Machine-learning technology powers many aspects of modern society. Compared to the conventional machine learning techniques that were limited in processing natural data in the raw form, deep learning allows computational models to learn representations of data with multiple levels of abstraction. In this study, an improved deep learning model is proposed to explore the complex interactions among roadways, traffic, environmental elements, and traffic crashes. The proposed model includes two modules, an unsupervised feature learning module to identify functional network between the explanatory variables and the feature representations and a supervised fine tuning module to perform traffic crash prediction. To address the unobserved heterogeneity issues in the traffic crash prediction, a multivariate negative binomial (MVNB) model is embedding into the supervised fine tuning module as a regression layer. The proposed model was applied to the dataset that was collected from Knox County in Tennessee to validate the performances. The results indicate that the feature learning module identifies relational information between the explanatory variables and the feature representations, which reduces the dimensionality of the input and preserves the original information. The proposed model that includes the MVNB regression layer in the supervised fine tuning module can better account for differential distribution patterns in traffic crashes across injury severities and provides superior traffic crash predictions. The findings suggest that the proposed model is a superior alternative for traffic crash predictions and the average accuracy of the prediction that was measured by RMSD can be improved by 84.58% and 158.27% compared to the deep learning model without the regression layer and the SVM model, respectively.

Given the enormous traffic issues, such as congestion and crashes, resulting from the conflicts between trucks and passenger cars, an accurate and reliable prediction of truck traffic flow is needed to enhance the traffic flow efficiency and safety in the mixed traffic condition. Enabled by emerging sensing technologies, the GPS data become available and will reveal some insights to improve the understanding of truck traffic flow prediction. In the paper, a novel method of truck traffic flow prediction is proposed by using sampled GPS data in the roadway network. The proposed method consists of two phases, which are expansion and prediction. In the data expansion phase, a piece-wise constant coefficient method is designed to minimize errors between the sampled truck flow and the actual truck flow, where the coefficients are determined according to road grades and traffic flow size. In the prediction phase, Long Short Term Memory (LSTM) and Gated Recursive Unit (GRU) neural network methods are first time employed to improve the prediction accuracy. Considering that the sequence of the expansion and prediction could have different prediction performance, approaches using both ‘previous-prediction’, ‘post-expansion’ and ‘previous-expansion’, ‘post-prediction’ were used and the results compared with the survey data from traffic flows. The results demonstrate that LSTM and GRU have a superior performance compared to existing approaches using SRV and ARIMA for truck traffic flow prediction. For the whole prediction period, LSTM has better prediction results than GRU overall with an accuracy which is 4.10% better than that of GRU. Furthermore, the accuracy of the ‘previous-prediction’, ‘post-expansion’ is 8.26% greater than that of the ‘previous-expansion’, ‘post-prediction’.

Understanding the factors that contribute to crash frequency and severity will assist better highway design and develop appropriate countermeasures for hot spots, thereby improving the safety of the road system. This study explores the influences of risk factors on frequency and severity of large truck–involved crashes. Multinomal logit (MNL) and negative binomial (NB) models are proposed to analyze crash severity and frequency, respectively. The explanatory factors include characteristics of the vehicles, drivers, traffic, environment, and roadway geometric design features. To obtain better parameter estimation results, the maximum likelihood (ML) method and Bayesian method are employed. The results show that the MNL and NB models have a better goodness of fit under the Bayesian estimation framework. Using Bayesian MNL and NB models, factors that significantly affect crash frequency and severity outcomes are analyzed. Some critical factors that contribute significantly to both crash severity and frequency estimation of large truck–involved crashes, including truck percentage, annual average daily traffic (AADT), driver condition, and weather condition, are identified and discussed. Driver age, speed limit, and location type are found to have significant effects only on the frequency of large truck–involved crashes. Seat belt usage, light condition, and terrain type are found to have significant effects only on the severity of large truck involved crashes. The results from this study will be valuable in transportation policy making, improvement of carrier operation, and crash-cost reduction.

The Regenerative Braking Energy (RBE) of metro trains plays an important role in metro energy saving. If the regenerative energy can be directly absorbed by the adjacent trains, the investment in other RBE usage equipments like super capacitors will be reduced. To increase the RBE usage, the overlaps of traction and braking processes of adjacent trains can be increased by timetable optimization or real-time train regulation. But if the overlaps cannot be realized, the RBE cannot be efficiently absorbed by the other trains. To solve this problem, this paper proposes an optimal train control method to absorb the RBE of the braking train by adjusting the speed profile of an adjacent train that runs at non-braking stage. A speed profile adjustment algorithm for the adjacent train is given considering the speed limits and gradients, in which a traction process is incorporated in the adjusted speed profile. Connected and unconnected DC power supply networks are both considered. In the connected DC power supply network, the RBE of the braking train can be absorbed by the adjacent adjusted train, which runs in the different power supply section. To test the robustness of the proposed algorithm, random train delays are considered in the simulations. The interstation running time will be regulated to reduce the train delay by a linear programming model, based on which, the speed profile of the cooperative train will be adjusted. Simulations are conducted based on a whole day data of Beijing Metro Line 4. The simulation results show that the proposed method improves the energy efficiency over 13% by improving RBE usage, and the proposed method also shows strong robustness in the train delay simulations, which makes it a promising method to be applied in real metro system.