提升國道事件應變效率:事件派遣資料與處理時間分析

Exploring incident response data on freeways : analysis of incident dispatch and processing time

陳薇亘(Wei-Huan Chen)

指導教授 ： 許聿廷

國立臺灣大學/工學院/土木工程學研究所/碩士(2019年)

https://doi.org/10.6342/NTU201902297

摘要

交通壅塞是影響高速公路服務水準的一大主因，而壅塞相當主要的一部分原因是在於非預期的事件，包含車輛事故、散落物或動物闖入等，因此需要透過有效率的事件應變系統，降低事件持續時間，以減少事件對於車流的影響程度。各國高速公路管理單位積極研究、開發事件應變系統，而臺灣地區國道系統亦持續精進事件應變效率，並提出應變車隊之動態派遣系統，透過考量即時的路況資訊以及車隊位置，以最佳化模型決策各應變車隊的調度派遣，以降低總應變時間。然而，在事件集中發生的狀況下，模式中還涉及事件處理的排程，而每一事件的處理時間與下一事件的應變時間皆有相關，會進一步影響整體事件應變、處理的效率。因此，本研究之主要目標在於分析、估計事件的處理時間，進而反饋至派遣系統，以進一步提升應變效率。本研究利用模糊理論建立模糊邏輯模型，釐清事件處理時間的程序以及特性，針對事故類、散落物類和動物闖入三類事件，分別建立事件處理時間估計之模型。模型除考量事件類別之外，也納入車流狀況和事件發生地點之時空特性等因素，使模型能夠更準確估計事件的處理時間。本研究以臺灣北區高速公路路網上歷史事件資料為研究案例，分析模糊邏輯模型預測結果與實際處理時間的差異，並與存活分析和排序性常態機率模型進行結果比較。從三種類型事件的分析結果中可以發現模糊邏輯模型預測準確度相對高於其他統計模型，而針對長處理時間之事件也提供較良好的預測能力。本研究所建立處理時間估計的模糊邏輯模型，未來可納入動態車隊派遣系統，以更行提升更調度策略和事件應變之效率，並能夠提供相關即時資訊予用路人參考。

關鍵字

高速公路事件應變 ； 應變車隊派遣系統 ； 事件處理時間 ； 模糊邏輯模型

並列摘要

Traffic congestion can lead to significant impact on decreasing the service level on freeways, and one of the major causes for traffic congestion is unexpected incidents, including car accidents, scattered objects, and intrusion of animals. Therefore, it is necessary to develop efficient incident response system to decrease incident duration and associated impact. Research efforts have been made to develop dynamic dispatch systems which considers real-time traffic conditions and the location of response teams, seeking to assign appropriate response teams so as to decrease total response time. However, when there are multiple incidents occurring temporarily closely, the assignment problem further involves the scheduling of incident responses, and the incident processing time can affect the next incident to be responded, which may overall influence incident response efficiency. Hence, this study aims at better estimating the processing time to enhance the efficiency of the dispatch systems. Rule-based fuzzy logic models are developed according to the fuzzy theory. This study identifies the procedures and the characteristic of estimating incident processing time and constructs the fuzzy logic models for three major types of freeway incidents, car accidents, scattered objects, and intrusion of animals, to estimate the associated processing time. In addition to considering the incident types, the fuzzy logic model also incorporates ambient traffic conditions and the spatiotemporal characteristics of incident occurrence to enable more accurate estimation of incident processing time. This study uses the historical incident records of Northern Taiwan Freeway Network for case study and analyzes the accuracy of the estimation results from the fuzzy logic models against the actual processing time. The fuzzy logic models are also compared with the hazard-based duration model and ordered probability model. It can be found that the fuzzy logic models result in the highest accuracy in contrast other statistical models for each incident type, and it also provides better performance in estimating longer processing time. The fuzzy logic models proposed by this research can be used to estimate incident processing time for the development of a dynamic dispatch system to determine more efficient response strategies.

並列關鍵字

freeway incident response ； dispatch system ； incident processing time ； fuzzy logic model

參考文獻

Zografos, K. G., Nathanail, T., & Michalopoulos, P. (1993). Analytical framework for minimizing freeway-incident response time. Journal of Transportation Engineering, 119(4), 535-549.

Khattak, A., Wang, X., & Zhang, H. (2009). Are incident durations and secondary incidents interdependent?. Transportation Research Record, 2099(1), 39-49.

Oh, C., Oh, J. S., Ritchie, S., & Chang, M. (2001, January). Real-time estimation of freeway accident likelihood. In 80th Annual Meeting of the Transportation Research Board, Washington, DC.

Kim, H., Kim, W., Chang, G. L., & Rochon, S. M. (2014). Design of Emergency Response System to Minimize Incident Impacts: Case Study for Maryland District 7 Network. Transportation Research Record, 2470(1), 65-77.

Zografos, K. G., Androutsopoulos, K. N., & Vasilakis, G. M. (2002). A real-time decision support system for roadway network incident response logistics. Transportation Research Part C: Emerging Technologies, 10(1), 1-18.