- 作者: 陳惠國; 王中允
- 作者服務機構: 中央大學土木工程學研究所
- 中文摘要: 本研究針對變動需求下之非對稱性交通量指派模型發展出兩個新的求解演算法,即:混合法(Hybrid method)與EHB法,此二演算法在路段旅行成本績效函數之Jacobian矩陣與負需求函數的反函數均為正定的前題假設下保證收斂。由於所發展的二個演算法在求解問題時所需的電腦記憶體較少,故可在一般個人電腦的環境下進行中、大型路網問題的測試與求解。本文除了對混合法及EHB法的求解過程做了詳盡的介紹外,並舉一簡例驗證,經與常用之對角化法(Diagonalization method)之收斂解進行比較,混合法及EHB法的正確性已無庸置疑。為了提高所提演算法之實用性,未來,將提高其運算績效與收斂速度並進行大型實際路網測試。
- 英文摘要: Two newly developed methods are described in this study for solution of the asymmetric trafficequilibrium problem with variable demand, i.e. the Hybrid and EHB methods. The convergence of thetwo methods is guaranteed under positive definite assumption for the Jacobian matrices of the cost functionand negative inverse of demand function. The primary advantage of these two methods is that only moderatememory is required. Thus, larger networks than those presented in previous literature can be tested ina PC environment. To demonstrate the correctness of the two methods, a numerical example is included.An intensive study is currently underway to speed up the convergence, and provide a comparison withother available solution algorithms.
- 中文關鍵字: traffic eduilibrium; asymmetric traffic assignment; monotonicity; Frank-Wolfe method; Evan's algorithm; Hybrid method; EHB method
- 英文關鍵字: --