Analytic Methods in Accident Research最新文献

This paper seeks to identify factors that influence driver injury severities in single-vehicle freeway crashes when airbags deployed and when airbags did not deploy. Injury-severity models were estimated using random parameters logit models with consideration given to possible heterogeneity in means and variances of the random parameters to account for unobserved heterogeneity. Three years of pre-COVID pandemic crash data (2016, 2017 and 2018) from the state of Alabama were used in the model estimations. Models were estimated with data from all years, but the model formulation allowed the estimated parameters to vary by year. The model estimation results show that there are fundamental differences in crashes where airbags deployed (which tend to be crashes associated with greater energy transfers and variance in such transfers across crashes) relative to crashes where airbags did not deploy (which tend to be crashes associated with lower-speed impacts with less variance in energy transfers across crash observations). Moreover, the effects of most of the explanatory variables on resulting injury severities were found to vary significantly over time. However, explanatory variables such as shoulder and lap belt use, driver gender, newer model year vehicles, passenger car vehicle types, urban-located crashes, collisions with deer, collisions with trees and collisions with cable barriers did not vary significantly over time in either the airbag or non-airbag deployed models, or both. The findings of this study suggest that there is a potential for advances airbag systems to substantially improve safety by closing the injury-severity gap observed between men and women in particular, and that there is a need to further explore the evolution of driver behavior over time, which ultimately determines the effectiveness of ongoing improvements in vehicle and highway safety systems.

Challenges addressing overdispersion, unobserved heterogeneity, the preponderance of zeros, and correlation in the dependent variables of crash count models are of significant interest. Accounting for all these data issues simultaneously is few and far between. This study proposes a new mixing distribution model that accounts for overdispersion and the preponderance of zeros in crash count models. The proposed mixing distribution model extends to the multivariate structure to account for correlations between dependent variables and unobserved heterogeneity. The empirical analysis is conducted on crash data of Bruce highway involving single-vehicle and multi-vehicle crash types by “fatal and severe injury” and “moderate and minor injury” severity levels on aggregated data over three analysis years (2016, 2017, and 2018). The study demonstrates superior goodness of fit of the proposed multivariate random parameters Poisson lognormal-Lindley model compared to its restricted models. Moreover, pooling the crash data as repeated measures of crash types helped formulate a pooled-univariate random parameters Poisson-Lindley model to estimate multiple crash types by severity. The results showed the pooled-univariate model offers comparable goodness of fit and averaged marginal effects as the complex multivariate modeling structure. Moreover, the proposed pooled-univariate model reduced the model complexity to a one-dimensional integral and offered more efficient parameter estimates. In the empirical context, the modeling results showed that single-vehicle and multi-vehicle crashes by severity are linked with different causality.

Speeding behaviors can be classified into different patterns according to both speeding-range and speeding-distance. Among the speeding patterns, some are more frequently observed in specific traffic scenarios, implying that the likelihood of speeding behaviors may vary across the speeding patterns due to the inconsistent impact of temporal, road, environmental, and other traffic factors. Additionally, the trigger of speeding is a complex process so the researchers may not have access to all the critical information associated with the speeding behaviors. This issue may bring about not only independent heterogeneity but also multi-dimensional heterogeneities that are mutually correlated when modeling speeding behaviors by patterns. However, the joint solution to the above challenges is rarely seen in past studies. To fill the knowledge gaps, this study uses taxi GPS trajectories to extract speeding behaviors and classify them into four patterns. The speeder’s percent of speeding distance for each speeding pattern is respectively measured to represent the likelihood of speeding behaviors by patterns. Afterwards, we compare the data-fitting between the models combined with different beta-gamma mixture distributions and a multivariate Gaussian error in modeling the four patterns of speeding likelihood. The combination with the best fitness is used to incorporate jointly correlated random parameters. The results show that the model with beta-gamma-gamma-gamma mixed distributions performs better than other combinations. The model with jointly correlated random parameters outperforms models with other random parameters. The factor analysis reveals that percent of driving at night, percent of driving on roads with a low-speed limit (≤30 km/h), average delays in junctions along the trips, and hourly income have consistent effects on the likelihood of speeding behaviors in all patterns, while the effects of the remaining factors are inconsistent across the speeding patterns. Furthermore, the heterogeneity unveiled by the model parameters is discussed. The study highlights the necessity of considering mixed distributions and multi-dimensional heterogeneities in modeling speeding likelihood by different patterns.

