明確型動態用路人均衡模型係假設用路人具有完整資訊，然而實際上用路人並不一定具有完整資訊，而機率型動態用路人均衡模型以機率分配的形式來處理用路人路徑選擇問題，使得模型更趨向一般化。 機率型動態用路人均衡模型依據機率分配的不同而分為羅吉特模型及普羅比模型。羅吉特模型因為假設方案之間具有獨立且不相關(IIA)的特性，而具有效求解的優點，但卻無法考慮用路人選擇方案間的相關性，而普羅比機率型動態用路人均衡模型，雖然缺乏求解效率，但可以有效解決羅吉特模型的爭議 由於普羅機率型動態用路人均衡模型中，必須考慮時空間的相關性，本研究接續馮君惠(1998)的研究，更加深入探討普羅比機率型動態用路人均衡模型，並試圖將不同的求解演算法應用至動態路徑選擇中，並以不同的測試例比較各演算法的求解效率，提供日後相關應用的參考建議。 The basic assumption with Wardrop’s principles is that the traffic information is perfect. However, this assumption is generally not correct because the traffic information is likely to be imperfect. It is realistic that stochastic dynamic user equilibrium model hypothesizes that the error term of an actual route travel time is randomly distributed. We assume the distribution is the Gumbel distribution which yield the logit model or Normal distribution which yield the probit model. The logit model hypothesizes the independent and irrelative between alternative and can solve the problem effectively. The probit model is realistic but it solve the problem ineffectively. We must consider the relation between time and space in stochastic dynamic user equilibrium model with probit model and develop different algorithms to solve the probit model. Finally, we suggest several indices to evaluste these algorithms and make conclusions to provide a reference for further improvement.
