基于轨迹数据和深度学习的CNG出租车CO2排放微观模型构建及碳减排效益评估方法
Construction of a Micro Model for CO2 Emissions from CNG Taxi Based on Trajectory Data and Deep Learning Method and Evaluation of Carbon Reduction Benefits
通讯作者:
收稿日期: 2023-05-31 修回日期: 2023-09-2
基金资助: |
|
Corresponding authors:
Received: 2023-05-31 Revised: 2023-09-2
Fund supported: |
|
作者简介 About authors
刘琪(1994—),男,河北邯郸人,博士,主要从事城市绿色交通研究。E-mail:
为准确评价压缩天然气(CNG)出租车的二氧化碳(CO2)减排效益,以武汉市为例,提出了一种基于深度学习的车辆微观CO2排放模型来准确对城市内出租车的CO2排放做时空分析,探究出租车在不同燃料情景下CO2排放时空规律。路测实验中使用便携式排放测量系统(PEMS)收集车辆的CO2排放数据,考虑车辆驾驶特征序列和燃料类型,借助BiLSTM算法构建了车辆微观CO2排放模型,并验证其精度;利用提出的CO2排放模型和武汉市15 752辆出租车轨迹数据估算了武汉市出租车使用92#汽油和CNG的CO2排放,探索CNG出租车的CO2减排效益。结果表明,模型精度优于目前常用SVR、LSTM等回归算法和IVE、CMEM等物理模型,能够拟合真实车辆的CO2排放变化,满足大范围估算城市出租车CO2排放的精度需求,为车辆排放估算提供更好思路;实证结果发现,一天内,15 752辆武汉市出租车全面使用CNG取代92#汽油可以减少22.05%的CO2排放,同时揭示了CNG出租车在时间空间角度的CO2排放规律以及CO2减排效益。结果对政府交通部门推广车辆使用CNG燃料提供依据。
关键词:
Many large cities have been actively promoting the policy of "replacing oil with gas" for taxis. Taxis are converted from traditional gasoline consumption to Compressed Natural Gas (CNG) to achieve energy conservation and emission reduction goals. To accurately evaluate the carbon dioxide (CO2) emission reduction benefits of CNG taxis, taking Wuhan as an example, a vehicle microscopic CO2 emission model based on deep learning method and trajectory data was proposed to investigate the spatial-temporal characteristics of CO2 emissions of taxis under different fuel scenarios. Considering the driving feature sequence and fuel type of vehicles, the Portable Emission Measurement System (PEMS) was used to collect vehicle CO2 emission data in the road test experiment, then we constructed a vehicle microscopic CO2 emission model by the BiLSTM algorithm and further verified its accuracy. Based on the proposed CO2 emission model and the trajectory data of 15 752 Wuhan taxis, the CO2 emissions throughout the entire lifecycle of urban taxis by 92# gasoline and CNG were estimated respectively to quantify the CO2 emission reduction benefits of CNG taxis. The results show that the proposed model had a higher accuracy than common regression algorithms such as SVR and LSTM, and the predictions matched well with real vehicle CO2 emission changes, meeting the accuracy for a large-scale estimation of urban taxi CO2 emissions. In addition, the accuracy of taxi CO2 emission estimation based on deep learning methods was also higher than that of physical microscopic models such as IVE and CMEM. Especially, when using CNG as vehicle fuel, the physical models had significant computational errors due to not involving technical parameters. The empirical results show that, taxi CO2 emissions using CNG were reduced by 22.05% during the PTW process and by 49.45% during the WTP process, compared to emissions using 92 # gasoline. Our results reveal both the temporal and spatial patterns of taxi CO2 emission as well as the CO2 emission reduction benefits of CNG taxis. The outperformance of deep learning methods over other methods for estimating vehicle CO2 emissions provides new ideas for large-scale and high-precision estimation of vehicle emissions. The CO2 emission reduction benefits of using CNG as fuel in taxis are significant, which provides a reference for the government to formulate relevant energy-saving and CO2 emission reduction policies.
Keywords:
本文引用格式
刘琪, 陈碧宇, 李歆艺.
LIU Qi, CHEN Biyu, LI Xinyi.
