載貨汽車(chē)的車(chē)身結(jié)構(gòu)設(shè)計(jì)【帶開(kāi)題報(bào)告+文獻(xiàn)綜述】【QX系列】
載貨汽車(chē)的車(chē)身結(jié)構(gòu)設(shè)計(jì)【帶開(kāi)題報(bào)告+文獻(xiàn)綜述】【QX系列】,帶開(kāi)題報(bào)告+文獻(xiàn)綜述,QX系列,載貨,汽車(chē),車(chē)身,結(jié)構(gòu)設(shè)計(jì),開(kāi)題,報(bào)告,講演,呈文,文獻(xiàn),綜述,qx,系列
附錄一：外文原文
A Comprehensive Study of Single and Multiple Truck Crashes Using Violation and Crash Data
Abstract
Around 4,000 people died in crashes involving trucks in 2016 alone in the U.S., with 21 percent of these fatalities involving only single-unit trucks. Many studies have identified the underlying factors for truck crashes. However, few studies detected the factors unique to single and multiple crashes, and none have examined these underlying factors in conjunction with violation data. The current research assessed all of these factors using two approaches to improve truck safety. An injury/fatal crash was defined as a crash that results in an injury or fatality. The first approach investigated the contributory factors that increased the odds of injury/fatal single truck and multiple vehicle crashes with involvement of at least one truck. The literature has indicated that previous violations can be used to predict future violations and crashes. Therefore, the second approach used violations related to driver actions that could result in truck crashes. The analysis for the first approach indicated that driving on dry-roadway surfaces, driver distraction, and rollover/jackknife types of truck crashes, speed compliance failure, and higher posted speed limits are some of the factors that increased the odds of injury/fatal single and multiple vehicle crashes. With the second approach, the violations related to risky driver actions, which were underlying causes of truck crashes, were identified and analyses were run to identify the groups at increased risk of truck involved crashes. The results of violations indicated that being nonresident, driving off peak hours, and driving on weekends could increase the risk of truck involved crashes.
1. INTRODUCTION
Trucks are a crucial part of the United States economy. Trucks transport 80% of all freight in the
U.S. annually, which accounts for over $700 billion worth of goods [1]. The trucking industry in the U.S. moves about 10.5 billion tons annually, which is expected to increase to 27 billion tons by 2040 [2]. Moreover, seven million people, including more than three million drivers, are employed through this industry. However, truck crashes place a huge burden on the nation in terms of death, injury, and lost productivity. According to the Federal Motor Carrier Safety
24
Administration (FMCSA), there were 667 truck occupant deaths (driver and passenger), and of those 667 deaths, 398 deaths occurred in single-vehicle crashes [3].
Wyoming has the highest fatality rate (24.7 death per 100,000 population) in the nation [4]. Wyoming also has the highest truck crash rate in the United States [5]. These high truck crash and fatality rates result from the high amount of through truck traffic on Wyoming interstates, adverse weather conditions, and mountainous geometric conditions.
However, truck crashes can be mitigated by improving truck safety through policies and regulations, which enhance the performance of the trucking industry without compromising safety. Various countermeasures have been taken in the United States organized into 4 E’s of safety. The 4 E’s include enforcement, education, engineering, and emergency response. Enforcement is one of the 4 E’s that can improve traffic safety. The performance of Wyoming highway patrol (WHP), and consequently road safety, could be improved by identification of the factors that increase the odds of future violations, and consequently future crashes [6, 7]. Thus, this study incorporates violation data, in addition to crash data, to identify the contributory factors to the violations that are likely to increase the odds of future crashes. Identification of these factors can help the WHP to put more emphasis on the contributory factors of risky violations resulting in traffic safety.
Truck crashes are complex events. They can involve single vehicles or two or more vehicles. Out of 700 truck occupant deaths that occur every year in the U.S., about 60% occur in single-vehicle truck crashes [8]. For each type of event, different contributory factors may play roles. Previous research indicated that there are significant differences between single and multiple vehicle crashes [9, 10]. Therefore, this study analyzed single truck and multiple vehicle crashes, with truck involvement, separately. This study investigated factors impacting different types of truck crashes by including vehicle, driver, and environmental factors. In addition, this study included violation data to identify the groups at higher risk of truck crashes by including only the violations contributing to truck crashes in this state. For the purpose of this study, a truck is defined as a commercial vehicle with gross vehicle weight rating greater than 10,000 pounds.
