Bus Dynamic Travel Time Prediction: Using a Deep Feature Extraction Framework Based on RNN and DNN

Accurate real-time bus arrival information is essential for an efficient public transport network, as it significantly impacts passenger experience, system reliability, reduced waiting times, dwell times, and operational efficiency. In Sri Lanka's public transport system, current bus arrival time computations primarily rely on static data, neglecting real-time information and critical factors influencing travel times. This highlights the need to identify the unique variables affecting bus arrival times within the Sri Lankan context and to develop robust prediction models that account for these influences. While traditional methods such as Historical Average Models, Regression, Time Series Analysis, and Kalman Filtering have been used in previous research for short-term travel time predictions, Machine Learning (ML) approaches have proven to deliver superior accuracy. ML models are regarded as the most effective for heterogeneous, lane-less traffic conditions with varying traffic volumes, such as those found in Sri Lanka. ML techniques excel in processing large, high-quality datasets and provide accurate predictions by accounting for all relevant variables influencing travel times. Although research has been conducted on developing various basic ML models for travel time prediction, there is a noticeable gap in studies comparing these models to determine the most suitable one for the Sri Lankan context. A Long- Short Term Memory (LSTM) neural network is a deep learning model that is capable of handling long-term dependencies. In the context of bus travel time prediction, LSTMs can leverage historical traffic and travel data to capture temporal patterns and fluctuations that influence travel times. By evaluating LSTMs against basic machine learning models, this study seeks to explore the advantages of applying deep learning techniques to transportation forecasting, ultimately contributing to more accurate and efficient predictive systems in transit planning. The ML models selected in this study include two basic traditional models K- Nearest Neighbours (KNN) and Support Vector Regression (SVR) and four advanced models that utilize ensemble techniques and advanced optimization as Random Forest Regression (RFR), Ada Boost, XG Boost and Gradient Boosting Machine (GBM). The performance of these models was compared with the LSTM model to identify the gap in their accuracies.

Reliable short-term prediction of bus travel time on signalized urban arterials is essential for improving service reliability and may provide a useful forecasting basis for prediction-informed transit signal priority (TSP) and arterial coordination applications. However, bus operations on urban arterials are highly variable due to stop dwell times, signal delays, and interactions with mixed traffic, leading to nonlinear and nonstationary travel time patterns with strong spatiotemporal dependence. This study proposes a hybrid KNN–ConvLSTM framework for short-term arterial bus travel time prediction using real-world field data. A K-nearest neighbors (KNNs) module is first employed to retrieve historical operation sequences that are most similar to the current corridor state, thereby reducing interference from mismatched traffic regimes and improving robustness. Smart-card (IC card) transaction data are incorporated as demand-related features to represent passenger activity and its impact on dwell time and travel time variability. The selected sequences are then organized into a corridor-ordered spatiotemporal representation and further refined by lightweight temporal enhancement operations, including relevance gating, multi-scale aggregation, adaptive feature fusion, and residual enhancement, before being fed into the convolutional long short-term memory (ConvLSTM) predictor. The proposed approach is evaluated using weekday service-hour data extracted from 30 days of real-world bus operation records collected from a typical urban arterial corridor in Changchun, China, and is compared with several benchmark models, including ARIMA, KNN, LSTM, CNN, ConvLSTM, Transformer, and DCRNN. The results indicate that the proposed KNN–ConvLSTM framework achieves an MAE of 40.1 s, an RMSE of 55.8 s, a SMAPE of 10.7%, and an R2 of 0.878, outperforming all benchmark models. Specifically, compared with the Transformer baseline, the proposed framework reduces MAE by 1.5%, RMSE by 5.1%, and SMAPE by 7.0%, while increasing R2 by 0.014. Compared with the DCRNN baseline, it reduces MAE by 10.7%, RMSE by 1.9%, and SMAPE by 2.7%, while increasing R2 by 0.008. These findings demonstrate that similarity-aware retrieval combined with spatiotemporal deep learning can substantially enhance short-term bus travel time prediction on signalized urban arterials. More accurate short-term forecasts may support prediction-informed transit signal priority and arterial coordination by providing more reliable downstream arrival-time estimates. However, the generalizability of the reported results is still constrained by the relatively short 30-day observation period and the single-corridor case setting, and the operational and environmental effects of downstream applications remain to be validated through dedicated closed-loop control evaluation in future work.