Count regression models have been applied to model expected crash frequency at individual roadway locations. Random parameters have been increasingly integrated into these models to account for unobserved heterogeneity. However, the introduction of random parameters might also mask issues in the model specification, leading to inaccurate relationships and model interpretation. Two of these specification-related issues are: (1) not considering the appropriate functional form of explanatory variables; and, (2) ignoring the best set of significant explanatory variables. To better examine the need for careful model specification, this study uses synthetic data to demonstrate that the consideration of random parameters does not address the two model specification issues identified. The results from the simulation study illustrate that (a) model specification issues cannot be circumvented by random parameters alone and (b) random parameter models including the exhaustive set of explanatory variables available offer significant model improvements.

The adverse road surface condition has been identified as an important factor resulting in serious casualties and property losses in traffic accidents, and there is a tremendous need to uncover the interaction mechanism between deteriorating road surfaces and vehicle impact locations on the driver injury severity at a disaggregate level. In this paper, three groups of random parameters logit models with heterogeneity in means (and variances) are developed to investigate the unobserved heterogeneity and temporal stability of the determinants affecting driver injury severity outcomes across different damage locations among single-vehicle crashes that occurred under adverse weather conditions. A three-year crash dataset gathered from January 1, 2015, to December 31, 2017, in Ohio is utilized. Three crash injury severity categories including no injury, minor injury, and severe injury are identified as outcome variables, while crash characteristics, driver characteristics, temporal characteristics, vehicle characteristics, roadway characteristics, and environment characteristics are regarded as potential predictors influencing driver injury severities. Additionally, likelihood ratio tests and marginal effects are used to assess the temporal instability and impact location non-transferability of the explanatory variables. The results indicate an overall temporal and locational instability of model estimates while several determinants are identified to have consistent effects on injury severity outcomes such as animal-involved collisions, old drivers, safety restraint usage, female drivers, physically impaired drivers, and vehicles with insurance. This study also quantifies and characterizes the net effect of year-to-year and location-to-location shifts through probability differences between out-of-sample predictions and within-sample observations. Varying magnitudes and inconsistent directions of distribution characteristics (mean, skewness, kurtosis, and prediction accuracy) in the probability differences across different impact locations over time are captured. Moreover, this study indicates that the non-transferability of collision locations has a greater impact on the prediction accuracy than the temporal instability. The findings could potentially serve as a reference for transportation administrators to formulate effective safety strategies to better protect drivers from adverse-road-related crashes.

Real-time prediction of crash risk is an effective method for enhancing traffic safety, but it is not fully explored in freeway tunnels. A two-stage deep learning modeling framework comprising a preliminary exploration stage and a prediction and analysis stage is proposed for real-time crash risk prediction in freeway tunnels. A random parameters logit model with heterogeneity in means and variances is used in the preliminary exploration stage to investigate the unobserved heterogeneity and influence mechanism of precursors on real-time crash risk. In the prediction and analysis stage, a random deep and cross network model considering feature interactions and unobserved heterogeneities is developed to predict and analyze real-time crash risk, which is interpreted by the shapley additive explanations approach. The multi-source fusion dataset, collected from the Caltrans performance measurement system and the weather information website, is used to validate the proposed framework for exploring real-time crash risk in freeway tunnels. Results reveal that: (1) the random parameters logit model with heterogeneity in means and variances outperforms the traditional logit model in terms of the model fitting, providing a reference for deep learning modeling that may be able to improve model performance by addressing heterogeneity; (2) the important crash precursors such as the average difference in speed between detectors of tunnel entrance and exit are discovered based on the marginal effect analysis of the random parameters logit model with heterogeneity in means and variances; (3) the random deep and cross network model yields the best prediction performance compared to its counterparts (some other data-driven models), demonstrating the superior performance of deep learning models for real-time risk prediction tasks. It also indicates that considering feature interaction and heterogeneity in deep learning modeling can improve prediction performance; and (4) the important precursors found in the random deep and cross network model using the shapley additive explanations approach are close to those discovered in the statistical model, indicating that the proposed deep learning model can capture the similar effects of precursors as the statistical models, and the precursor interactions and heterogeneities also can be observed by the shapley additive explanations approach.

Using traffic conflict-based extreme value theory (EVT) models to quantify real-time crash-risk of road facilities is a promising direction for developing proactive traffic safety management strategies. Existing EVT real-time crash-risk analysis studies have only focused on using block maxima models. This study proposes a dynamic Bayesian hierarchical peak over threshold modeling approach to estimate real-time crash-risk based on traffic conflicts. The proposed approach combines quantile regression, dynamic updating approach, Bayesian hierarchical structure, and the peak over threshold method to generate time-varying generalized Pareto distributions to derive real-time crash-risk measures (i.e., crash probability and return level). The derived real-time crash-risk measures are applied to estimate cycle-level crash-risk at three signalized intersections in Surrey, British Columbia. Five approaches are used to dynamically update the model parameters, including time trend model, generalized autoregressive conditional heteroskedasticity process approach, as well as the first-order, second-order, and third-order dynamic linear models. For comparison, static models are also developed. All the developed models are compared in terms of statistical fit and predictive performance. Based on the best fitted dynamic model, cycle-level crash probability and return level are calculated to measure signalized intersection safety at cycle level. The results show that dynamic models considerably outperform static models in terms of statistical fit and predictive performance. Further, the third-order dynamic model has the best performance, which is probably due to that the model incorporates two linear trends to respectively describe the variation of the coefficients as well as its change to better account for the variation in the effect of time-varying covariates. However, it should be noted that the third-order dynamic model development needs more computation time than other dynamic models, which may limit the application of the model.