2. BACKGROUND
Based on FMCSA, the critical reasons for large truck crashes can be assigned to driver (87%), non-performance (12%), recognition (28%), decision (38%), performance (9%), and vehicle (10%) [11]. Lemp et al. (2011) used the ordered probit model to investigate the impact of vehicle, occupant, driver, and environmental characteristics on crash severity for those involved in heavy-duty truck crashes [12]. The results indicated that the odds of fatalities increase with the number of trailers and fall as the truck gross vehicle weight rating decreases. Khattak et al. (2003) used crash data in North Carolina during 1996-1998 to investigate the impact of truck rollovers and occupant injuries in single-vehicle crashes [8]. The results indicated that higher risk factors in single-truck-crashes include risky driving, speeding, alcohol and drug use, traffic control violations, truck exposure to dangerous road geometry, and trucks that transport hazardous materials.
25
Moomen et al. (2018) investigated the influential factors of downgrade truck crashes in Wyoming using logistics regression. They found that driver gender, speed compliance, weather, lighting and road condition, shoulder and lane width, number of sag and crest curves, roadway grade and length are the contributory factors to truck related crashes [13]. Schneider et al. (2009) developed multinomial logit models to investigate driver injury severity resulting from single-vehicle crashes [14]. Different driver, vehicle, and environmental characteristics were found to increase injury severity. Being female, older, unbuckled, fatigued, and under the influence led to increase in the odds of injury. Zhu and Srinivasan (2011) investigated the factors impacting the injury severity in truck crashes [15]. Truck driver distraction, alcohol use, and emotional factors of car drivers were associated with higher severity crashes. Pahukula et al. (2015) investigated the contributory factors to injury severity of truck crashes using data from Texas during 2006 to 2010 [16]. The results indicated that different time periods in a day have different contributing impacts on truck crash severity.
However, in the majority of the studies, researchers mostly looked at the injury severity of both multiple-vehicle and single truck crashes as a whole. Thus, they did not identify the variables unique to single and multiple crashes. Zou et al. (2017) carried out a study in New York City to investigate the differences between single-vehicle and multiple-vehicle truck crashes [17]. The results indicated that there are substantial differences between factors affecting single and multiple truck crashes. Thus, this study examined truck crash severity separately for single-vehicle and multiple-vehicle truck crashes.
Many studies have identified correlations between previous violations and future crash risk. A previous study carried out by Li and Baker (1994) indicated that conviction records can be used to identify groups with greater odds of involvement in fatal crashes [18]. Similarly, Elliott (2001) investigated the ability of previous violations to predict future offenses and crashes [19]. The results indicated that the drivers with previous ticketed offenses are at greater risk for future crashes. Rezapour et al. (2017) used violation data, in addition to crash data, to assess unsafe driver actions to reduce crashes [6]. Chen et al. (1995) carried out a study by examining driver records to investigate the relationship between crashes and past records of crashes and convictions [20]. The authors found a positive correlation between pre-period crashes per driver and pre-period number of convictions. In this study, failure to yield and disobeying traffic signals were two violations that best predict crashes. Lantz and Loftus (2006) carried out a study to use an analytical model for predicting future crash involvement based on the history of driver information and also identifying effective enforcement actions that can predict driver behavior and future crash involvement [21]. The results indicated that reckless driving and improper turn violations are the violations that have the highest increase in likelihood of a future crash. Also, failure to keep proper lane was some of the convictions with the highest likelihood of a future crash. A study by Terrill et al. (2016) investigated the impact of traffic citations on the number of crashes on an interstate in Wyoming [22]. The results indicated that an increased number of citations issued is a preventive measure for the number of crashes.