Optimizing the bus operation plan Based on Deep Learning

Multi-attention graph neural networks for city-wide bus travel time estimation using limited data

Ensuring accurate and dependable predictions of bus arrival times is essential to improving public transportation services and maintaining their appeal in urban settings. Such predictions, whether displayed on electronic boards or integrated into mobile applications, enable passengers to make better travel decisions, such as choosing alternate routes, anticipating delays, or avoiding missed connections. Furthermore, advanced Intelligent Transport Systems (ITS) utilize this information to facilitate smoother passenger transfers by holding delayed services within predefined limits. However, as urban congestion and travel time unpredictability grow, traditional methods face significant challenges in providing reliable predictions, making the problem increasingly complex. This research focuses on developing a robust system for forecasting bus arrival times in Astana city, utilizing extensive spatio-temporal data from two datasets. Multiple machine learning and deep learning models are implemented and compared to achieve this goal. These include K-means clustering to classify bus routes, K-Nearest Neighbors (KNN) for predictions based on proximity, and a Conv-LSTM model, which integrates convolutional and long short-term memory layers to address intricate temporal and spatial correlations. Support Vector Machines (SVM) and regression models are also incorporated to establish benchmarks and comparative insights. Through empirical evaluation, the proposed models demonstrate varying strengths, with the Conv-LSTM model showing exceptional performance in adapting to dynamic urban conditions and detecting subtle fluctuations in bus travel times. The findings highlight the transformative potential of sophisticated predictive modeling techniques to enhance urban transit systems, ensuring passengers receive timely and accurate information while improving overall operational efficiency.

Bus running state is always influenced by traffic flow. Traffic flow state evolution directly leads to the variation of bus speed, thereby affecting bus progress. Some investigations implied that there is a certain relationship between the bus speed and other traffic vehicle speed on the road. However, the relationship is different due to individual forms of road cross-sections, the number of lanes, and the traffic flow characteristics. Based on this mechanism and the forecast of road traffic flow data provided by intelligent transportation system (ITS), a state recognition-based model for travel time prediction is proposed, and three running states are introduced in the model. Finally, the bus travel time predicted by the proposed model are assessed with real-world data collected from bus route number 53 in Wuxi city. Results show that the proposed model outperforms the history data-based model, especially when the bus running state falls unstable; the model achieves satisfactory performance.

Ensemble machine learning for travel time prediction: An Active learning-assisted approach

Many intelligent transportation systems (ITS) often rely on prediction of traffic variables in the near future to provide useful information for the users. Therefore, accurate traffic prediction is essential in the development of ITS. In this study, we are comparing prediction models for traffic speed and travel time on urban traffic network. We present three ways to build travel time predictors: by using recent travel time and speed of neighboring links, by using recent travel time of individual links, and by using speed predictors. Model-tree ensemble is implemented for both traffic speed and travel time predictions. Experimental result shows that the predictors perform better on travel time prediction than on traffic speed prediction. Among the three discussed travel time prediction methods and two baseline predictors, travel time prediction via speed predictors achieves best accuracy in all tested paths and prediction horizons.

Information about public transport travel time is a key indicator of service performance, and is valued by passengers and operators. Among many different approaches, Support Vector Machines (SVM) has recently gained attention in predicting bus travel times. The training process of SVMs involves solving a quadratic programming problem which is slow when dealing with large training data. This paper proposes a Least Squares SVM (LS-SVM) method that expedites the training process by simplifying the quadratic programming problem using a linear regression technique. Also, to ensure the accuracy of the prediction results, a Genetic Algorithm (GA) is used to determine the optimal set of model parameters. The GA based LS-SVM approach is tested using real-world travel time data from a bus route in Melbourne, Australia. The comparison of the results in this paper to those obtained in a previous study using artificial neural networks shows that the proposed method produces more accurate results.

Prediction concept needs human-like thinking and ideas. ANN helps make a program for describing the solutions to the prediction problems. To gain accurate prediction, there is a need for deep thinking in the knowledge system discovered. Deep learning helps form such next-level type of ANN thinking. Deep learning models do not require humans for programming the problems. It can determine the prediction by itself by learning the dataset. This paper describes the prediction by considering sequential model for continuous accessing of data with the help of multi-layers. Deep learning is very helpful for predicting larger dataset. This paper proposes an algorithmic model known as a Comprehensive Deep Recurrent Artificial Neural Networks (CDRANN) by hybridizing LSTM (long short-term memory) with RNN (recurrent neural network) backtrack solver for better future stock prediction. This paper also explains the need for hybridizing evolutionary model CDRANN to create a novel change in future prediction.KeywordsCDRANN (Comprehensive Deep Recurrent Artificial Neural Networks)LSTM (long short-term memory)RNN (recurrent neural network)Stock marketPrediction

