An Expert Authentication System for the Two-Vehicle Crash Accident
碩士 === 逢甲大學 === 交通工程與管理所 === 92 === When a traffic accident happens, usually arguments on accident responsibility also accrued. In order to clarify the responsibility and settle the arguments, accident authentication committees are authorized to review these cases and conclude the allocation of resp...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Published: |
2004
|
Online Access: | http://ndltd.ncl.edu.tw/handle/89764157123524849716 |
id |
ndltd-TW-092FCU05118029 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-092FCU051180292015-10-13T13:01:03Z http://ndltd.ncl.edu.tw/handle/89764157123524849716 An Expert Authentication System for the Two-Vehicle Crash Accident 兩車碰撞事故之肇事鑑定專家系統 Shou-Chieh Fang 方守潔 碩士 逢甲大學 交通工程與管理所 92 When a traffic accident happens, usually arguments on accident responsibility also accrued. In order to clarify the responsibility and settle the arguments, accident authentication committees are authorized to review these cases and conclude the allocation of responsibility. However, in comparing with over 15,000 accident authentication cases in a year in Taiwan, there are rather few experts in reviewing these cases. Therefore, in order to ease the burden of these experts and utilize their historical authentication cases, an efficiency expert system for accident authentication is worthy of developing. Besides, because highly professional knowledge and experience which are accumulated from long term training is needed for conducting accident authentication, it is also important for the system to effectively educate junior reviewers with these knowledge and experience. Based on that, this study employs artificial neural network (ANN) to develop an expert system for two-vehicle crash accident authentication. A total of 538 two-vehicle crash accident cases, i.e. 1,076 drivers involved, from 2000 to 2002 are selected, which have same judgment between local committee and reviewing committee for accident authentication. These cases are randomly divided into to two sets: 70% for training and 30% for validating. The input variables for ANN are selected by using contingent table analysis and stepwise discriminate analysis from a total of 24 accident variables, such as area, driver gender, driver age, vehicle type, right of the way, etc. The output variable is set as, of course, the degree of responsibility that the driver is judged to take. Then, different network structures and settings of parameters are tested and analyzed for proposing a comprising model. For the sake of comparison, a statistical discriminating analysis model is also calibrated. The results show that the ANN model can achieve 77.19% and 72.67% of correctness rate in training and validating, respectively. The correctness rates of discrimination analysis model are only 61.84% in training and 58.39% in validating. Obviously, it indicates that the ANN model is more suitable to be the expert system of accident authentication. Moreover, in order to measure the influence of each input variable on judging the accident responsibility, an index named as general influence index (GI) is computed based on the ANN trained weights. The most influential variable is highway type, with GI=0.227, followed by right of way (GI=0.200), degree of drinking (GI=0.189), relative driving direction (GI=0.097), types of vehicles (GI=0.056), types of road (GI=0.048), degree of speeding (GI=0.047). It is also in accordance with the prior knowledge in accident authentication. none 邱裕鈞 2004 學位論文 ; thesis 127 zh-TW |
collection |
NDLTD |
language |
zh-TW |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 逢甲大學 === 交通工程與管理所 === 92 === When a traffic accident happens, usually arguments on accident responsibility also accrued. In order to clarify the responsibility and settle the arguments, accident authentication committees are authorized to review these cases and conclude the allocation of responsibility. However, in comparing with over 15,000 accident authentication cases in a year in Taiwan, there are rather few experts in reviewing these cases. Therefore, in order to ease the burden of these experts and utilize their historical authentication cases, an efficiency expert system for accident authentication is worthy of developing. Besides, because highly professional knowledge and experience which are accumulated from long term training is needed for conducting accident authentication, it is also important for the system to effectively educate junior reviewers with these knowledge and experience.
Based on that, this study employs artificial neural network (ANN) to develop an expert system for two-vehicle crash accident authentication. A total of 538 two-vehicle crash accident cases, i.e. 1,076 drivers involved, from 2000 to 2002 are selected, which have same judgment between local committee and reviewing committee for accident authentication. These cases are randomly divided into to two sets: 70% for training and 30% for validating. The input variables for ANN are selected by using contingent table analysis and stepwise discriminate analysis from a total of 24 accident variables, such as area, driver gender, driver age, vehicle type, right of the way, etc. The output variable is set as, of course, the degree of responsibility that the driver is judged to take. Then, different network structures and settings of parameters are tested and analyzed for proposing a comprising model. For the sake of comparison, a statistical discriminating analysis model is also calibrated.
The results show that the ANN model can achieve 77.19% and 72.67% of correctness rate in training and validating, respectively. The correctness rates of discrimination analysis model are only 61.84% in training and 58.39% in validating. Obviously, it indicates that the ANN model is more suitable to be the expert system of accident authentication. Moreover, in order to measure the influence of each input variable on judging the accident responsibility, an index named as general influence index (GI) is computed based on the ANN trained weights. The most influential variable is highway type, with GI=0.227, followed by right of way (GI=0.200), degree of drinking (GI=0.189), relative driving direction (GI=0.097), types of vehicles (GI=0.056), types of road (GI=0.048), degree of speeding (GI=0.047). It is also in accordance with the prior knowledge in accident authentication.
|
author2 |
none |
author_facet |
none Shou-Chieh Fang 方守潔 |
author |
Shou-Chieh Fang 方守潔 |
spellingShingle |
Shou-Chieh Fang 方守潔 An Expert Authentication System for the Two-Vehicle Crash Accident |
author_sort |
Shou-Chieh Fang |
title |
An Expert Authentication System for the Two-Vehicle Crash Accident |
title_short |
An Expert Authentication System for the Two-Vehicle Crash Accident |
title_full |
An Expert Authentication System for the Two-Vehicle Crash Accident |
title_fullStr |
An Expert Authentication System for the Two-Vehicle Crash Accident |
title_full_unstemmed |
An Expert Authentication System for the Two-Vehicle Crash Accident |
title_sort |
expert authentication system for the two-vehicle crash accident |
publishDate |
2004 |
url |
http://ndltd.ncl.edu.tw/handle/89764157123524849716 |
work_keys_str_mv |
AT shouchiehfang anexpertauthenticationsystemforthetwovehiclecrashaccident AT fāngshǒujié anexpertauthenticationsystemforthetwovehiclecrashaccident AT shouchiehfang liǎngchēpèngzhuàngshìgùzhīzhàoshìjiàndìngzhuānjiāxìtǒng AT fāngshǒujié liǎngchēpèngzhuàngshìgùzhīzhàoshìjiàndìngzhuānjiāxìtǒng AT shouchiehfang expertauthenticationsystemforthetwovehiclecrashaccident AT fāngshǒujié expertauthenticationsystemforthetwovehiclecrashaccident |
_version_ |
1716870103246569472 |