This paper proposes a convolutional neural network (CNN)-based method that learns traffic as images and predicts large-scale, network-wide traffic speed with a high accuracy. Spatiotemporal traffic dynamics are converted to images describing the time and space relations of traffic flow via a two-dimensional time-space matrix. A CNN is applied to the image following two consecutive steps: abstract traffic feature extraction and network-wide traffic speed prediction. The effectiveness of the proposed method is evaluated by taking two real-world transportation networks, the second ring road and north-east transportation network in Beijing, as examples, and comparing the method with four prevailing algorithms, namely, ordinary least squares, k-nearest neighbors, artificial neural network, and random forest, and three deep learning architectures, namely, stacked autoencoder, recurrent neural network, and long-short-term memory network. The results show that the proposed method outperforms other algorithms by an average accuracy improvement of 42.91% within an acceptable execution time. The CNN can train the model in a reasonable time and, thus, is suitable for large-scale transportation networks.

This paper presents a vehicle detection and classification system using distributed fiber-optic acoustic sensing (DAS) technology and describes a comprehensive classification method including signal processing and feature extraction. This sensing device is based on Rayleigh scattering light and is used for real-time vehicle detection, classification, and speed estimation. Distributed acoustic signals from an arbitrary point can be detected and located through DAS technology which can provide fully distributed acoustic information along the entire fiber link. This technology utilizes sensing fiber in the form of distributed sensors to collect traffic vibration signals and then extracts several features from the signals to estimate the vehicle count and identify vehicle categories. According to the vehicle vibration signal characteristics, the wavelet-denoising algorithm and dual-threshold algorithm are improved. The improved algorithm is used to reconstruct the signal for feature extraction, and the vehicle count and speed are obtained. When all features have been extracted, the classification of vehicle types is implemented by a support vector machine classifier. The validation data (using a distributed fiber-optic acoustic sensor) demonstrate that the vehicle detection accuracy is higher than 80%, the speed estimation error is less than 5%, and the vehicle classification accuracy is higher than 70%.

This paper proposes a novel method for traffic flow detection using distributed optical fiber acoustic sensing (DAS). Different from the traditional traffic flow detection method, this method detects the traffic vibration signal using a fiber optic cable. Distributed fiber-optic acoustic sensing technology can provide fully distributed acoustic information along the entire fiber link, and thus external acoustic signals from an arbitrary point can be detected and located. This study uses DAS to obtain traffic vibration data. Using the characteristics of traffic vibration data, this paper proposes an improved wavelet threshold algorithm and an improved dual-threshold algorithm and verifies the feasibility and effectiveness of these methods. Finally, the experimental results from a vehicle-counting test show that the counting error is smaller for a single vehicle passing through the detection area and that the counting error is larger if multiple vehicles pass through the detection area continuously. In vehicle speed estimation, the results show good accuracy, and the error range is controlled to less than 6%.

In recent years, an innovative public transportation (PT) mode known as the customized bus (CB) has been proposed and implemented in many cities in China to efficiently and effectively shift private car users to PT to alleviate traffic congestion and traffic-related environmental pollution. The route network design activity plays an important role in the CB operation planning process because it serves as the basis for other operation planning activities, for example, timetable development, vehicle scheduling, and crew scheduling. In this paper, according to the demand characteristics and operational purpose, a methodological framework that includes the elements of large-scale travel demand data processing and analysis, hierarchical clustering-based route origin-destination (OD) region division, route OD region pairing, and a route selection model is proposed for CB network design. Considering the operating cost and social benefits, a route selection model is proposed and a branch-and-bound-based solution method is developed. In addition, a computer-aided program is developed to analyze a real-world Beijing CB route network design problem. The results of the case study demonstrate that the current CB network of Beijing can be significantly improved, thus demonstrating the effectiveness of the proposed methodology.

With the development of big data and deep learning, bus passenger flow prediction considering real-time data becomes possible. Real-time traffic flow prediction helps to grasp real-time passenger flow dynamics, provide early warning for a sudden passenger flow and data support for real-time bus plan changes, and improve the stability of urban transportation systems. To solve the problem of passenger flow prediction considering real-time data, this paper proposes a novel passenger flow prediction network model based on long short-term memory (LSTM) networks. The model includes four parts: feature extraction based on Xgboost model, information coding based on historical data, information coding based on real-time data, and decoding based on a multi-layer neural network. In the feature extraction part, the data dimension is increased by fusing bus data and points of interest to improve the number of parameters and model accuracy. In the historical information coding part, we use the date as the index in the LSTM structure to encode historical data and provide relevant information for prediction; in the real-time data coding part, the daily half-hour time interval is used as the index to encode real-time data and provide real-time prediction information; in the decoding part, the passenger flow data for the next two 30 min interval outputs by decoding all the information. To our best knowledge, it is the first time to real-time information has been taken into consideration in passenger flow prediction based on LSTM. The proposed model can achieve better accuracy compared to the LSTM and other baseline methods.

To alleviate traffic congestion and traffic-related environmental pollution caused by the increasing numbers of private cars, public transport (PT) is highly recommended to travelers. However, there is an obvious contradiction between the diversification of travel demands and simplification of PT service. Customized bus (CB), as an innovative supplementary mode of PT service, aims to provide demand-responsive and direct transit service to travelers with similar travel demands. But how to obtain accurate travel demands? It is passive and limited to conducting online surveys, additionally inefficient and costly to investigate all the origin-destinations (ODs) aimlessly. This paper proposes a methodological framework of extracting potential CB routes from bus smart card data to provide references for CB planners to conduct purposeful and effective investigations. The framework consists of three processes: trip reconstruction, OD area division and CB route extraction. In the OD area division process, a novel two-step division model is built to divide bus stops into different areas. In the CB route extraction process, two spatial-temporal clustering procedures and one length constraint are implemented to cluster similar trips together. An improved density-based spatial clustering of application with noise (DBSCAN) algorithm is used to complete these procedures. In addition, a case study in Beijing is conducted to demonstrate the effectiveness of the proposed methodological framework and the resulting analysis provides useful references to CB planners in Beijing.