1 引言
随着我国汽车保有量的快速提升,交通行业二氧化碳(Carbon Dioxide, CO2)排放量急剧增长。据统计,汽车等交通运输部门的CO2排放约占全球总CO2排放量的四分之一[1-2],是造成温室效应的主要原因之一。降低道路交通CO2排放已经列入到我国碳达峰、碳中和战略规划中,为此我国为节能减排做了一系列努力[3],例如很多大城市积极推行出租车“油改气”政策:鼓励出租车由传统的消耗汽油改装成使用压缩天然气(Compressed Natural Gas, CNG),以实现节能减排的目标[4-5]。出租车作为城市交通重要组成部分,对其实现全方位的节能减排可为城市绿色交通的发展做出重要贡献[6-7]。研究高精度的车辆CO2排放模型对于定量评价城市节能减排政策的节能环保效益具有重要意义。
国内外学者针对车辆CO2排放估算研究开展了很多工作,传统的CO2排放模型主要分为宏观模型和微观模型,宏观模型如MOVES(Motor Vehicle Emission Simulator)模型[8]、COPERT (Computer Program To Calculate Emissions From Road Transport)模型[9]等,主要以城市路网各路段的平均速度为主要特征估算大尺度下的CO2排放,忽略了现实中车辆驾驶的实际工况和不同路段路况的差异而导致精度不够高。微观模型考虑了更精细的车辆参数包括车型、质量、发动机参数、速度、加速度等,能够更精准地估计车辆的排放,早期一些学者利用底盘测功机的方法在实验室环境下测试车辆的CO2排放[10-11],但是没有考虑车辆的实际驾驶工况。一些学者利用便携式排放测量系统(Portable Emissions Measurement Systems, PEMS)通过实际路测实验[12-13],在车辆上安装PEMS设备,实时收集车辆的排放。使用PEMS尽管考虑了实际驾驶工况,但是由于设备昂贵且安装复杂,大范围地使用PEMS测算城市中出租车的CO2排放难以实现。
定位技术的高速发展以及GPS采样技术的日渐成熟,使得收集出租车的轨迹数据变得容易,出租车 GPS 数据包含了车辆在行驶过程中详细而丰富的运动状态信息[14⇓-16],为大范围估算城市内出租车的CO2排放提供了契机。一些学者结合物理微观模型在车辆排放估算方面做了很多工作[16],例如赵永明等[17]利用排放物理模型估算车辆排放,构建了车辆出行排放知识图谱,表征车辆行为与排放的关联关系。Liu等[18]借助出租车GPS轨迹数据和车牌识别数据,利用物理模型估算车辆的CO2排放,以杭州为例揭示了排放热点。Kan等[19]基于出租车GPS数据提取了车辆的活动轨迹,使用CMEM (Comprehensive Modal Emissions Model)模型估算车辆在不同运动状态下的CO2排放,以武汉市为例,探索城市排放的时空分布规律。Chang等[20]以北京市出租车轨迹数据为例,计算了出租车的工况信息,利用IVE (International Vehicle Emissions Model)模型估算了城市内出租车的CO2排放,并通过时空统计分析,探索北京市出租车的CO2排放分布规律。然而这些物理微观模型大多都是发达国家提出,尚不清楚其CO2排放估算精度是否满足中国的出租车,尤其是这些物理微观模型很少涉及CNG车辆的技术参数。
综上所述,针对CNG出租车CO2排放估算微观模型的不足,本文做了3点工作:① 通过实际路测实验,采集车辆驾驶轨迹数据和CO2排放数据,借助深度学习模型构建数据驱动下的CNG出租车排放微观模型,以实现对出租车在不同燃料情景下CO2排放的精确计算,为碳排放估算提供新思路;② 验证了传统微观物理模型如CMEM和IVE模型估算CNG出租车CO2排放的不足,提出了改进方向;③ 利用提出的微观模型对武汉市的出租车分别计算其不同油品下的CO2排放,评价出租车“油改气”背景下的CO2减排效益和时空规律。
2 研究方法
本文的技术路线如图1所示,首先设计了路测实验采集车辆的轨迹数据和CO2排放数据,对其数据预处理并实现地图匹配,接着利用滑动时间窗口提取车辆驾驶特征序列与CO2排放数据形成映射关系,借助BiLSTM深度学习方法构建车辆微观CO2排放模型并评价验证其精度,最后使用城市浮动车数据和提出的模型计算和评价CNG出租车的CO2减排效益。
图1
2.1 基于深度学习的车辆CO2排放微观模型
2.1.1 车辆驾驶特征序列提取
准确估算CO2排放量是评价CNG出租车减排效益的前提,车辆的驾驶条件影响CO2排放水平,车辆驾驶的运动过程具有时间序列属性,碳排放不仅与当前的瞬时驾驶条件相关,还与过去的驾驶条件序列有关。为更精准地估算碳排放,本文考虑排放与车辆驾驶条件的时空相关性,借助深度学习方法建立车辆驾驶特征序列与碳排放的映射关系,以此实现车辆排放的估算。
车辆在
式中:
