随着交通压力的逐步上升，利用好交通客流的历史数据，分析其内在规律，对未来时段的客流进行准确预测，不仅为交通运输管理的优化奠定了基础，同时也具有重要的理论意义。近年来，在地铁营运过程中自动售票检票系统（AFC）的普及，可以帮助我们获得充分的数据基础，而群智能优化算法和神经网络的组合模型在复杂数据的拟合上表现更好，因此本文结合黑猩猩优化算法和BP神经网络构成组合模型，来对地铁短时客流进行预测，并进一步利用杭州市地铁客流数据对该模型进行实例分析，评估其预测效果，具体内容如下：

首先，分析并处理地铁客流数据。基于所收集到的杭州市地铁客流数据，对原数据进行预处理，最后将连续日期的地铁AFC刷卡数据合并在一个文件中，做好模型预测前的数据准备工作。在数据分析阶段，主要包括数据中的异常值处理、空值填充并进一步分析数据的整体特征和分线路站点客流数据的特征。

其次，优化黑猩猩算法。深入学习群智能优化算法的基础理论，分析并改进黑猩猩优化算法。在初始化种群阶段，本文引入Circle混沌映射，得到分布更为均匀的初始种群；在迭代寻优过程中，引入Levy飞行来对寻优过程优化，并分析改进后的寻优是否优于原算法，最终构建得到引入Levy飞行的黑猩猩优化算法（Levy-Choa）。

再次，构建Levy-Choa-BP的组合预测模型。利用引入Levy飞行的黑猩猩优化算法（Levy-Choa）对BP神经网络的权重进行优化，寻找能使预测误差最小的最优权重值，并分析该组合模型的网络设计，得到最优节点数等参数。

最后，利用Levy-Choa-BP组合预测模型对地铁客流数据进行预测并评价预测效果。在评价预测效果时，共选用三个预测模型以及四个评价指标对模型进行评估，三个对比预测模型包括LSTM神经网络、原始BP神经网络以及未改进的黑猩猩优化算法（Choa）和BP神经网络结合构建的对比模型（Choa-BP），四个评价指标分别为拟合度（R²）、平均绝对百分比误差（MAPE）、平均绝对误差（MAE）、均方误差（MSE）。最终本文通过实例分析证实了所构建的Levy-Choa-BP组合模型的预测效果更好。

As traffic pressures gradually increase, utilizing historical traffic passenger data to analyze its inherent patterns and accurately predict passenger flow in future periods not only lays the foundation for the optimization of transportation management but also holds significant theoretical importance. In recent years, the widespread adoption of automatic ticketing and fare inspection systems (AFC) during the operation of subways has enabled the acquisition of a sufficient data basis. Furthermore, the combination model of swarm intelligence optimization algorithms and neural networks exhibits superior performance in fitting complex data. Consequently, this paper proposes a combined model incorporating the chimpanzee optimization algorithm and the BP neural network to predict short-term subway passenger flow. Additionally, the model is exemplarily analyzed using Hangzhou subway passenger data to evaluate its predictive effectiveness, the details of which are provided as follows:

Firstly, analyze and process urban rail transit passenger flow data. Utilizing the gathered Hangzhou subway passenger flow data, the raw information was subjected to analytical and contaminant removal processes, and finally the subway AFC card swiping data of consecutive dates were combined into a file to prepare the data before the model prediction. In the data analysis stage, it mainly includes outlier processing and null value filling of data types, and further analysis of the overall characteristics of the data and the characteristics of sub-line stations.

Secondly, optimize the chimpanzee optimization algorithm. By learning the basic theory of swarm intelligence optimization algorithm, analyzing the optimization process of chimpanzee optimization algorithm, this paper introduces circle chaos map to obtain a more uniformly distributed initial population, in the iterative process, Levy flight is introduced to optimize the optimization process, and analyzes whether the new solution is better than the iterative solution of the original algorithm, and finally constructs the chimpanzee optimization algorithm introduced by Levy flight (Levy-Choa).

Thirdly, construct a Levy-Choa-BP combined prediction model. The weight of the BP neural network was optimized by using the chimpanzee optimization algorithm introduced into Levy flight, and the optimal weight value that could minimize the prediction error was found, and the network design of the combined model was analyzed to obtain the optimal number of nodes and other parameters.

Finally, use the Levy-Choa-BP combined prediction model to forecast the subway passenger flow data and evaluate the prediction effectiveness. In assessing the forecasting performance, three predictive models and four evaluation indicators are employed to appraise the models’ performance. The three comparative prediction models include the LSTM neural network, the original BP neural network, and the unimproved Chimpanzee Optimization Algorithm (Choa) combined with the BP neural network to construct the comparative model (Choa-BP). The four evaluation indicators are fitness（R²）, Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and Mean Squared Error (MSE). Ultimately, this paper confirmes through case analysis that the constructed Levy-Choa-BP combined model has better predictive performance.

[1]2023年城市轨道交通运营数据统计分析[J]. 现代城市轨道交通, 2024, (03):131-132.

