| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 14.28 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Os locais onde decorrem as ações de construção, manutenção e reabilitação necessárias
à conservação dos activos rodoviários, designados de zonas em obras, são pontos sensíveis
da rede em termos de sinistralidade, os quais exigem a necessária atenção por parte das
agências de transporte em todo mundo. Estes locais são considerados perigosos tanto
para os trabalhadores como para o tráfego que os atravessa, já que envolvem o
encerramento de vias e desvios do tráfego, variação dos limites de velocidade,
congestionamentos do trânsito e distracções.
Tendo em conta as particularidades destas zonas, o trabalho proposto tem como
objectivo identificar os principais fatores com impacto na gravidade e na natureza de
acidentes ocorridos nas zonas em obras, bem como suas consequências.
Para atingir esse objectivo será aplicada a técnica estatística de regressão logística
multinomial, aos dados de natureza mista oficiais da sinistralidade rodoviária nacional
(Portugal - Continental) relativos ao período compreendido entre os anos de 2010 a 2015.
Os modelos desenvolvidos, natureza do acidente e tipo de envolvimento, apresentam
globalmente alguns fatores contribuintes predominantes como o ambiente urbano,
traçado em planta (reta), boas condições de aderência, envolvimento de pesados,
conduzir em marcha normal e velocidade excessiva para as condições existente.
A identificação destes fatores possibilita a programação das intervenções na rede em
ambientes mais seguros, minimizando assim a probabilidade de ocorrência destes
acidentes e por consequência, o número de vítimas envolvidas. Esse tipo de análises
evidenciam as lacunas críticas existentes no conhecimento das questões de segurança nas
zonas em obras, bem como os desafios da abordagem associada à modelação deste tipo
de dados.
The places where the maintenance and rehabilitation actions necessary for the conservation of the road assets, designated as works zones, are network sensitive points in terms of accident rates. In this sense, work zones require special attention from transport agencies worldwide. These locations are dangerous to both workers and the traffic, as they involve the closing of traffic lanes and diversions, variation of speed limits, traffic congestion and distractions. Considering the particularities of these zones, the proposed work aims to identify the main factors; that have an impact on the severity and type of accidents happening in work zones. In order to achieve this goal, the multinomial logistic regression statistical technique will be applied to the mainland Portuguese official road accidents mixed data from 2010 to 2015. The developed models of, crash and involvement type, present some predominant contributing factors such as the road environment (urban area), alignment (straight), good grip conditions, heavy vehicle involvement, the driver action (running straightly) and excessive speed for existing conditions. The identification of these factors allows the programming of interventions in the network in order to promote safe environments, minimizing the probability of occurrence of this type of accidents. Analyzes such as this highlight the critical gaps in works zones accident knowledge, as well as, the challenges associated with the modeling approach adopted.
The places where the maintenance and rehabilitation actions necessary for the conservation of the road assets, designated as works zones, are network sensitive points in terms of accident rates. In this sense, work zones require special attention from transport agencies worldwide. These locations are dangerous to both workers and the traffic, as they involve the closing of traffic lanes and diversions, variation of speed limits, traffic congestion and distractions. Considering the particularities of these zones, the proposed work aims to identify the main factors; that have an impact on the severity and type of accidents happening in work zones. In order to achieve this goal, the multinomial logistic regression statistical technique will be applied to the mainland Portuguese official road accidents mixed data from 2010 to 2015. The developed models of, crash and involvement type, present some predominant contributing factors such as the road environment (urban area), alignment (straight), good grip conditions, heavy vehicle involvement, the driver action (running straightly) and excessive speed for existing conditions. The identification of these factors allows the programming of interventions in the network in order to promote safe environments, minimizing the probability of occurrence of this type of accidents. Analyzes such as this highlight the critical gaps in works zones accident knowledge, as well as, the challenges associated with the modeling approach adopted.
Description
Keywords
Regressão Logística Multinomial Sinistralidade Rodoviária Zonas Em Obras