为了捕捉车辆驾驶的时序演化特征,使用滑动时间窗口提取驾驶特征序列
式中:m是滑动时间窗口的长度。
2.1.2 基于BiLSTM的车辆CO2排放模型建立
为了更好地捕捉车辆CO2排放与驾驶特征序列的非线性关系,利用双向长短期记忆网络(Bidirectional Long Short-Term Memory,BiLSTM)[30-31]学习车辆排放的时序演化机制,构建CO2排放微观模型。LSTM网络常用于解决时间序列建模和分类任务,例如在车辆轨迹分析[32⇓-34]、车辆排放估算等方面有许多应用。BiLSTM 网络由一个前向LSTM 和一个反向LSTM 组成,能够解决长时间依赖的序列建模问题,车辆CO2排放与车辆的驾驶条件时间序列相关,用BiLSTM能更好地捕捉2者的相关性,提高模型性能。用滑动时间窗口提取到车辆驾驶特征序列
LSTM单元利用3个门限控制结构(输入门、输出门以及遗忘门)增加神经网络的时序记忆能力,模型基于LSTM对车辆驾驶特征序列进行建模。图2展示了其中第
图2
式中:
式中:
式中:
设计的损失函数利用平均绝对误差的方法,表示为:
式中:
算法1:基于BiLSTM的车辆CO2排放估算 |
---|
Input:车辆驾驶特征 |
Output:CO2排放估算模型M |
begin |
1.计算和提取车辆的驾驶特征序列 |
2. 生成训练集和测试集 Dtrain 和Dtest 从数据集 |
3. For mini batch d in Dtrain |
4. For T time window data dT in d |
5. for i=1 to T: |
6. Input Xt into forward LSTM to encoder |
7. Using the backword LSTM to encoder |
8. End For |
9. Compute |
10.End For |
11.训练模型M 用反向LSTM算法 |
12. End For |
13.存储车辆CO2排放估算模型为 M |
2.2 碳减排效益评估方法
燃料产生CO2排放的生命周期过程包括2个阶段,一个是能源投入产出过程,包括原料开采、加工和运输过程,称为WTP(Well-to-Pump)阶段;另一个是车辆驾驶过程中的燃料燃烧过程,称为PTW(Pump-to-Wheels)阶段。为充分分析CNG出租车的CO2减排效益,使用出租车的轨迹数据估算不同燃料2个阶段的CO2排放。
在PTW阶段,借助本文提出的CO2排放模型,来评估城市CNG出租车在车辆驾驶过程中的CO2减排效益,基于轨迹数据,在相同的车辆驾驶条件下,改变出租车的燃料类型参数,利用提出的CO2排放估算模型计算这些出租车在2种燃料情景下的CO2排放,进而计算CNG充当燃料时的CO2减排效益,并从时空角度分别展开分析。空间上,分析不同道路类型下,CNG与92#油的CO2排放统计结果,以及在一天内,出租车在不同燃料类型下CO2排放的空间分布规律。时间上,分析出租车在一天 24 h、一周的CNG相对92#油CO2减排程度。
CNG和汽油在WTP阶段的排放率不同,GREEN(The Greenhouse gases, Regulated Emissions, and Energy use in Transportation Model)用于车辆和燃料技术生命周期评价的模型,经前人工作验证[35], CNG和汽油燃料在WTP阶段的CO2排放率分别为
3 研究区数据及模型训练
3.1 研究区概况和数据来源
本文研究区域为武汉市,自2004年起,武汉市政府启动CNG出租车政策,推动出租车以CNG替代汽油。截至2018年底,武汉市机动车保有量约300万辆,其中出租车16 747辆, 94%的出租车使用CNG燃料,其他1 000余辆出租车为电动汽车。武汉市最新的统计数据为2022年公布的出租车数量有17 778辆,新增1 000余辆电动出租车,2018—2022年CNG出租车的数量保持稳定。
数据集分为2个部分。第一部分数据集用于CO2排放模型的建立和评价,是在武汉市经过路测试验采集的出租车GPS轨迹数据、CO2排放数据和路网信息。武汉市出租车型号为东风雪铁龙爱丽舍,重量为1 125 kg,可以使用汽油和CNG作为燃料。利用高频GPS接收器采集到逐秒的车辆位置、高程、速度和加速度信息,并在出租车排气管配备了便携式车载排放检测系统(PEMS)收集CO2排放率(g/m3)和尾气流速(m3/s),计算可得到逐秒CO2排放值cet。使用相关设备在武汉市实际路网中分别采集使用92#汽油和CNG燃料的车辆驾驶工况和CO2排放量,一共采集到2023年6月的共2周内580个行程的数据,涵盖了武汉市不同路况的详细信息。 2种燃料下共采集约120万个轨迹点。第二部分数据用于评价武汉市CNG出租车的CO2减排效益,考虑到2020—2022年正值新冠疫情时期,此段时间与非疫情时期交通出行量差异大,本文使用疫情前的2018年8月2日至8月8日共15 752辆CNG双燃料出租车轨迹信息,此部分数据集由武汉市某匿名出租车运营公司提供,数据的GPS采样频率为30 s一次,数据集中的武汉市路网数据来源于OpenStreetMap网站,包括19 354个节点和26 437条路段来提取道路类型信息。由于出租车车载GPS精度不够高,采样频率有限,为保证CO2排放计算的准确性,首先要对轨迹数据进行预处理,包括轨迹插值和地图匹配,结果如图3所示,插值后的轨迹更加完整,采样频率更高。
