Chen Dongwei

Bio: Chen Dongwei is an academic researcher from Beihang University. The author has contributed to research in topics: Bus bunching & Smart card. The author has an hindex of 4, co-authored 5 publications receiving 111 citations.

Headway-based bus bunching prediction using transit smart card data

Abstract: Bus bunching severely deteriorates the quality of transit service with poor on-time performance and excessive waiting time. To mitigate bus bunching, this paper presents a predictive framework to capture the stop-level headway irregularity based on transit smart card data. Historical headway, passenger demands, and travel time are utilized to model the headway fluctuation at the following stops. A Least Squares Support Vector Machine regression is established to detect bus bunching with the predicted headway pattern. An empirical experiment with two bus routes in Beijing is conducted to demonstrate the effectiveness of the proposed approach. The predictive method can successfully identify more than 95% of bus bunching occurrences in comparison with other well-established prediction algorithms. Moreover, the detection accuracy does not significantly deteriorate as the prediction lead time increases. Instead of regularizing the headways at all costs by adopting certain correction actions, the proposed framework can provide timely and accurate information for potential bus bunching prevention and inform passengers when the next bus will arrive. This feature will greatly increase transit ridership and reduce operating costs for transit authorities.

Probabilistic Prediction of Bus Headway Using Relevance Vector Machine Regression

Abstract: Bus headway regularity heavily affects transit riders’ attitude for choosing public transportation and also serves as an important indicator for transit performance evaluation. Therefore, an accurate estimate of bus headway can benefit both transit riders and transit operators. This paper proposed a relevance vector machine (RVM) algorithm to predict bus headway by incorporating the time series of bus headways, travel time, and passenger demand at previous stops. Different from traditional computational intelligence approaches, RVM can output the probabilistic prediction result, in which the upper and lower bounds of a predicted headway within a certain probability are yielded. An empirical experiment with two bus routes in Beijing, China, is utilized to confirm the high precision and strong robustness of the proposed model. Five algorithms [support vector machine (SVM), genetic algorithm SVM, Kalman filter, k-nearest neighbor, and artificial neural network] are used for comparison with the RVM model and the result indicates that RVM outperforms these algorithms in terms of accuracy and confidence intervals. When the confidence level is set to 95%, more than 95% of actual bus headways fall within the prediction bands. With the probabilistic bus headway prediction information, transit riders can better schedule their trips to avoid late and early arrivals at bus stops, while transit operators can adopt the targeted correction actions to maintain regular headway for bus bunching prevention.

Method for predicting bus passenger waiting time range based on IC card data

Abstract: The invention discloses a method for predicting bus passenger waiting time range based on IC card data. The method comprises the following steps: step 1, a sample set is created; step2, RVM is modeling and the parameter calibration thereof are performed, steps, the time headways of the Bus A and Bus B on the second target station are predicted; the biggest characteristic of the invention comprises that the method can predict that the next bus which will arrive at some station at some point to some point, the blank of the bus arrival time range prediction is filled, the method has great innovative significance, and provides a more reasonable prediction method than the other patents; the method adopts IC card data to extract a mass of passenger information from a plurality of buses without using the vehicle-mounted GPS, the method is convenient and fast, the data processing cost is reduced, the IC card data can obtain the number of people who get on or off the bus at every station, so that the bus stay time at every station is obtained, and the above number and time can not be provided by the GPS data.

Bus bunching predication method based on IC card data

Abstract: The invention discloses a bus bunching prediction method based on IC card data, and belongs to the technical field of public transport information processing The prediction method comprises the steps of bus IC card data collection, data processing, practical bus bunching state detection, data training and learning and bus bunching prediction, wherein bus bunching prediction uses a least square support vector machine (LS-SVM) algorithm IC card data of buses is combined, a lot of passenger information is extracted from multiple buses, a vehicle-mounted GPS system is not needed, operation is convenient and rapid, and the data processing cost is reduced; the LS-SVM algorithm can be used to realize bus bunching prediction more rapidly and more effectively, passengers can better understand operation state of buses and regulate travel time reasonably, and a public transport operation department can adjust the departure interval of buses timely and improve the service level of public transport; and data processing is simple, the cost is low, and the prediction precision is higher

Bus punctuality prediction method based on GPS data

Abstract: The invention discloses a bus punctuality prediction method based on GPS data, and belongs to the field of public transport information processing. The method comprises the steps: the collection and processing of bus GPS data and departure data, the determining of a bus punctuality value, the extraction of bus punctuality influence factors and the prediction of bus punctuality. The method employs an SVM (support vector machine) algorithm. The method combines the bus GPS data, extracts the track information and departure information of a plurality of buses, is convenient and quick, and reduces the data processing cost. Moreover, the method employs the SVM algorithm for the two-class prediction of the punctual conditions of a downstream stop, and enables passengers to know the bus operation conditions better to reasonably adjust the travel time. Meanwhile, the method enables a bus operation department to be able to timely adjust the departure interval of buses, and improves the bus service level.

