一级一片aⅴ在线观看| 亚洲自偷自拍三级| 亚洲欧洲日产国产| 欧美一区二区亚洲| 午夜福利视频精品| 亚洲va在线va天堂va国产| 人妻少妇偷人精品九色| 国产精品一二三区在线看| 女人被狂操c到高潮| 欧美成人午夜免费资源| 又爽又黄a免费视频| 亚洲天堂国产精品一区在线| 亚洲av在线观看美女高潮| videos熟女内射| 国产免费又黄又爽又色| 久久久精品免费免费高清| 人妻一区二区av| 成人免费观看视频高清| 夫妻性生交免费视频一级片| 三级男女做爰猛烈吃奶摸视频| 高清午夜精品一区二区三区| 天天躁夜夜躁狠狠久久av| 亚洲国产高清在线一区二区三| 亚洲av欧美aⅴ国产| 亚洲av国产av综合av卡| 亚洲一区二区三区欧美精品 | 深爱激情五月婷婷| av在线播放精品| 国产精品久久久久久av不卡| 禁无遮挡网站| 亚洲精品成人av观看孕妇| 热re99久久精品国产66热6| 高清在线视频一区二区三区| 成人午夜精彩视频在线观看| 99re6热这里在线精品视频| 欧美最新免费一区二区三区| av在线蜜桃| 亚洲成人中文字幕在线播放| 免费不卡的大黄色大毛片视频在线观看| 亚洲熟女精品中文字幕| 汤姆久久久久久久影院中文字幕| 亚洲精品成人久久久久久| 亚洲成人中文字幕在线播放| 国产爱豆传媒在线观看| 国产免费一级a男人的天堂| av福利片在线观看| 日韩欧美 国产精品| 黄片无遮挡物在线观看| 亚洲伊人久久精品综合| 久久久精品94久久精品| 久久97久久精品| 亚洲av免费高清在线观看| 欧美成人精品欧美一级黄| 涩涩av久久男人的天堂| 一级毛片aaaaaa免费看小| 白带黄色成豆腐渣| 亚洲精品亚洲一区二区| 国产精品秋霞免费鲁丝片| 成人特级av手机在线观看| 天天躁日日操中文字幕| 亚洲成人久久爱视频| 欧美高清性xxxxhd video| 国产 一区 欧美 日韩| 亚洲性久久影院| 一级毛片黄色毛片免费观看视频| 少妇人妻久久综合中文| av在线天堂中文字幕| 亚洲国产最新在线播放| 蜜桃亚洲精品一区二区三区| 国产色爽女视频免费观看| 一个人看视频在线观看www免费| 制服丝袜香蕉在线| 国产精品秋霞免费鲁丝片| 激情 狠狠 欧美| 插阴视频在线观看视频| 麻豆精品久久久久久蜜桃| 亚洲天堂国产精品一区在线| 欧美激情国产日韩精品一区| 国产午夜福利久久久久久| 五月伊人婷婷丁香| 噜噜噜噜噜久久久久久91| 男插女下体视频免费在线播放| 免费看av在线观看网站| 亚洲av中文字字幕乱码综合| 色综合色国产| av网站免费在线观看视频| 少妇人妻 视频| 精品国产三级普通话版| 久久精品人妻少妇| 99久久人妻综合| 欧美性感艳星| 精品视频人人做人人爽| 国产淫片久久久久久久久| 91在线精品国自产拍蜜月| 国产成人精品婷婷| 欧美少妇被猛烈插入视频| av卡一久久| 女人被狂操c到高潮| 看黄色毛片网站| 天堂中文最新版在线下载 | 日韩精品有码人妻一区| 久久人人爽人人爽人人片va| 嫩草影院入口| 美女高潮的动态| 亚洲人成网站在线观看播放| 岛国毛片在线播放| 日产精品乱码卡一卡2卡三| 中国三级夫妇交换| 亚洲国产精品成人综合色| 日韩一区二区三区影片| 中文字幕久久专区| 大陆偷拍与自拍| 午夜福利高清视频| 菩萨蛮人人尽说江南好唐韦庄| 日韩成人av中文字幕在线观看| 男人爽女人下面视频在线观看| 九九爱精品视频在线观看| 中国美白少妇内射xxxbb| 日本免费在线观看一区| 久久精品国产自在天天线| 久久综合国产亚洲精品| 国产精品久久久久久精品电影小说 | 国产午夜福利久久久久久| 国产精品秋霞免费鲁丝片| 日韩成人av中文字幕在线观看| 国产人妻一区二区三区在| 老师上课跳d突然被开到最大视频| 网址你懂的国产日韩在线| 黄片wwwwww| 国产高清国产精品国产三级 | 久久人人爽av亚洲精品天堂 | 在线观看免费高清a一片| 麻豆成人午夜福利视频| 国产综合懂色| 国产白丝娇喘喷水9色精品| eeuss影院久久| 在线观看美女被高潮喷水网站| 乱系列少妇在线播放| 真实男女啪啪啪动态图| .