图3
3.2 模型训练和评价
上述的第一部分路测数据集用于模型的构建与评价,利用十折交叉验证方法将数据集分成十份,轮流将其中9份作为模型的训练数据,1份作为验证模型精度的测试数据。模型的滑动时间窗口步长为1 s,如图4(a)所示,滑动时间窗口大小影响模型的效果,当设置
图4
图4
不同参数下的实验结果对比
Fig. 4
Comparison of experimental results under different parameters
式中:
4 结果及分析
4.1 模型评价结果
本文用MAE和RMSE 2个指标来评价不同算法的表现,表1展示了3种机器学习算法和2种物理模型在逐秒级的估算精度,可以看出SVR因其不能捕捉时间序列特征,效果相对LSTM和BiLSTM较差,当使用92#油时, SVR的MAE为2.29, BiLSTM将其估算的MAE降低了51.96%,达到1.10,当使用CNG时, SVR的MAE为3.17, BiLSTM将其估算的MAE降低了60.88%,达到1.24。且BiLSTM的结果也均高于LSTM。基于BiLSTM的CO2估算模型由于其更好地捕捉车辆驾驶过程的时间序列特征而显著提高了估算精度,满足CNG出租车减排效益评价的精度需求。
表1 3种算法的CO2排放估算性能评价结果
Tab. 1
模型 | 92#汽油 | CNG | |||
---|---|---|---|---|---|
RMSE | MAE | RMSE | MAE | ||
SVR | 4.31 | 2.29 | 5.29 | 3.17 | |
LSTM | 2.22 | 1.21 | 2.53 | 1.38 | |
BiLSTM | 1.98 | 1.10 | 2.45 | 1.24 | |
IVE | 3.62 | 1.38 | 6.25 | 4.31 | |
CMEM | 3.24 | 1.29 | 5.95 | 3.74 |
注:加粗数值表示5种模型评价结果中的最优值。
IVE和CMEM 2种物理模型在92#油情景下的估算结果优于SVR模型,与LSTM和BiLSTM相比,精度略低;但是在CNG作为燃料时, 2种物理模型的精度较差,均低于3种机器学习。可见,物理微观模型由于没有考虑车辆驾驶的时序演化特征,效果不及深度学习方法;另外,由于缺乏CNG出租车的详细技术设置, 2种物理模型在CNG情景下的精度较差。后续2种物理模型可以进一步考虑车辆的时间序列特征而非仅考虑局部空间属性,可以捕捉到更详细的车辆驾驶规律,从而提升模型的精度;另外,CMEM和IVE模型需要分析和更新更多技术标准,例如改装的CNG出租车的油品和车型,以完善模型的泛化能力。
为了更详细展示提出的算法估算CO2排放过程,分别随机选择92#汽油和CNG燃料下的1段车辆路测轨迹,用不同算法逐秒的估算值和真值比较结果如图5所示,可以看出,虽LSTM和BiLSTM的估算结果和真值的趋势相近,优于不能捕捉时间序列特征的SVR算法,LSTM算法相较于BiLSTM,会出现高估或者低估的情况。详细的估算过程展示了本文提出的基于BiLSTM排放估算模型能够很好地拟合真实CO2排放变化。
图5
图5
2种油品下各方法估算CO2排放过程
Fig. 5
Estimating CO2 emission processes using various methods under two fuel types
4.2 CNG出租车CO2减排效益时空分析
图6展示了2018年8月3日基于15 752辆出租车轨迹估算得到的不同燃料情景下车辆驾驶过程的CO2排放结果。这一天内,每个轨迹点的CO2排放值被映射到武汉市路网地图中,CO2排放量以不同的颜色分层显示。可以看出,相比于使用汽油,使用CNG时,地图中CO2排放的红色区域明显减少。经统计,CNG情景下出租车在这一天内的总CO2排放量有358.76 t,比92#油情景下的460.26 t降低了22.05%。另外,由于武汉市沿江的主要商业圈道路相对拥堵,车辆运行速度慢,导致CO2排放高,使用CNG时,中心区域的排放热点也会有所减少。
图6
图6
2018年8月3日出租车使用不同燃料类型时CO2排放的空间分布可视化
Fig. 6