However, none of the aforementioned studies used violation or conviction data to investigate groups of truck drivers with an increased risk of being involved in truck crashes. On the basis of
26
the discussed studies, violations can be used as an indication of the groups that are at greater risk of being involved in future crashes.
This current study was set forward to fulfill two main objectives:
1. Conduct crash analysis to determine the factors impacting injury single truck and multiple vehicle, truck involved, crashes. In order to determine these factors, two analyses were carried out:
1.1 Injury/fatal single truck crash analysis.
1.2 Injury/fatal multiple vehicle crash analysis with involvement of at least one truck.
2. Conduct violations analysis to identify the groups who are more likely to violate the laws that are the main causes of single and multiple truck crashes. Two analyses were carried out to fulfill this objective:
2.1 Analysis of those types of violations associated with single truck crashes.
2.2 Analysis of those types of violations associated with multiple vehicle crashes involving at least one truck.
A crash in this study is one that results in an injury or fatality. Due to the low number of fatality crashes, these crashes were aggregated with injury crashes.
3. METHODS
Logistic regression is used in many studies involving binary crash outcomes [23-27]. For the logistic regression model, the binary response variable Y is assumed to have a Bernoulli distribution with probability π [28].
(1)
where x is a vector of explanatory variables and is a vector of unknown regression coefficients.
Equation (1) can be solved for π which gives
(2)
For modeling truck crash severity, the binary response variable was 1 for injury/fatal and 0 for a property damage only (PDO) truck involved crashes. The response is conditioned on a crash that has occurred, and then looking at its binary classification (fatal/injury (F+I) or PDO). Separate models were developed for single truck crashes and for multiple vehicle, truck involved, crashes. The probability (π) of either a single or multiple truck crash being injury was modeled using
various risk factors as explanatory variables .
Logistic regression was also used for analyzing violations. The purpose of these analyses was to identify drivers who are more at risk of committing particular traffic law violations, which can
27
lead to truck involved crashes. Here, the response (Y) had the value 1 if a driver received a citation of a particular type and a 0 if a driver did not receive a citation of a particular type. The response is conditioned on drivers who had received a violation and then looking at its binary classification for the citation type (received or did not receive citation of particular type). Two different citation analyses were considered. The first analysis involved only those violations more commonly observed with single-truck crashes. The second analysis involved those violations most commonly observed in multiple-vehicle crashes involving at least one truck. The probability (π) of a driver receiving a citation of a particular type was examined in relation to the explanatory
variables involving driver characteristics such as gender and residency, and temporal characteristics such as time of day and day of the week.
Stepwise model selection was used to select explanatory variables for a final logistic regression model. A significance level of 0.10 was pre-specified for entering the model and a significance level of 0.05 was pre-specified for staying in the model. All analyses were performed using the Statistical Analysis System (SAS) [29].
4. DATA PREPARATION
The data was combined from the three interstates in Wyoming, I-80, I-25, and I-90, with the highest truck related crash rates. Crash data was obtained from the Wyoming Department of Transportation (WYDOT) using the Criticial Analysis Reporting Environment (CARE) from 2011 to 2014. This study used various variables, which can be categorized under driver, environmental, vehicle, temporal, crash, and driver behaviors. Driver characteristics included age, gender, residency, violation (conviction) record, and speed limit compliance at the time of crash. Environmental characteristics included weather and roadway-surface conditions. Weight of a truck was categorized as a vehicle characteristic. Day of week and time of crash were organized under temporal characteristics. Roadway characteristics included posted speed limit of a location where a crash occurred. Driver actions at the time of crash, number of vehicles, and pre-collision vehicle action were categorized under crash characteristics. Driver distraction, driver under influence (DUI) suspicion, fatigue, and the use of safety technology were categorized under driver behaviors. In this study, distraction is defined as any type of distraction such as TV, cell pager, or wireless communication inside the cabin at the time of crashes. Truck crash analyses were divided into two parts: Single truck and multiple vehicles, truck involved, crashes. Single truck crashes were investigated separately as more than 50% of all the truck crashes were single truck crashes.