Big data analytics architecture for real-time traffic control

Abstract: The advent of Big Data has triggered disruptive changes in many fields including Intelligent Transportation Systems (ITS). The emerging connected technologies created around ubiquitous digital devices have opened unique opportunities to enhance the performance of the ITS. However, magnitude and heterogeneity of the Big Data are beyond the capabilities of the existing approaches in ITS. Therefore, there is a crucial need to develop new tools and systems to keep pace with the Big Data proliferation. In this paper, we propose a comprehensive and flexible architecture based on distributed computing platform for real-time traffic control. The architecture is based on systematic analysis of the requirements of the existing traffic control systems. In it, the Big Data analytics engine informs the control logic. We have partly realized the architecture in a prototype platform that employs Kafka, a state-of-the-art Big Data tool for building data pipelines and stream processing. We demonstrate our approach on a case study of controlling the opening and closing of a freeway hard shoulder lane in microscopic traffic simulation.

Mining Smart Card Data for Transit Riders’ Travel Patterns

Abstract: To mitigate congestion caused by the increasing number of privately owned automobiles, public transit is highly promoted by transportation agencies worldwide. With a better understanding of the travel patterns and regularity (the “magnitude” level of travel pattern) of transit riders, transit authorities can evaluate the current transit services to adjust marketing strategies, keep loyal customers and improve transit performance. However, it is fairly challenging to identify travel pattern for each individual transit rider in a large dataset. Therefore, this paper proposes an efficient and effective data-mining approach that models the travel patterns of transit riders using the smart card data collected in Beijing, China. Transit riders’ trip chains are identified based on the temporal and spatial characteristics of smart card transaction data. Based on the identified trip chains, the Density-based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is used to detect each transit rider’s historical travel patterns. The K-Means++ clustering algorithm and the rough-set theory are jointly applied to clustering and classifying the travel pattern regularities. The rough-set-based algorithm is compared with other classification algorithms, including Naive Bayes Classifier, C4.5 Decision Tree, K-Nearest Neighbor (KNN) and three-hidden-layers Neural Network. The results indicate that the proposed rough-set-based algorithm outperforms other prevailing data-mining algorithms in terms of accuracy and efficiency.

Dynamic headway control for high-frequency bus line based on speed guidance and intersection signal adjustment

Abstract: Computer-Aided Civil and Infrastructure Engineering To prevent bus bunching, a dynamic headway control method in the V2I (vehicle to infrastructure) environment for a high-frequency route with bus lane is developed. Bus operating speed guidance on the mid-blocks and intersection signal adjustment are two main strategies in the proposed method. A forecasting model of bus travel time under the dynamic control method is developed. The objective function is set up by taking into account differences between actual bus headways and dispatching headways, and the scaling ratios of intersection cycle lengths. The optimization model is solved using genetic algorithm. The proposed method is applied to a real bus route in Meihekou city, China, and compared with the current control plan as well as holding strategy. Results show that the proposed method can reduce bus headway deviations in all investigating periods; negative impacts on cars can be limited by setting reasonable values for the parameters.

Comparing bus holding methods with and without real-time predictions

Abstract: On high-frequency routes, transit agencies hold buses at control points and seek to dispatch them with even headways to avoid bus bunching. This paper compares holding methods used in practice and recommended in the literature using simulated and historical data from Tri-Met route 72 in Portland, Oregon. We evaluated the performance of each holding method in terms of headway instability and mean holding time. We tested the sensitivity of holding methods to their parameterization and to the number of control points. We found that Schedule-Based methods require little holding time but are unable to stabilize headways even when applied at a high control point density. The Headway-Based methods are able to successfully control headways but they require long holding times. Prediction-Based methods achieve the best compromise between headway regularity and holding time on a wide range of desired trade-offs. Finally, we found the prediction-based methods to be sensitive to prediction accuracy, but using an existing prediction method we were able to minimize this sensitivity. These results can be used to inform the decision of transit agencies to implement holding methods on routes similar to TriMet 72.