Spatial distribution visualization of CO2 emissions when taxis use different fuel types on August 3, 2018
如表2所示,不同类型的道路车流量和规定的车速不同,为从空间角度评价不同道路类型CNG出租车的CO2减排效益。本文将武汉市路网的道路类型分为5个等级,从空间上分析2种油品在不同道路类型上的CO2排放特征。
表2 武汉市路网道路类型信息
Tab. 2
道路等级 | 道路类型 | 限速/(km/h) |
---|---|---|
1 | 支路 | 30 |
2 | 次干道 | 40 |
3 | 主干道 | 60 |
4 | 快速路 | 80 |
5 | 高速路 | 120 |
通过计算不同类型道路的每个路段排放率来对比分析,可表示为:
式中:
图7展示了92#油和CNG作为燃料时,不同道路类型在这一天内的平均CO2排放率(g/km),可以看出道路等级越低,排放率更高,速度更快的快速路和高速路上出租车的CO2排放率显著较低。低等级的道路在城市中数量占比大,车辆运行数量大,总排放相较于高等级道路更高,这也是排放率高的原因。另外,不同等级道路限速不同,低等级道路在上下班高峰期间拥堵,车辆运行速度慢会造成车辆排放更高,这也是排放率高的重要原因。在各个道路类型中,CNG作为燃料时,CO2排放率始终低于汽油。统计不同道路类型的车辆排放率是从空间角度分析车辆的排放特征,分析CNG在不同道路类型的减排率,有助于交管部门合理安排制定限速限流政策,进而从道路端节能减排。
图7
图7
2018年8月3日92#汽油和CNG燃料情景下不同道路类型的CO2排放率
Fig. 7
Average CO2 emissions of paths of different types under different fuel types on August 3, 2018
图8是武汉市出租车在2种燃料情景下在2018年8月3日CO2排放一天24 h的分布情况,从0:00开始,城市出租车的CO2排放逐渐减少,在凌晨3:00达到最低值;此后,随着道路车辆的增加,CO2排放逐步升高。在早高峰8:00左右、午高峰12:00、晚高峰17:00出租车的CO2排放分别达到峰值。使用CNG时,出租车的CO2排放在不同时间都要远低于使用汽油。
图8
图8
2018年8月3日武汉市出租车在不同油品下CO2排放单日24 h变化
Fig. 8
Hourly variations of CO2 emissions under different fuel types of taxis in Wuhan on August 3, 2018
图9(a)是PTW阶段,武汉市出租车在2种燃料情景下在2018年8月3日—8月9日CO2排放一周的分布情况,在这一周内,节假日周日和周六的CO2排放略高于工作日,这和周末市民出行量大有关。使用92#汽油时,出租车的CO2排放量在这一周内总量为3 325.83 t,使用CNG时,出租车的CO2排放量在这一周内总量为2 628.58 t,相较于汽油降低了20.96%的CO2排放。图9(b)是WTP阶段基于2018年8月3日—8月9日一周的武汉市出租车轨迹,根据GREET模型得到的2种油品在WTP阶段的CO2排放率计算得到的一周内2种燃料情景下CO2排放分布情况。假设全部使用92#汽油时,WTP阶段的CO2排放总量约为325.23 t,全部使用CNG时,WTP阶段的CO2排放总量约为164.39 t,比92#汽油减少160.84 t CO2排放,减排率达49.45%,在燃料的生产投入阶段,使用CNG也可显著降低CO2排放成本。
图9
图9
2018年8月武汉市出租车不同油品下CO2排放一周变化
Fig. 9
Changes in CO2emissions of taxis in Wuhan over a week in August 2018
5 结论
本文在数据驱动下提出了一种基于深度学习方法的车辆CO2排放估算模型,并以武汉市的CNG出租车为例,计算并评价武汉市CNG出租车的CO2减排效益。得出以下结论:
(1)本文以武汉市出租车为例,设计实际路测实验采集92#汽油和CNG燃料情景下的车辆轨迹和CO2排放量,考虑车辆排放的时序演化特性,借助BiLSTM算法构建微观CO2排放模型,模型可以捕捉车辆驾驶的时间序列特征,顾及不同油品实现了CNG和汽油燃料情景下CO2排放的精确估算。并与SVR和LSTM进行对比验证,基于BiLSTM的模型精度有明显提升,估算的CO2排放量与真实值更加吻合。
(2)基于实际路测实验验证了基于深度学习方法估算出租车CO2排放的精度优于物理微观模型IVE和CMEM,尤其是以CNG为车辆燃料时,物理模型由于未涉及其技术参数,导致计算误差较大。因此,使用物理模型时应考虑车辆驾驶的时间序列特征,弥补局部空间属性带来的不足,也需进一步调试补充CNG燃料出租车的技术参数,提高其精度。