The violation data was obtained from the Wyoming court reported violation database from 2011 to 2014. For single truck crash analysis, truck drivers were at fault in the crashes. Therefore, the violation data for this analysis was filtered to include just truck driver violations to identify groups that are more at risk of single truck crashes. There were 121,680 violations filtered to 17,239 truck violations. However, all violations were used for investigating the groups that were at higher risk of multiple vehicles crashes, involving at least one truck. This is due to the fact that both truck and no truck drivers could be at fault in these crashes. Only violation types: follow too closely, failure to drive within single lane, and speed too fast for conditions that resulted in truck
28
crashes were included in this study (Table 1). Violations related to the main causes of truck crashes were identified among 800 types of violations and presented in Table 1. For instance, only one violation type: driving too fast for conditions was identified as a contributing factor to a crash type: “drove too fast for conditions”.
5. RESULTS
5.1. Descriptive Analysis, Crash Data
Fig. (1) presents general characteristics of truck crashes on Wyoming interstates. As can be seen from Fig. (1a), most of the truck crashes (52%) involved just a single truck. Including both single (52%) and multiple vehicle, truck involved crashes (26%), about 78% of the truck drivers were at fault for truck crashes on Wyoming interstates. Driver actions with highest percentage, for both single and multiple truck related crashes, are included in Figs. (1b and c).As can be seen from Figs. (1b and c), the main causes of truck crashes, driver actions, include failing to keep proper lane, driving too fast for conditions, and following too close. Based on Figs. (1b and c), no improper driving, failure to keep proper lane, following too close account for 78% of multiple vehicle, truck involved, crashes and 65% of single truck crashes. That is the reason why the violations related to these driver actions are included in the violation analyses (Table 1). It should be noted that “no improper driving” action is a crash in which a driver had no improper driving, but was involved in a crash. Therefore, no violation was identified related to this driver action and this driver action was excluded from violation analyses.
For the first analysis, data were filtered from the original file to include only single truck crashes. Summary statistics of significant variables that impacted the severity of single truck crashes are presented in Table 2. As can be seen from the table, most of the single truck crashes (85%) involved property damage only. Most of the drivers in single truck crashes were male (94%) compared with only 6% involving female drivers. Speed limit of 65 mi/hr was chosen as a threshold for the speed limit variable. This is because most speed limits on the included highways in Wyoming are greater than 70 mi/hr and 65 mi/hr was found to be the best threshold that can divide the crashes into similar categories. Although the majority of single truck crashes (71%) occurred at a posted speed limit of greater than 65 mi/hr, a rather large proportion (29%) occurred at a lower speed limit. Most of the single truck crashes (67%) occurred on not-dry-road conditions. Most single truck crashes (72%) were rollover or jackknife. In 16% of all the single truck crash cases, truck drivers had some type of distraction in the cabin. Distraction was defined as any distraction in a cabin such as wireless communication or TV.
For the second analysis, the data was filtered from the original file to include crashes involving at
29
least one truck. Due to the involvement of at least two vehicles, only the summary statistics of a vehicle at fault is presented in Table 2. Seven percent of truck crashes occurred while the driver at fault did not follow the posted speed limit. To provide more insight about this variable,
Table 2 also includes more statistics on the circumstance in which speed limit compliance was not fulfilled. In these crashes, 62% of the at-fault vehicles did not follow the posted speed limit while driving on not-dry-road conditions. Not-dry-road conditions include the road conditions other than dry, such as rainy/snowy. About 29% of truck crashes occurred at the locations with a speed limit of less than 65 mi/hr. More detailed summary is also provided for this variable in Table 2.
The results indicated that most of the lower speed crashes occurred when the road was not dry (56%), which might be an indication that the locations were equipped with Variable Speed Limits (VSL). Forty four percent of the lower speed limits were also related to dry-road condition, which might be due to driving through work zones.
5.2. Statistical Modeling, Crash Data
5.2.1. Factors Associated with Higher Risk of Injury Single Truck Crashes
This first modeling approach investigated the variables that increase the odds of fatal/injury truck crashes compared to PDO truck crashes. Table 3 shows the variables included in the full model and the estimates for those variables remain
收藏