Terminal building short-term passenger flow forecast based on two-tier K-nearest neighbor algorithm
-
摘要:
航站楼离港客流量在短时期内呈现准周期性规律变化,易受航班计划、天气等多种因素影响,表现出复杂的非线性特点。为了实现航站楼短时客流量的准确预测,在传统
K 近邻(KNN)算法基础上增加了航班计划状态模式匹配方法,以航班计划包含的多维属性作为特征选取相似历史运营日作为预测基准向量,建立基于航站楼短时客流量预测的双层K 近邻模型。通过实例分析,与ARIMA模型和传统K 近邻模型等进行比较,证明双层K 近邻模型预测误差更小,精度更高,模型拟合度相对传统K 近邻模型提高了8%~10%,为航站楼短时客流量精确预测提供了一种新的解决思路。Abstract:Outbound passenger flow of terminal building shows the quasi-periodic variation in a short period of time and also shows complex nonlinear characteristics because of many factors such as flight schedule and weather. In order to accurately predict the short-term passenger flow of terminal building, the flight schedule state pattern matching procedure is added on the basis of the traditional
K -nearest neighbor (KNN) algorithm. The flight schedule including multi-dimensional attributes is taken as a feature to select historical similar operation days as forecast reference vectors. The two-tierK -nearest neighbor model based on terminal building short-term passenger flow forecast is built. Through instance analysis and comparison with ARIMA model and traditionalK -nearest neighbor model, it is proved that two-tierK -nearest neighbor model has smaller prediction error and higher precision, and the model fitting degree increases by 8%-10% compared with traditionalK -nearest neighbor model. Thus the model provides a new solution for accurately forecasting terminal building short-term passenger flow. -
表 1 K近邻模型预测精度分析
Table 1. KNN model prediction accuracy analysis
日期 MSE MAE R2/% 2016-09-09 351.893 0 11.451 1 83.65 2016-09-10 386.675 2 12.254 9 82.33 2016-09-11 345.365 1 10.931 5 89.31 2016-09-12 342.478 3 10.547 2 89.54 2016-09-13 411.579 2 13.367 3 79.14 表 2 不同模型预测精度分析
Table 2. Different models prediction accuracy analysis
日期 模型 MSE MAE R2/% 2016-09-09 ARIMA 393.735 7 13.195 3 80.27 KNN 351.893 0 11.451 1 83.65 TD-SFAPM 411.358 6 13.258 9 79.11 SVM 343.256 8 12.158 9 83.35 T-KNN 273.253 5 10.332 5 90.31 2016-09-10 ARIMA 423.658 1 14.652 8 78.13 KNN 386.675 2 12.254 9 82.33 TD-SFAPM 422.598 7 14.857 0 77.28 SVM 379.876 3 12.268 9 83.22 T-KNN 289.326 5 10.659 9 90.21 2016-09-11 ARIMA 387.365 7 13.986 3 81.55 KNN 345.365 1 10.931 5 89.31 TD-SFAPM 404.586 13.896 7 79.58 SVM 385.897 11.857 0 81.80 T-KNN 271.325 9 9.587 9 91.13 2016-09-12 ARIMA 435.578 9 14.587 3 77.97 KNN 342.478 3 10.547 2 89.54 TD-SFAPM 412.583 0 14.058 0 78.20 SVM 378.368 7 11.235 8 82.58 T-KNN 286.687 2 10.253 1 90.63 2016-09-13 ARIMA 426.875 3 13.087 5 78.96 KNN 411.579 2 13.367 3 79.14 TD-SFAPM 385.350 13.589 7 80.25 SVM 365.257 11.587 0 82.58 T-KNN 268.657 8 9.324 6 91.35 表 3 不同模型时间维度预测精度分析
Table 3. Time dimension prediction accuracy analysis of different models
日期 模型 MSE MAE R2/% 2016-04-05 ARIMA 372.354 6 13.257 9 79.32 KNN 331.389 6 12.132 4 82.65 TD-SFAPM 393.251 14.258 9 77.52 SVM 345.367 4 13.235 7 81.25 T-KNN 252.178 6 10.258 9 90.71 2016-05-20 ARIMA 365.578 9 13.189 6 80.56 KNN 342.236 5 11.438 1 81.36 TD-SFAPM 362.576 8 13.025 7 81.03 SVM 332.216 0 11.268 7 82.03 T-KNN 265.796 3 10.568 6 89.62 2016-07-15 ARIMA 363.589 7 12.328 6 81.36 KNN 342.358 6 11.327 3 83.22 TD-SFAPM 421.354 0 14.258 9 78.70 SVM 378.235 5 12.963 0 80.25 T-KNN 275.265 8 10.981 1 90.26 2016-08-08 ARIMA 378.998 5 14.265 7 78.25 KNN 353.865 7 11.188 2 83.55 TD-SFAPM 423.587 9 15.025 7 77.25 SVM 373.568 0 12.524 0 82.56 T-KNN 266.788 4 9.712 3 91.68 2016-10-01 ARIMA 355.562 3 11.589 6 80.33 KNN 324.337 8 11.045 1 82.44 TD-SFAPM 380.257 9 13.257 0 79.33 SVM 352.248 7 11.568 7 81.57 T-KNN 258.365 7 9.865 2 91.70 -