[2]姚俊峰, 何瑞, 史童童, 等. 基于机器学习的交通流预测方法综述[J]. 交通运输工程学报, 2023, 23(03):44-67.

[3]Mcculloch W.S, Walter P. A logical calculus of the ideas immanent in nervous activity[J]. Bulletin of Mathematical Biophysics, 1943, 5(4):115-133.

[4]Mitchell T.M, Does machine learning really work[J]. AI Magazine, 1997, 18(3):11.

[5]Rumelhart D.E, Hinton G.E, Willians R.J. Learning representations by back propagating errors[J]. Nature, 1986, 323(6088):533-536.

[6]Hinton G.E, Osindero S, Teh Y.W. A fast learning algorithm for deep belief nets[J]. Neural Computation, 2006, 18(07):1527-1554.

[7]陈龙, 高韩, 倪婕, 等. 鼻咽癌IMRT剂量验证中BP神经网络的应用[J]. 中华肿瘤防治杂志, 2022, 29(10):736-741.

[8]朱旭辉, 沈国娇, 夏平凡, 等. 基于螺旋进化萤火虫算法和BP神经网络的模型及其在PPP融资风险预测中的应用[J]. 计算机科学, 2022, 49(S1):667-674.

[9]谢从珍, 卢伟民, 马康, 等. 结合MIDAS仿真和BP神经网络的输电杆塔基础滑坡风险评估[J]. 科学技术与工程, 2023, 23(16):6923-6930.

[10]曾诚, 吴佳媛, 罗霞. 城市轨道交通短时客流预测文献综述[J]. 铁道运输与经济, 2021, 43(08):105-111.

[11]Smith B.L, Demetsky M.J. Multiple-interval freeway traffic flow forecasting[J]. Transportation Research Record, 1996, 1554(01):136-141.

[12]杨兆升, 王媛, 管青. 基于支持向量机方法的短时交通流量预测方法[J]. 吉林大学 学报(工学版), 2006, 36(06):881-884

[13]张利, 李星毅, 施化吉. 一种基于ARIMA模型的短时交通流量改进预测算法[C]//科学技术部全国智能运输系统协调指导小组办公室. 2007第三届中国智能交通年会论文集, 2007, 770-776.

[14]常国珍, 张前登. 基于时间序列的我国铁路客流量预测[J]. 统计与咨询, 2008(04):20-21.

[15]Ge S.Y, Zheng C.J, Hou M. Forecast of bus passenger traffic based on exponential smoothing and trend moving average method[J]. Applied Mechanics and Materials, 2013, 433:1374-1378.

[16]Zhao Z, Chen W, Yue H, et al. A novel short-term traffic forecast model based on travel distance estimation and ARIMA[C]. Chinese Control and Decision Conference, 2016:6270-6275.

[17]Ding C, Wang D, Ma X, et al. Predicting short-term subway ridership and prioritizing its influential factors using gradient boosting decision trees[J]. Sustainability, 2016, 8(11):1100.

[18]Yu S, Shang C, Yu Y, et al. Prediction of bus passenger trip flow based on artificial neural network[J]. Advances in Mechanical Engineering, 2016, 8(10):16878140.

[19]Liu S, Yao E. Holiday passenger flow forecasting based on the modified least square support vector machine for the metro system[J]. Journal of Transportation Engineering, 2017, 143(02):04016005.

[20]Pompigna A, Rupi F. Comparing practice-ready forecast models for weekly and monthly fluctuations of average daily traffic and enhancing accuracy by weighting methods[J]. Journal of Traffic and Transportation Engineering, 2018, 5(04):239-253.

[21]丁聪, 倪少权, 吕红霞. 基于梯度提升的城市轨道交通客流量预测分析[J]. 城市轨道交通研究, 2018, 21(09):60-63.

[22]Zhang J, Chen F, Guo Y, et al. Multi‐graph convolutional network for short‐term passenger flow forecasting in urban rail transit[J]. IET Intelligent Transport Systems, 2020, 14(10):1210-1217.

[23]王秋雯, 陈彦如, 刘媛春. 基于卷积长短时记忆神经网络的城市轨道交通短时客流预测[J]. 控制与决策, 2021, 36(11):2760-2770.

[24]Wu Z.L, Lu M, Wang C.Y. Forecasting metro rail transit passenger flow with multiple attention deep neural networks and surrounding vehicle detection devices[J]. Neural Networks and Complex Problem Solving Technologies, 2023, 53(15):18531-18546.

[25]远祯, 罗波. BP网络的改进研究[J]. 信息技术, 2006, 30(02):88-91.

[26]Bishop G. Neural networks for pattern recognition[M]. Oxford University Press, 1995.

[27]戚德虎, 康继昌. BP神经网络的设计[J]. 计算机工程与设计, 1998, 19(02):48-50.

[28]杨洋, 陈家俊. 基于群智能算法优化BP神经网络的应用研究综述[J]. 电脑知识与技术, 2020, 16(35):7-10.