This study presents a new safety measure derived from field theory. It evaluates the risk arising from the various concurrent conflicts within a platoon that can occur on high-speed highway driving situations, such as car-following, yielding, and lane changing. We defined the risk field as a finite scalar field produced by traveling vehicles on the road, and we defined the conflict field as the overlapping risk field between any vehicles in proximity on the roadway. The study used a probabilistic trajectory prediction model to construct risk fields and an approximation method to reduce the computational time for real-time applications. To demonstrate the applicability of the proposed new measure, we applied it to real-world trajectory data (NGSIM data from US Highway 101). We compared the results with three traditional conflict-based safety measures: post-encroachment time (PET), modified time-to-collision (MTTC), and deceleration rate to avoid a crash (DRAC). The new measure produced seamless and continuous risk estimations even during time windows when the other measures could not estimate the risk between vehicles. This is a major advantage over traditional measures. The study also developed visual displays of the estimated conflict fields to provide safety analysts with an intuitive and fast understanding of the results of the safety assessments made using the conflict field measure. We conclude that the proposed new safety measure provides a robust, reliable, and improved assessment of the risk involved in expected future mixed-traffic environments that involve both human-driven vehicles and automated vehicles in the future.

With the recent advancements in computer vision and artificial intelligence, traffic conflicts occurring at an intersection and associated traffic characteristics can be obtained at the granular level of a signal cycle in real-time. This capability enables the estimation of the real-time crash risk using sophisticated modelling techniques, e.g., extreme value theory. However, these models are inherently incapable of forecasting the crash risk of future time periods based on the temporal dependency of crash risks. This study proposes a unified framework of extreme value theory and autoregressive integrated moving average models for forecasting crash risks at signalised intersections. At the first level of this framework, a non-stationary generalised extreme value model has been developed to estimate the real-time rear-end crash risk at the signal cycle level using the video data collected from three signalised intersections in Queensland, Australia. To capture the time-varying effect of different traffic conditions on conflict extremes, traffic flow, speed, shockwave area, and platoon ratio covariates are incorporated into the generalised extreme value model. The signal cycle-level crash risks obtained from the first level form a univariate time series, which is modelled using two variants of autoregressive integrated moving average model to forecast the crash risk of future signal cycles. Results reveal that the autoregressive integrated moving average model with exogenous variables outperforms the model without exogenous variables and can forecast the crash risk for the next 30–35 min with reasonable accuracy. Similarly, results also demonstrate that different crash risk patterns within a typical day are accurately predicted. The proposed framework helps identify the spatiotemporal windows where safety gradually deteriorates over time, thus enabling proactive safety assessment.

Due to visibility limitations and other factors, the injuries sustained by bicyclists in nighttime vehicle-bicycle crashes tend to be more severe than daytime crashes. This paper seeks to provide insights into this day/night injury severity phenomenon by studying how day/night bicyclist injury severities have changed in crashes that occurred before, during, and after the COVID-19 lock downs. Using data from vehicle-bicycle crashes in the state of Florida over a three-year period (from 2019 to 2021 inclusive), separate yearly models of bicyclist-injury severities (with possible outcomes of severe injury, minor injury, and no visible injury) were estimated using a random parameters logit approach with possible heterogeneity in the means and variances of random parameters. Likelihood ratio tests were conducted to examine the overall stability of model estimates across the studied years as well as day/night differences, and a comparison of partially constrained and unconstrained temporal modeling approaches was undertaken. A wide range of variables potentially affecting resulting bicyclist injury severities in vehicle/bicycle crashes was considered including bicyclist and vehicle driver information, vehicle features, roadways and environmental conditions, temporal characteristics, and roadway features. The findings show statistically significant injury-severity differences between daytime and nighttime before, during and after the COVID-19 pandemic. Out-of-sample simulation results suggest that improving the visibility of bicyclist through mandated reflectivity, improved roadway illumination, undertaking public awareness campaigns relating to nighttime bicyclist safety, and vulnerable road user detection sensors in vehicles can all contribute to substantially improving nighttime bicyclist safety.