[1] GROSCHE T, ROTHLAUF F, HEINZL A.Gravity models for airline passenger volume estimation[J].Journal of Air Transport Management, 2007, 13(4):175-183. doi: 10.1016/j.jairtraman.2007.02.001 [2] LETAVKOVA D, MATUSKOVA S, KEBO V, et al.Simulation model for regional airport passenger throughputs[C]//International Carpathian Control Conference.Piscataway, NJ: IEEE Press, 2015: 295-299. [3] 黄飞虎, 彭舰, 由明阳.航空旅客群体移动行为特性分析[J].物理学报, 2016, 65(22):2289011. http://d.old.wanfangdata.com.cn/Periodical/wlxb201622032HUANG F H, PENG J, YOU M Y.Analyses of characetristics of air passenger group mobility behaviors[J].Acta Physica Sinica, 2016, 65(22):2289011(in Chinese). http://d.old.wanfangdata.com.cn/Periodical/wlxb201622032 [4] NDOH N N, ASHFORD N.Evaluation of airport access level of service[J].Transportation Research Record, 1993, 1423:34-39. http://cn.bing.com/academic/profile?id=6bd9f25c65e3cda2ca2952dccc6b3dc2&encoded=0&v=paper_preview&mkt=zh-cn [5] KIM W, PARK Y, KIM B J.Estimating hourly variations in pa-ssenger volume at airports using dwelling time distributions[J].Journal of Air Transport Management, 2004, 10(6):395-400. doi: 10.1016/j.jairtraman.2004.06.009 [6] 邢志伟, 文涛, 罗谦, 等.基于效用价值驱动的旅客出行动力学研究与建模[J].北京航空航天大学学报, 2018, 44(2):250-256. http://bhxb.buaa.edu.cn/CN/abstract/abstract14420.shtmlXING Z W, WEN T, LUO Q, et al.Utility value driven passenger travel dynamic study and modeling[J].Journal of Beijing University of Aeronautics and Astronautics, 2018, 44(2):250-256(in Chinese). http://bhxb.buaa.edu.cn/CN/abstract/abstract14420.shtml [7] 邢志伟, 冯文星, 罗谦, 等.基于航班离港时刻主导的单航班离港旅客聚集模型[J].电子科技大学学报, 2015, 44(5):719-724. doi: 10.3969/j.issn.1001-0548.2015.05.014XING Z W, FENG W X, LUO Q, et al.Arrived passenger model in single flight based on the time of departure[J].Journal of University of Electronic Science and Technology of China, 2015, 44(5):719-724(in Chinese). doi: 10.3969/j.issn.1001-0548.2015.05.014 [8] TIAN Z D, LI S J, WANG Y H, et al.A prediction method based on wavelet transform and multiple models fusion for chaotic time series[J].Chaos, Solitons & Fractals, 2017, 98:158-172. http://cn.bing.com/academic/profile?id=0f5af4d0139fc358a3fec5a579396ac5&encoded=0&v=paper_preview&mkt=zh-cn [9] 田中大, 李树江, 王艳红, 等.基于ARIMA与ESN的短期风速混合预测模型[J].太阳能学报, 2016, 37(6):1603-1610. doi: 10.3969/j.issn.0254-0096.2016.06.037TIAN Z D, LI S J, WANG Y H, et al.Short-term wind speed hybrid prediction model based on ARIMA and ESN[J].Acta Energiae Solaris Sinica, 2016, 37(6):1603-1610(in Chinese). doi: 10.3969/j.issn.0254-0096.2016.06.037 [10] TIAN Z D, LI S J.A network traffic prediction method based on IFS algorithm optimised LSSVM[J].International Journal of Engineering Systems Modelling and Simulation, 2017, 19(4):200-213. http://cn.bing.com/academic/profile?id=a8e2b61a863fc22c57c97211d5044ab2&encoded=0&v=paper_preview&mkt=zh-cn [11] 田中大, 李树江, 王艳红, 等.基于混沌理论与改进回声状态网络的网络流量多步预测[J].通信学报, 2016, 37(3):55-70. doi: 10.3969/j.issn.1001-2400.2016.03.010TIAN Z D, LI S J, WANG Y H, et al.Network traffic multi-step prediction based on chaos theory and improved echo state network[J].Journal on Communications, 2016, 37(3):55-70(in Chinese). doi: 10.3969/j.issn.1001-2400.2016.03.010 [12] 田中大, 李树江, 王艳红, 等.高斯过程回归补偿ARIMA的网络流量预测[J].