(3)评价了武汉市CNG出租车的CO2减排效益, 2018—2022年武汉市CNG出租车数量保持稳定,为避免新冠疫情期间车辆出行量与非疫情期间差异大带来的影响,本文利用出租车微观CO2排放模型和非疫情期间2018年武汉市15 752辆出租车的运行轨迹数据,实现了PTW阶段出租车在汽油和CNG情景下的CO2排放的精确估算,并从燃料的WTP和PTW阶段分别对CNG出租车的CO2减排率进行了分析。实验结果表明,武汉市车辆的CO2排放分布不均匀,城市中心的排放更集中;不同道路等级的排放率也有所不同,低等级道路由于其车速低和车流量大等原因其排放率更高,等级越高(如高速公路)车流量越小且车速越快,其排放率越低。另外,武汉市15 752辆CNG出租车的CO2排放量为358.76 t/日,在汽油情景下的CO2排放量为460.26 t/日,全部出租车由CNG替代汽油可实现22.05%的CO2减排。在能源的WTP阶段,本文借助GREET模型,基于轨迹数据计算了一周内2种燃料下的CO2排放,一周内所有出租车全部使用CNG时,WTP阶段的CO2排放总量约为164.39 t,比92#汽油减少160.84 t CO2排放,在WTP阶段,CNG减排率达49.45%。出租车使用CNG作为能源,在两个阶段均有显著CO2减排效益。
综上所述,基于深度学习方法估算车辆的CO2排放可以实现低成本高精度的估算结果,符合大数据时代对轨迹数据挖掘车辆排放的要求,为大范围高精度估算车辆排放提供新思路。其次,实证结果可以看出,出租车使用压缩天然气作为燃料时的CO2减排效益十分可观,为政府制定相关节能减排政策提供依据。
此外,未来关于评价CNG出租车的减排效益还有几点可以深入探讨,比如本文只针对CO2这一排放气体进行了讨论,未涉及其他尾气如氮化物或硫化物等,借助深度学习同时估算多种尾气排放仍面临着较大的挑战;环境因素如温度和湿度等条件也会对发动机尾气排放产生影响,提出的模型未涉及环境条件,这也是未来继续改进的地方。另外,本文涉及的能源WTP阶段的排放估算过程较为粗略,后续需要更多调研,以实现精细化计算。未来将开展相关研究讨论多种类尾气排放的估算方法,以支持更全面评价CNG出租车的社会经济环保效益。
参考文献
A big data approach to improving the vehicle emission inventory in China
[J].Estimating truck emissions accurately would benefit atmospheric research and public health protection. Here, we developed a full-sample enumeration approach TrackATruck to bridge low-frequency but full-size vehicles driving big data to high-resolution emission inventories. Based on 19 billion trajectories, we show how big the emission difference could be using different approaches: 99% variation coefficients on regional total (including 31% emissions from non-local trucks), and ± as large as 15 times on individual counties. Even if total amounts are set the same, the emissions on primary cargo routes were underestimated in the former by a multiple of 2-10 using aggregated approaches. Time allocation proxies are generated, indicating the importance of day-to-day estimation because the variation reached 26-fold. Low emission zone policy reduced emissions in the zone, but raised emissions in upwind areas in Beijing's case. Comprehensive measures should be considered, e.g. the demand-side optimization.