[29]姚俊峰, 何瑞, 史童童, 等. 基于机器学习的交通流预测方法综述[J]. 交通运输工程学报, 2023, 23(03):44-67.

[30]Landahl H.D, McCulloch W.S, Pitts Walter. A statistical consequence of the logical calculus of nervous nets[J]. The Bulletin of Mathematical Biophysics, 1943, 5(4):135-137.

[31]孙洲鑫, 王文松. 深度学习中激活函数权重初始值的选取[J]. 信息技术与信息化, 2020, (09):100-102.

[32]王春枝, 邢绍文, 高榕, 等. 基于预训练交互式图神经网络的多元时间序列异常检测[J]. 中南民族大学学报(自然科学版), 2023, 42(04):541-550.

[33]杨彦霞, 王普, 高学金, 等. 基于混合双层自组织径向基函数神经网络的优化学习算法[J]. 北京工业大学学报, 2024, 50(1):38-49.

[34]张彦, 史天运, 李仕达, 等. AFC 技术及铁路自动售检票系统研究[J]. 中国铁路, 2009(03):50-55.

[35]曾平利. 城市轨道交通运输计划编制方法研究[J]. 铁路技术创新, 2017(03):18-24.

[36]张文. 城市轨道交通客流分析及系统设计研究[D]. 西安：长安大学，2022

[37]胡明伟, 何国庆, 吴雯琳, 等. 基于改进Logistic-SSA-BP神经网络的地铁短时客流预测研究[J]. 重庆交通大学学报(自然科学版), 2023, 42(03):90-97.

[38]曾诚, 吴佳媛, 罗霞. 城市轨道交通短时客流预测文献综述[J]. 铁道运输与经济, 2021, 43(8):105-111.

[39]胡聪丛, 胡桓. 深度神经网络的发展现状[J]. 电子技术与软件工程, 2017(04):29-31.

[40]Khishe M, Mosavi M.R. Chimp optimization algorithm[J]. Expert Systems with Applications, 2020, 149:113338.

[41]Fukushima K. A hierarchical neural network model for associative memory[J]. Biological Cybernetics, 1984, 50(02):105-113.

[42]Le Cun Y. Using curvature information to improve back-propagation[J]. Neural Networks, 1988, 1:168-171.

[43]Zhang Jinlei. Multi-graph convolutional network for short-term passenger flow forecasting in urban rail transit[J]. IET Intelligent Transport Systems, 2020, 14(10):1210-1217.

[44]Fan J, Zheng J, Wu L, et al. Estimation of daily maize transpiration using supportvector machines,extreme gradient boosting,artificial and deep neural networks models[J]. Agricultural Water Management, 2021, 245:106547.

[45]Kruthiventi, Srinivas S.S, Ayush K, et al. DeepFix:A fully convolutional neural network for predicting human eye fixations[J]. IEEE Transactions on Image Processing, 2017, 26(9):4446-4456.

[46]Liu D, Wu Z, Sun S. Study on Subway passenger flow prediction based on deep recurrent neural network[J]. Multimedia Tools and Applications, 2020, 81(14):18979-18992.

[47]蔡昌俊. 城市轨道交通进出站短时客流预测模型研究[J]. 城市轨道交通研究, 2021, 24(09):14-19.

[48]Hubel D.H，Wiesel T.N. Integrative action in the cat's lateral geniculate body[J]. The Journal of Physiology, 1961, 155(2):385.

[49]Jeffrey L, Elman. Finding structure in time[J]. Cognitive Science, 1990, 14(2):179-211.

[50]Hochreiter S，Schmidhuber J. Long short-term memory[J]. Neural Computation, 1997, 9(8):1735-1780.

[51]Cho K,Van Merrienboer B, Gulcehre C, et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation[J]. Arxiv Preprint Arxiv, 2014, 1406.

[52]Clerc M. The swarm and the queen:towards a deterxministic and adaptive particle swarm optimization[C]//Proceedings of the 1999 Congress on Evolutionary Computation, 1999, 3:1951-1957.

[53]Kao Y.T, Zahara E. A hybrid genetic algorithm and particle swam optimization for multimodal functions[J]. Applied Soft Computing, 2008, 8(2):849-857.

[54]Liu T，Yin S. An improved particle swarm optimization algorithm used for BP neural network and multimedia course-ware evaluation[J]. Multimedia Tools and Applications, 2017, 76(09):11961-11974.

[55]Li C, Gao L, Li X, et al. A hybrid multi-objective grey wolf optimizer for dynamic scheduling in a real-world welding industry[J]. Engineering Applications of Artificial Intelligence, 2017, 61-79.

[56]陈峰, 丁泉, 吴乐, 等. 混合驱动的粒子群算法[J]. 计算机工程与应用, 2024, 60(08):78-89.

[57]温惠英, 张东冉. 基于Bi-LSTM模型的高速公路交通量预测[J]. 公路工程, 2019, 44(06):51-56