北京邮电大学学报, 2017, 40(6):65-73. http://d.old.wanfangdata.com.cn/Periodical/bjyddx201706010TIAN D Z, LI S J, WANG Y H, et al.Network traffic prediction based on ARIMA with Gaussian process regression compensation[J].Journal of Beijing University of Posts and Telecommunications, 2017, 40(6):65-73(in Chinese). http://d.old.wanfangdata.com.cn/Periodical/bjyddx201706010 [13] DAVIS G, NIHAN N.Nonparametric regression and short-term freeway traffic forecasting[J].Journal of Transportation Engineering, 1991, 117(2):178-188. doi: 10.1061/(ASCE)0733-947X(1991)117:2(178) [14] 林川.基于K近邻非参数回归的短时交通流预测算法研究[D].成都: 电子科技大学, 2015: 30-40. http://cdmd.cnki.com.cn/Article/CDMD-10614-1015711959.htmLIN C.Short-term traffic flow prediction algorithm based on K-nearest neighbor[D].Chengdu: University of Electronic Science and Technology of China, 2015: 30-40(in Chinese). http://cdmd.cnki.com.cn/Article/CDMD-10614-1015711959.htm [15] 张涛, 陈先, 谢美萍, 等.基于K近邻非参数回归的短时交通流预测方法[J].系统工程理论与实践, 2010, 30(2):376-384. http://d.old.wanfangdata.com.cn/Periodical/xtgcllysj201002025ZHANG T, CHEN X, XIE M P, et al.K-NN based nonparametric regression method for short-term traffic flow forecasting[J].Systems Engineering-Theory and Practice, 2010, 30(2):376-384(in Chinese). http://d.old.wanfangdata.com.cn/Periodical/xtgcllysj201002025 [16] 于滨, 邬珊华, 王明华, 等.K近邻短时交通流预测模型[J].交通运输工程学报, 2012, 12(2):105-111. doi: 10.3969/j.issn.1671-1637.2012.02.017YU B, WU S H, WANG M H, et al.K-nearest neighbor model of short-term traffic flow forecast[J].Journal of Traffic and Transportation Engineering, 2012, 12(2):105-111(in Chinese). doi: 10.3969/j.issn.1671-1637.2012.02.017 [17] 谢海红, 戴许昊, 齐远, 等.短时交通流预测的改进K近邻算法[J].交通运输工程学报, 2014, 14(3):87-94. doi: 10.3969/j.issn.1671-1637.2014.03.017XIE H H, DAI X H, QI Y, et al.Improved K-nearest neighbor algorithm for short-term traffic forecasting[J].Journal of Traffic and Transportation Engineering, 2014, 14(3):87-94(in Chinese). doi: 10.3969/j.issn.1671-1637.2014.03.017 [18] 豆飞, 贾利民, 秦勇, 等.铁路客运专线模糊k近邻客流预测模型[J].中南大学学报(自然科学版), 2014, 45(12):4422-4430. http://www.cnki.com.cn/article/cjfdtotal-zngd201412044.htmDOU F, JIA L M, QIN Y, et al.Fuzzy k-nearest neighbor passenger flow forecasting model of passenger dedicated line[J].Journal of Central South University(Science and Technology), 2014, 45(12):4422-4430(in Chinese). http://www.cnki.com.cn/article/cjfdtotal-zngd201412044.htm [19] 张晓利, 贺国光, 陆化普.基于k-邻域非参数回归短时交通流预测方法[J].系统工程学报, 2009, 24(2):178-183. http://d.old.wanfangdata.com.cn/Periodical/xtgcxb200902008ZHANG X L, HE G G, LU H P.Short-term traffic flow forecasting based on k-nearest neighbors nonparametric regression[J].Journal of Systems Engineering, 2009, 24(2):178-183(in Chinese). http://d.old.wanfangdata.com.cn/Periodical/xtgcxb200902008 [20] CAI P L, WANG Y P, LU G Q, et al.A spatiotemporal correlative k-nearest neighbor model for short-term traffic multistep forecasting[J].Transportation Research Part C:Emerging Technologies, 2016, 62:21-34. doi: 10.1016/j.trc.2015.11.002 [21] 陈通, 孙国强, 卫志农, 等.基于相似日和CAPSO-SNN的光伏发电功率预测[J].电力自动化设备, 2017, 37(3):66-71. http://d.old.wanfangdata.com.cn/Periodical/dlzdhsb201703012CHEN T, SUN G Q, WEI Z N, et al.Photovoltaic power generation forecasting based on similar day and CAPSO-SNN[J].Electric Power Automation Equipment, 2017, 37(3):66-71(in Chinese). http://d.old.wanfangdata.com.cn/Periodical/dlzdhsb201703012 [22] WANG Y S, WU D L, GUO C X, et al.Short-term wind speed prediction using support vector regression[C]//IEEE Power and Energy Society General Meeting.Piscataway, NJ: IEEE Press, 2010: 1-6.