| 摘要: | 國道高速公路為陸路運輸動脈，交通事故為主要壅塞原因之一，若提升事故處理績效，能減少延誤與降低衍生事故發生風險。故本研究在不增加警力等資源條件下，以警車排程最適化規劃，改善勤務編排與派遣方式，縮短事故到場時間，加速事故處理，達成交通安全與順暢等目的。 本研究先以95年國道1號及3號事故，作巨觀角度統計分析並對其事故特性、派遣作業及勤務狀況作相關、單因子變異數分析。得除到場時間與事故嚴重性無關、勤務班次數與事故次數有明顯相關外，餘者未發現清晰關聯；此外，發現事故處理時各階段所耗用時間等，可作為後續研究參數。 續以微觀角度剖析警車到場影響因素，得勤務編排、派遣及事故特性等均為影響因素，且亦會相互影響，難以單一方案有效處理。故以二階段處理較為允當，先建構勤務編排模式，後發展調整派遣模擬，解決警車排程問題。 第一階段勤務編排模式，目的為編排足夠勤務班次，以因應事故隨機發生處理。當模式求解後，其結果得編排勤務時數及班表。第二階段勤務派遣模擬，可將前得勤務班表，並配合所擬研動、靜2類型中各4種勤務派遣方案，併同輸入並進行模擬。此外，此模擬亦能依設定分配，產生事故模擬資料，以測試勤務模擬績效。而模擬績效以到場時間、支援班次、加班時數及效益等為指標。 最後，以泰山分隊作為實例分析，經模擬後相關指標綜合排序，得信賴水準為99%時勤務班表派遣模擬績效最佳及依事故率、依事故次數派遣方案二者併列最優。此外，亦發現靜態勤務不易受勤務編排影響及到場時間、支援班次、加班時數正相關係數甚高，顯示三者具有共同趨勢。本研究所建構模式與模擬亦可運用於高速公路其他分隊。 Freeways form the primary traffic corridor in Taiwan, and traffic accidents are a major cause of traffic congestion. By increasing the speed of the processes for handling traffic accidents, unexpected delays and the risk of derivational accidents can be reduced. To shorten response time and improve the efficiency of accident processing, this study schedules and allocates police patrols using an optimization method to enhance traffic safety and order. First, data of all accidents occurring on National Highway No. 1 and No. 3 in 2006 were collected. The accident characteristics, accident responses, and shift scheduling were analyzed using correlation coefficients and ANOVA. Following these analyses, this study determined the following two conclusions: (1) No relationship exists between the response time and accident severity; and (2) a significant and positive quantity relationship exists between shift scheduling and accidents. Additionally, every stage of the accident processing procedure was identified. These data will form the parameters of future research. Second, through further research, the response time was found to be influenced by three factors, as follows: the accident characteristics, accident response, and shift scheduling, which interact with each other. Achieving optimal scheduling and dispatching under complex factors using only one method is difficult. Therefore, this study employs a two-stage method to solve this problem. The first stage involves creating a chance-constrained optimization model for scheduling by using the minimal patrol units to meet the demands of accident handling. The second stage involves creating a dispatch simulation, which can simulate the operation of a police car under current conditions and evaluate the effectiveness. Additionally, eight patrol patterns, including moving and stationary patterns, are also inputted in the simulation. Indices are established for response time, extra supportive patrol units, extra work hours, and utility. Third, the Taishan platoon was chosen as the case study. The simulation results led to two conclusions: (1) the optimal patrol schedule occurs at a confidence level of 99 %; and (2) the optimal patrol patterns are achieved by c based on accident frequency and rate. These values are evaluated using a composite index that integrates response time, extra supportive patrol units, extra work hours. Additionally, the correlation coefficients among above three indices are positive and near 1. These results also imply that they follow the same trend. The results of this study can provide a useful reference for other traffic enforcement agencies affected by similar conditions. |