Global perspective on CO2emissions of electric vehicles
[J].Plug-in electric vehicles (PEVs) are a promising option for greenhouse gas (GHG) mitigation in the transport sector - especially when the fast decrease in carbon emissions from electricity provision is considered. The rapid uptake of renewable electricity generation worldwide implies an unprecedented change that affects the carbon content of electricity for battery production as well as charging and thus the GHG mitigation potential of PEV. However, most studies assume fixed carbon content of the electricity in the environmental assessment of PEV and the fast change of the generation mix has not been studied on a global scale yet. Furthermore, the inclusion of up-stream emissions remains an open policy problem. Here, we apply a reduced life cycle assessment approach including the well-to-wheel emissions of PEV and taking into account future changes in the electricity mix. We compare future global energy scenarios and combine them with PEV diffusion scenarios. Our results show that the remaining carbon budget is best used with a very early PEV market diffusion; waiting for cleaner PEV battery production cannot compensate for the lost carbon budget in combustion vehicle usage.
基于居民出行行为的北京市交通碳排放影响机理
[J].
The mechanism of CO2emissions from urban transport based on individuals' travel behavior in Beijing
[J].
实际道路工况下“油改气”双燃料出租车的排放特征
[J].
Emission characteristics of “oil-to-gas” dual fuel taxis under actual road conditions
[J].
出租车智能调度系统温室气体减排方法学研究
[J].
Research on methodology to reduce greenhouse gas emission by means of taxi intelligent dispatching system
[J].
基于MOVES模型本地化的轻型车排放因子估计方法
[J].
Localization of light-duty vehicle emission factor estimation based on MOVES
[J].
Detailed estimation of fuel consumption and emissions during aircraft taxi operations at Dallas/Fort Worth International Airport
[J].
Environmental implications of converting light gas vehicles: The Brazilian experience
[J].
Comparative engine performance and emission analysis of CNG and gasoline in a retrofitted car engine
[J].
On-road performance and emission characteristics of CNG-gasoline bi-fuel taxis/private cars at the roadside environment
[J].
On-road emission characteristics of CNG-fueled bi-fuel taxis
[J].
出租车GPS轨迹集聚和精细化路网提取
[J].
DOI:10.11947/j.AGCS.2019.20180256
[本文引用: 1]
针对利用GPS数据提取双向路网和交叉路口转向信息精度低的不足,本文提出一种顾及位置与行驶方向的轨迹集聚和精细化路网提取方法,实现了精细化路网提取。为提高交叉路口路网的提取精度,首先剔除原始轨迹中的离散和异常轨迹点,并按一定的步长对轨迹段进行加密;然后引入行驶方向角来表达车辆在轨迹点处的行驶方向,顾及位置和行驶方向获取每个轨迹点的相似轨迹点集合;接着依次计算每个轨迹点的偏移距离,通过对轨迹点的迭代偏移完成轨迹集聚;最后剔除未成功集聚的轨迹点,将完成集聚的轨迹点连成轨迹线并作缓冲区,运用栅格数据数字化方法提取得到能够反映道路精细转向关系的道路网。以福州市出租车GPS数据进行轨迹集聚和路网提取试验,结果表明:本文方法能有效地将GPS轨迹按车辆行驶方向分别进行集聚,提取的道路网为双向道路并且能反映交叉路口处道路的精细转向关系。
GPS trajectory agglomeration and refined road network extraction
[J].Aiming at the shortcomings of low-accuracy in the use of GPS data to extract bidirection a roads and intersections, this paper proposes a trajectory agglomeration and refined roads extraction method that takes into account the position and travel direction to extracts refined road network. First,we remove the discrete and abnormal trajectory points from the original trajectory and insert the trajectory points into the trajectory segments by a certain step size, in order to improve the extraction accuracy of the intersection network.Second,we introduce the driving direction angle to express the driving direction of the vehicle at the track point, obtain its similar trajectory points set by considering the position and direction of the track point, calculate the offset distance of each track point in turn, and complete the track aggregation by iteratively offsetting the track points.Finally, we eliminate the track points that have not been successfully gathered, and use the Grid digitization method to extract the road network that can reflect the fine steering relationship of the roads from the trajectory data after gather. The trajectory agglomeration and road network extraction experiments were carried out with GPS data of Fuzhou taxis. The experimental results show that this method can effectively gather the GPS trajectories according to the direction of vehicle travel and the extracted road network is bidirectional roads, and can finely reflect the steering relationship of the roads at the intersections.