国产精品久久| 美女主播在线视频| 少妇裸体淫交视频免费看高清| 天堂俺去俺来也www色官网| 精品一区二区三卡| 国产有黄有色有爽视频| 亚洲综合色惰| 欧美xxxx黑人xx丫x性爽| 免费黄网站久久成人精品| 国产午夜精品一二区理论片| 成人毛片a级毛片在线播放| 我的女老师完整版在线观看| 97热精品久久久久久| 尤物成人国产欧美一区二区三区| 成人高潮视频无遮挡免费网站| 国产一区二区三区av在线| 在线观看国产h片| 免费在线观看成人毛片| 免费看不卡的av| 午夜日本视频在线| 日韩强制内射视频| 51国产日韩欧美| 小蜜桃在线观看免费完整版高清| 国产女主播在线喷水免费视频网站| 亚洲综合色惰| 性色avwww在线观看| 新久久久久国产一级毛片| 欧美日韩综合久久久久久| 国产精品久久久久久精品电影| av卡一久久| 新久久久久国产一级毛片| 欧美三级亚洲精品| 日韩三级伦理在线观看| 国产亚洲av嫩草精品影院| 男女那种视频在线观看| 黄色日韩在线| 亚洲伊人久久精品综合| 午夜福利视频精品| 欧美极品一区二区三区四区| 国产毛片a区久久久久| 少妇的逼好多水| 国产成人aa在线观看| 亚洲人成网站在线观看播放| 亚洲精品乱码久久久久久按摩| 男女国产视频网站| 国产亚洲一区二区精品| 欧美激情国产日韩精品一区| 国产午夜精品久久久久久一区二区三区| 欧美性感艳星| 特大巨黑吊av在线直播| 黄色日韩在线| 亚洲一级一片aⅴ在线观看| 好男人在线观看高清免费视频| 精品久久久久久久人妻蜜臀av| 久久99精品国语久久久| 国精品久久久久久国模美| av黄色大香蕉| 身体一侧抽搐| 欧美国产精品一级二级三级 | 大陆偷拍与自拍| xxx大片免费视频| 亚洲欧美中文字幕日韩二区| 97超碰精品成人国产| 在线观看一区二区三区| 18禁在线无遮挡免费观看视频| 中文在线观看免费www的网站| 欧美精品一区二区大全| 嫩草影院精品99| 精品99又大又爽又粗少妇毛片| 成人亚洲欧美一区二区av| 97超视频在线观看视频| 国产片特级美女逼逼视频| 尾随美女入室| 九九久久精品国产亚洲av麻豆| 亚洲欧美清纯卡通| 国产午夜精品一二区理论片| 亚洲国产精品999| 九九在线视频观看精品| 色吧在线观看| 国产成人aa在线观看| 亚洲欧美日韩东京热| 少妇人妻久久综合中文| 亚洲精品日本国产第一区| 在线播放无遮挡| 久久精品久久久久久噜噜老黄| 久久99热这里只频精品6学生| 国产亚洲av片在线观看秒播厂| 成人毛片60女人毛片免费| 听说在线观看完整版免费高清| 中文字幕亚洲精品专区| 久久久久久久久久久免费av| 日日撸夜夜添| 国产精品一区www在线观看| 亚洲在久久综合| 色吧在线观看| 日韩 亚洲 欧美在线| 日本熟妇午夜| 各种免费的搞黄视频| 97精品久久久久久久久久精品| 亚洲在久久综合| 18禁裸乳无遮挡动漫免费视频 | 免费观看av网站的网址| 听说在线观看完整版免费高清| 91精品国产九色| 亚洲电影在线观看av| 久久久久久国产a免费观看| 嘟嘟电影网在线观看| 国产有黄有色有爽视频| 成人特级av手机在线观看| 天堂俺去俺来也www色官网| 三级男女做爰猛烈吃奶摸视频| 男女边吃奶边做爰视频| 免费看日本二区| 亚洲欧美精品自产自拍| 少妇裸体淫交视频免费看高清| av在线app专区| 久久精品人妻少妇| 国精品久久久久久国模美| 久久这里有精品视频免费| 亚洲精品成人久久久久久| 日韩亚洲欧美综合| 亚洲精品国产av成人精品| 女的被弄到高潮叫床怎么办| 三级国产精品欧美在线观看| 精品久久久久久久久av| 亚洲精品国产av蜜桃| 全区人妻精品视频| 