Pedestrians do not always comply with the crossing rules of when and/or where to cross the road at signalized intersections. This risky behavior tends to undermine greatly the effectiveness of safety countermeasures at such locations. Thus, it is very important to understand illegal behavior to develop more effective and targeting measures. In order to address the problem, this paper aimed to analyze characteristics of illegal crossings and their impact on behavior choice. Firstly, illegal crossing behaviors at signalized intersections were classified into two categories, including “crossing at a red light” and “crossing outside of a crosswalk.” Secondly, two sets of data were collected to understand the behaviors. One set of data was collected from video-based observation conducted at 3 signalized intersections in Guangzhou, China, capturing 3334 valid illegal crossing cases in total. Another set of data, from a questionnaire survey conducted online, resulted in 275 valid responses. Finally, presentational characteristics of illegal crossings at signalized intersection were analyzed and two Bayesian network-based behavior models were developed to investigate the characteristics and their impacts on the two types of illegal crossing behaviors, “crossing at a red light” and “crossing outside of a crosswalk,” respectively. Findings reveal that, (i) illegal crossings occur at various types of signalized intersections, with a higher probability for “crossing outside of a crosswalk” compared to “crossing at a red light;” (ii) Arc routing crossing has the highest probability to occur at signalized intersections compared to other types of out-side-crosswalk crossings. (iii) The location of origin and destination of a pedestrian has a significant effect on crossing outside of a crosswalk, the location of origin and destination of “one is inside of a crosswalk and another is outside of a crosswalk” has a highest proportion. These findings provide better understanding of illegal crossings and their impact factors so that the effectiveness of management and control of pedestrians at signalized intersections can be improved.

Traffic volume data are the important part of research and application of intelligent transportation systems (ITS). However, data loss often happens due to various factors in the real world, which may cause large deviations in prediction or bad accuracy of optimizations. Imputation is a valid way to handle missing values. A multilayer perceptron-multivariate imputation of chain equation (MLP-MICE) regression imputation method optimized by the limit-memory-BFGS algorithm is proposed, considering the temporal and spatial characteristics of traffic volume. Also, 32 groups of simulated imputation experiments based on the detected traffic volume of road sections in the Guangdong freeway system are conducted, which take the scenarios of continuous missing and jumped missing into account. The results of the experiments show that the MLP-MICE can optimize the imputation performance in the missing value of traffic volume with the MAPE of imputation results from 6.38% to 30%. Meanwhile, the proposed model has higher imputation accuracy for the traffic volume data with a lower degree of mutation. Lastly, the performance of the proposed model of imputation in short-term traffic volume prediction is discussed using the support vector machine. The results of it show that the MAPE of prediction under the proposed model is much lower than all-zero imputation. Therefore, the proposed model in this study is positive on improving the accuracy of traffic volume prediction and intelligent traffic control and management.

Modeling individual and crowd behaviors is the basis for evacuation research, which is essential to reduce casualties under emergency conditions in large venues. This paper focuses on the pedestrian evacuation of large-scale venues with multiple exits, and proposes an improved pedestrian evacuation model which uses a novel dual-strategy adaptive particle swarm optimization algorithm with affinity propagation clustering. Compared with the traditional models, the proposed model is demonstrated to have a better performance in simulating pedestrians’ herding behavior in panic situations, especially when not familiar with the environment, as i) individual heterogeneity is considered and affinity propagation clustering is used in population division to simulate the process of people spontaneously gathering into swarms; ii) A dual-strategy updating scheme is designed to balance the cognition difference between the leaders and the agents in the swarm; and iii) Adaptive control of parameters is used to simulate the human-surrounding interaction and psychological fluctuates when the evacuees are moving towards exits. Numerical examples, which simulated the evacuation of a rectangle venue with multiple exits, demonstrate the effectiveness and practicality of the proposed model. Moreover, the influences of pedestrian velocity, characteristics of exits, and leader movement on evacuation are analyzed. Experimental results show that the movement of leaders is different from other evacuees and the parameters of doors, such as width, quantity, and location all have a great influence on evacuation time.

The safety of signalized intersections is of great concern. To allow for an effective evaluation measure on the safety level of intersections, traffic conflict analysis methods are commonly used. However, the existing literature has mainly focused on the statistical prediction of conflicts by using surrogate measurements, among which the spatial-temporal characteristics of the potential conflicts have been less addressed. In addition, most of the relevant studies rely on precise trajectory data, and the results could be limited to engineering applications when real-time/comprehensive trajectory data are not available. To address these issues, this study proposes a SICP (signalized intersection conflict probability) model to predict a straight-left traffic flow conflict with a spatial-temporal distribution in the heat map, which could effectively evaluate the traffic safety of the existing or prebuilt signalized intersections on urban roads. Firstly, the impact of vehicle movement characteristics on traffic conflict at signalized intersections was considered by incorporating the vehicle movement trajectory. Secondly, the signal phase was categorized in several stages (each phase contains switching and nonswitching stages); then, a vehicle-vehicle conflicts probability prediction model was established by integrating both horizontal and vertical arrival probability. Finally, to validate the performance of the proposed model, the measured data were collected from the intersection of Wushan road and Yuehan road in Tianhe District, Guangzhou, China. SSAM（Surrogate Safety Assessment Model）traffic conflict simulation was used to analyze the traffic conflict in the actual data and compared to the SICP model. A case study was conducted to reveal the evolution mechanism of the conflict risk coefficient at the signalized intersection and to estimate the safety status under the various security optimization strategies. The experimental results verified the effectiveness of the SICP model, indicating that the proposed model is effective in evaluating the safety level of existing or prebuilt signalized intersections.

This paper presents a short-term traffic prediction method, which takes the historical data of upstream points and prediction point itself and their spatial-temporal characteristics into consideration. First, the Gaussian mixture model (GMM) based on Kullback–Leibler divergence and Grey relation analysis coefficient calculated by the data in the corresponding period is proposed. It can select upstream points that have a great impact on prediction point to reduce computation and increase accuracy in the next prediction work. Second, the hybrid model constructed by long short-term memory and K-nearest neighbor (LSTM-KNN) algorithm using transformed grey wolf optimization is discussed. Parallel computing is used in this part to reduce complexity. Third, some meaningful experiments are carried out using real data with different upstream points, time steps, and prediction model structures. The results show that GMM can improve the accuracy of the multifactor models, such as the support vector machines, the KNN, and the multi-LSTM. Compared with other conventional models, the TGWO-LSTM-KNN prediction model has better accuracy and stability. Since the proposed method is able to export the prediction dataset of upstream and prediction points simultaneously, it can be applied to collaborative management and also has good potential prospects for application in freeway networks.