反饋式的循序旅運需求預測之收斂性探討

The Convergence of the Travel Demand Forecasting Process with Feedback

10.29774/UT.200812.0002

Boyce, D.E.；陳惠國

循序性旅運預測程序 ； 常數權重法 ； 連續平均法 ； 簡單反饋法 ； sequential forecasting procedure ； constant weight ； method of successive averages ； naive feedback

都市交通

23卷2期（2008 / 12 / 01）

13 - 24

繁體中文

都市運輸規劃中四階段（含旅次產生、旅次分配、運具選擇、與路徑選擇在內），其循序性旅運預測程序必須引進反饋式的求解過程，才可能使得子問題之間的輸入與輸出旅運成本數據能夠符合一致性，從而達到合理的收斂精度。本文針對三種不同擁擠水準的交通情境進行測試，比較三種反饋式的方法發現：常數權重法優於連續平均法，而連續平均法又優於簡單反饋法。當常數權重法之權重值設定在0.2~0.5之間，執行5次反饋迴圈就可以使得總錯置流量(TMF)除以所有起點與迄點之總流量之比率達到小於1%的收斂精度，而其中又以權重值設定在0.25時，所獲得的最終解之收斂性最佳。

The four-step sequential travel demand forecasting process, consisting of trip generation, trip distribution, modal split and traffic assignment, has widely been used in transportation planning for many decades. With feedback, the travel demand forecast can have better chance to converge to a preset level of accuracy. This paper compares three different feedback strategies, i.e., constant weight, successive averages and naive feedback, for three different levels of traffic conditions. The results show that while the constant weight strategy outperforms, the successive averages strategy and the naïve feedback strategy are sequentially ranked as the second and third. It is also observed that for the constant weight strategy in the range of 0.2~0.5, the ratio of total misplaced flow to total origin-destination flow can fall within 1% in five iterations. The constant weight 0.25 is highly recommended because it consistently performed better than the other constant weights.

1. Bar-Gera, H.,Boyce, D.(2003).Origin-based algorithms for combined travel forecasting models.Transportation Research, Part B,37,405-422.
2. Bar-Gera, H.,Boyce, D.(2006).Solving the Sequential Procedure with Feedback.presented at the Sixth International Conference of Chinese Transportation Professionals,Dalian, China:
3. Bar-Gera, H.,Boyce, D.(2006).Solving a nonconvex combined travel forecasting model by the method of successive averages with constant step sizes.Transportation Research, Part B,40,351-367.
4. Bothner, P. ,Lutter, W.(1982).Ein direktes Verfahren zur Verkehrsumlegung nach dem 1. Prinzip von Wardrop.FB Verkehrssysteme:Universitaet Bremen.
5. Boyce, D.,Lupa, M.,Zhang, Y.(1994).Introducing ‘feedback’ into four-step travel forecasting procedure vs. equilibrium solution of combined model.Transportation Research Record,1443,65-74.
6. Boyce, D.,Xiong, C.(2007).Forecasting travel for very large cities: challenges and opportunities for China.Transportmetrica,3,1-19.
7. Carroll, Jr., J. D.,Bevis, H. W.(1957).Predicting local travel in urban regions.Papers and Proceedings, The Regional Science Association,3,183-197.
8. Comsis Corporation(1996).Incorporating Feedback in Travel Forecasting: Methods, Pitfalls and Common Concerns.Final Report, DOT-T-96-14, Federal Highway Administration.
9. PTV AG(2007).VISUM 9.5 Manual.Karlsruhe, Germany:
10. Schittenhelm, H.(1990).On the integration of an effective assignment algorithm with path and path-flow management in a combined trip distribution and traffic assignment algorithm.18th PTRC Summer Annual Meeting,Sussex, England:

1. 陳惠國、陳文婷(2012)。運輸需求預測整合模型統合架構之探討。運輸學刊，24(4)，467-496。