一级二级三级毛片免费看| 中文字幕制服av| 久久热精品热| 亚洲精品影视一区二区三区av| 国产精品国产av在线观看| 国产精品久久久久久精品电影| a级毛色黄片| 在线a可以看的网站| 成人漫画全彩无遮挡| 国产 一区 欧美 日韩| 国产黄频视频在线观看| 禁无遮挡网站| 天天一区二区日本电影三级| 国产成人精品婷婷| 亚洲美女搞黄在线观看| 婷婷色麻豆天堂久久| 日本一二三区视频观看| 高清视频免费观看一区二区| 高清日韩中文字幕在线| 欧美3d第一页| 成人毛片60女人毛片免费| 精品久久久久久久末码| 国产精品不卡视频一区二区| 大码成人一级视频| 久久精品国产亚洲av天美| 日韩av不卡免费在线播放| 亚洲国产最新在线播放| 91狼人影院| 一级片'在线观看视频| 久久精品国产亚洲av天美| 国产黄频视频在线观看| 人妻少妇偷人精品九色| 日韩欧美 国产精品| 在现免费观看毛片| 久久久久久久久久久免费av| 欧美性猛交╳xxx乱大交人| 观看免费一级毛片| 免费播放大片免费观看视频在线观看| 99热全是精品| 国产女主播在线喷水免费视频网站| 国产成人a∨麻豆精品| av在线亚洲专区| 色哟哟·www| 99热这里只有是精品50| 亚洲av成人精品一二三区| 性色av一级| 日日啪夜夜撸| 午夜爱爱视频在线播放| 精品国产一区二区三区久久久樱花 | 国产中年淑女户外野战色| 日本黄大片高清| 亚洲精品影视一区二区三区av| 国产成人一区二区在线| 久久ye,这里只有精品| 免费人成在线观看视频色| 精品一区二区免费观看| 午夜精品国产一区二区电影 | 国产亚洲午夜精品一区二区久久 | 亚洲欧洲国产日韩| 99热这里只有是精品在线观看| 纵有疾风起免费观看全集完整版| 丰满少妇做爰视频| 成人亚洲精品一区在线观看 | 国产精品成人在线| 丝袜脚勾引网站| 日本免费在线观看一区| av国产久精品久网站免费入址| 国产毛片a区久久久久| 亚洲av一区综合| 欧美变态另类bdsm刘玥| 麻豆国产97在线/欧美| 欧美97在线视频| 在线观看三级黄色| 日本黄大片高清| 亚洲欧美清纯卡通| 国产欧美亚洲国产| 免费观看性生交大片5| 在线精品无人区一区二区三 | 一本—道久久a久久精品蜜桃钙片 精品乱码久久久久久99久播 | 国产大屁股一区二区在线视频| 亚洲av成人精品一二三区| 国产91av在线免费观看| 国产精品国产三级专区第一集| 熟女人妻精品中文字幕| 纵有疾风起免费观看全集完整版| 婷婷色综合www| 美女脱内裤让男人舔精品视频| 亚洲高清免费不卡视频| 九草在线视频观看| 校园人妻丝袜中文字幕| 内射极品少妇av片p| 国产黄a三级三级三级人| 在线观看人妻少妇| 最近最新中文字幕免费大全7| 又爽又黄a免费视频| 在线观看一区二区三区| 欧美极品一区二区三区四区| 精华霜和精华液先用哪个| 亚洲成人一二三区av| 国产极品天堂在线| 建设人人有责人人尽责人人享有的 | 日本猛色少妇xxxxx猛交久久| 免费人成在线观看视频色| 国产亚洲91精品色在线| 亚洲在线观看片| 自拍欧美九色日韩亚洲蝌蚪91 | 精品一区二区免费观看| 在线亚洲精品国产二区图片欧美 | 精华霜和精华液先用哪个| 国产精品伦人一区二区| 亚洲欧美日韩无卡精品| 嘟嘟电影网在线观看| 婷婷色综合大香蕉| 欧美日韩亚洲高清精品| 亚洲天堂av无毛| 亚洲精品中文字幕在线视频 | 亚洲色图av天堂| 欧美精品国产亚洲| 99久久人妻综合| 日本与韩国留学比较| 在现免费观看毛片| 国产男女超爽视频在线观看| 啦啦啦中文免费视频观看日本| 国产欧美亚洲国产| 在线观看免费高清a一片| 欧美最新免费一区二区三区| 丰满少妇做爰视频| 精品熟女少妇av免费看| av女优亚洲男人天堂| 男人舔奶头视频| 热re99久久精品国产66热6| 精华霜和精华液先用哪个| 国产爽快片一区二区三区|