The accuracy of travel time prediction is not only decided by the state of the current journey but also affected by that of the transfer journey. This paper focuses on the transit transfer problem of urban public transport and aims to improve the performance of multi-route bus travel time prediction. In this paper, two kinds of infect factors, which are macro infect factors and local infect factors, are defined. Then, a pre-processing approach is proposed to construct a travel chain, which is used to compensate the shortage of raw data set. Multi-route bus travel time is divided into four parts, which are travel time of the current driving segment (CDS), bus stop dwelling time module, stop-stop travel time, and transfer point waiting time. A parrel processing architecture, which includes four learning blocks, is designed to fulfill the prediction process of the above four time parts. The classic LSTM model is used to predict the CDS travel time and transfer point waiting time. An improved DA-RNN model, the MDARNN model, is proposed to predict the bus stop dwelling time module, stop-stop travel time. Moreover, a real-time traffic flow model is used to calibrate the prediction time of all four sub-models. Experiment results prove the validity of the proposed methods.

Accurate travel time prediction for major freeways and corridors is crucial but challenging when road incidents happen. Data-driven models require a large set of historical data to estimate the spatial and temporal correlations between road incidents and traffic dynamics. More often than not, the amount of historical data under non-recurring conditions is limited when it comes to training the models. This paper investigates the application of data-driven models on an enriched database with simulated travel times. A well-calibrated traffic simulation is used to capture the artificial incident’s impact on a major urban corridor in Sydney, Australia. This procedure is repeated for multiple created incidents, resulting in a synthetic dataset validated by the available actual historical data. Several machine learning models, such as Regression Tree, Support Vector Regression, Extreme Gradient Boosting, and Recurrent Neural Networks are trained and tested based on the simulated travel time and incident information. As a baseline model for comparison, the measured travel time at the prediction time is considered equal to multi-step ahead travel time. Based on the results, the data-driven models developed with the simulated data outperformed the baseline, indicating that our approach can be effectively employed in the travel time prediction.

Travel time prediction is highly useful in traffic management and planning. The prediction's accuracy relies on the accuracy of the travel time data used. Various sensors are being used to collect travel time data, including link-based and node-based sensors. Recently, a new method in the collection of travel time data was introduced: Bluetooth technology, used to detect Bluetooth devices in vehicles and determine the vehicles' travel times. Bluetooth sensors are generally node-based sensors. Despite the amount of literature available on sensor location problems, the use of node-based sensors to collect travel time data is rarely discussed. Projects that looked at the collection of travel time data with Bluetooth sensors prompted this study of the problem of sensor location in the installation of Bluetooth sensors and, more generally, node-based sensors. The goal of this study is to find the optimal number of node-based sensors and their optimal deployment locations in a network for the collection of highly reliable travel time data. Two formulations are proposed for modeling this problem. The formulations consider a new set of reliability factors. Through these formulations, the sensor location problem can be solved optimally for large networks. The proposed formulations are not restricted to Bluetooth sensors and can be applied to any node-based sensor location problem. Various case studies that use real-world networks are conducted to compare the results obtained from both the proposed formulations with the methodologies available in the literature. The findings of the case studies are reported in the paper.

Average delays are an example of traffic network performance measures. They can be measured at intersections to estimate the average delay per vehicle at various levels, such as intersection, approach, or lane group. On the other hand, average delays at a given route are implicitly measured by estimating the difference between free-flow travel time to the observed ones. Different methods are used to estimate travel time for a given route, such as floating car, average speed, and vehicle tracking methods. This paper focuses on developing an artificial neural networks (ANN) model to predict travel time for specific routes based on field travel time measurements and other easy to measure characteristics, that are related to geometric layouts, peak-hour periods, posted speed limits, and route lengths. Travel time data for 75 different segments located in the State of Qatar were studied. Directional travel time data were measured in three different peak periods. A total of 450 travel time values were collected and analyzed. An ANN model was trained to estimate travel time. The results show that the ANN model was able to provide a reasonable estimation of travel time using limited information. The slopes of the regression plots between observed and predicted travel time values show a clear linear trend, with slopes around 0.85, and an intercept that is around 2.0.

Hybrid time series forecasting methods for travel time prediction