靜態號誌設計─以新竹科學園區路網為例

Abstract

新竹科學園區近年成為我國科技重鎮，發展迅速，導致交通需求量大增，當年規劃之道路規模至今已達飽和狀態，欲解決新竹科學園區尖峰時段交通擁擠的問題，可於園區興建新的道路，以增加道路之容量。然而園區內部之大部分土地用途均已確立，若欲在園區內興建新道路，必須從現有土地中將非道路使用土地更改為道路使用土地，此舉可能須拆除地上建物，成本甚鉅。因此在興建新路段成本過於高昂的情況之下，號誌之有效的規劃是較能改善路網中旅行成本的方法之ㄧ。 就路網規劃者而言，以整體使用者之旅行成本總和最小為目的，因此在設計路網號誌時，必須以總成本考量。然而道路之使用者在使用道路時會選擇對自己較有利之路徑，根據Wardrop之使用者均衡定律，使用者選擇旅行成本最短之路徑。因此規劃者在設計最佳化號誌時必須考慮使用者追求使用者均衡的行為，才能真正有效降低系統之成本。 本研究收集路段資料，如：路段長度、路旁干擾度高低與車道數等，求得路段之績效函數，並描述規劃者與道路使用者之雙層模式，以敏感性分析基礎之最陡下降演算法(GSDA)求解該模式，由得出之雙層模式之系統最佳解，並與現況號誌基礎下之使用者均衡解進行比較，期能有效降低系統之總旅行成本。

HsinChu Science Park has become the scientific and technological center in recent years. More traffic demand occurs due to the expansion of the facility, and leads to saturated road capacity and congestion. One alternative to improve the situation is to build new roads, but it may influence configuration of the existing buildings in the park. The other alternative, better and cheaper, is to design signal system efficiently. Network planners purpose to minimize the system cost, but we suppose that users adapt to the Wardrop’s first principle: the road users will minimize his own travel cost. The conflict between UE and SO makes the total travel cost higher. We describe this problem as a bilevel model. The upper is network planner whose objective function is to minimize total system cost. The lower is road users which stand to UE. It is a condition in this model. To solve the bilevel problem, we need to know the reactions between the roads users. It means that we need to calculate the sensitivity information to solve the bilevel network problem. We collect the attributes of the roads in the research range, such as road length, roadside interference, lane number, and etc., to derive the link performance function and calculate the coefficients in this study. Then formulate the network bilevel problem, and solved by the Gradient Steepest Descent Algorithm. Finally, we compare the result with the current signal.