A spatiotemporal data model and an index structure for computational time geography
[J].
交通仿真模型融合微观车辆排放模型研究综述
[J].
Review of studies that integrate traffic-simulation models with microscopic vehicle-emissions models
[J].
基于车辆身份检测数据的车辆出行排放知识图谱研究
[J].
Research on vehicle travel emission knowledge graph based on vehicle identification data
[J].
Spatial-temporal inference of urban traffic emissions based on taxi trajectories and multi-source urban data
[J].
Fine-grained analysis on fuel-consumption and emission from vehicles trace
[J].
Estimating real-time traffic carbon dioxide emissions based on intelligent transportation system technologies
[J].
基于LSTM神经网络的柴油机NOx排放预测
[J].
Prediction of diesel engine NOx emissions based on long-short term memory neural network
[J].
Measuring fuel consumption in vehicle routing: New estimation models using supervised learning
[J].
A review of recurrent neural networks: LSTM cells and network architectures
[J].Recurrent neural networks (RNNs) have been widely adopted in research areas concerned with sequential data, such as text, audio, and video. However, RNNs consisting of sigma cells or tanh cells are unable to learn the relevant information of input data when the input gap is large. By introducing gate functions into the cell structure, the long short-term memory (LSTM) could handle the problem of long-term dependencies well. Since its introduction, almost all the exciting results based on RNNs have been achieved by the LSTM. The LSTM has become the focus of deep learning. We review the LSTM cell and its variants to explore the learning capacity of the LSTM cell. Furthermore, the LSTM networks are divided into two broad categories: LSTM-dominated networks and integrated LSTM networks. In addition, their various applications are discussed. Finally, future research directions are presented for LSTM networks.
Use of artificial neural networks to predict fuel consumption on the basis of technical parameters of vehicles
[J].This publication presents a multi-faceted analysis of the fuel consumption of motor vehicles and the way human impacts the environment, with a particular emphasis on the passenger cars. The adopted research methodology is based on the use of artificial neural networks in order to create a predictive model on the basis of which fuel consumption of motor vehicles can be determined. A database containing 1750 records, being a set of information on vehicles manufactured in last decade, was used in the process of training the artificial neural networks. The MLP (Multi-Layer Perceptron) 22-10-3 network has been selected from the created neural networks, which was further subjected to an analysis. In order to determine if the predicted values match the real values, the linear Pearson correlation coefficient r and coefficient of determination R2 were used. For the MLP 22-10-3 neural network, the calculated coefficient r was within range 0.93–0.95, while the coefficient of determination R2 assumed a satisfactory value of more than 0.98. Furthermore, a sensitivity analysis of the predictive model was performed, determining the influence of each input variable on prediction accuracy. Then, a neural network with a reduced number of neurons in the input layer (MLP-20-10-3) was built, retaining a quantity of the hidden and output neurons and the activation functions of the individual layers. The MLP 20-10-3 neural network uses similar values of the r and R2 coefficients as the MLP 22-10-3 neural network. For the evaluation of both neural networks, the measures of the ex post prediction errors were used. Depending on the predicted variable, the MAPE errors for the validation sets reached satisfactory values in the range of 5–8% for MLP 22-10-3 and 6–10% for MLP 20-10-3 neural network, respectively. The prediction tool described is intended for the design of passenger cars equipped with internal combustion engines.
Data-driven estimation of energy consumption for electric bus under real-world driving conditions
[J].
Data-driven decomposition analysis and estimation of link-level electric vehicle energy consumption under real-world traffic conditions
[J].
基于BP神经网络的车辆碳排放测算研究
[J].
A research on vehicle carbon emission calculating method based on BP neural network
[J].
高速动态交通场景下自动驾驶车辆换道意图识别模型研究
[J].
Research on lane change intention recognition model of automated vehicle in high-speed dynamic traffic scenario
[J].
基于观测数据潜在特征与双向长短期记忆网络的车辆轨迹预测
[J].
Vehicle trajectory prediction based on potential features of observation data and bidirectional long short-term memory network
[J].
Interactive trajectory prediction of surrounding road users for autonomous driving using structural-LSTM network
[J].
The vehicle’s velocity prediction methods based on RNN and LSTM neural network
[C]//
Vehicle fuel consumption prediction method based on driving behavior data collected from smartphones
[J].
A microscopic model of vehicle CO2 emissions based on deep learning—a spatiotemporal analysis of taxicabs in Wuhan, China
[J].
Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance
[J].
/
〈 |
|
〉 |
