交通路網之適應模糊代理人控制系統

Abstract

交通路網是一個非線性的系統，因此以等待車輛的長度及燈號的Cycle Starting Time作為模糊控制的輸入參數，設計單一路口的模糊控制器，以紓解非線性路網的車流。由於所建立的控制採取快速切換號誌紓解尖峰時刻的車流，但此種方式不符合一般駕駛人的用路習慣；因此，本論文利用在不同Cycle Starting Time的條件下，給予不同的解模糊化切換值，設計適應模糊控制方法以解決此問題，相較於一般固定式的紅綠燈控制，使建成的單一路口延滯時間改善42.4%。 本研究進一步探討大型路網，由於其非線性程度與複雜度更勝於單一路口，本論文提出一個adaptive fuzzy multi-agent systems延伸adaptive fuzzy systems使用上的彈性，協調彼此交換控制信號的時間，以進一步改善交通流量，在本文之模擬結果中，進而在一個 交通路網的平均延滯時間也改善了45.7%。 VISSIM是一套應用於交通規劃的模擬軟體，利用它的使用者介面可以設定車輛型態、轉向比、車種組成等各種影響交通路網的因素。使用VisVAP的流程圖設計方式建立adaptive fuzzy agents的模糊規則；並以此控制器為基礎，而無須建構交通系統的數學模型，透過VISSIM與VisVAP內部函式建立adaptive fuzzy agents的溝通通道與協調機制，以驗證adaptive fuzzy agents於多重路口的控制成效。

In order to design fuzzy controllers to handle nonlinear traffic systems, the input parameters of the queuing length and the cycle starting time are adopted in the present simulation study. To deal with the traffic jam during the rush hours, the fuzzy controller is implemented and it results in more signal switching; however, it is not suitable for real traffic control. The increasing weighting of the threshold value for changing signals in the present fuzzy controller as an adaptive fuzzy controller to overcome the quick-switching problem. This fuzzy adaptive control method improves 42.7% in the delay time. Furthermore, because a large traffic network is more complicated than a single intersection, the adaptive fuzzy multi-agent system (MAS) is proposed to extend the applications of the adaptive fuzzy systems. By coordinating the control signal among neighboring intersections, the improvement of the delay time for the adaptive fuzzy agents in a traffic network is around 45%. VISSIM is a powerful and popular simulation software in traffic analysis. By applying the provided user interface, all vehicle types, turning ratio, traffic composition, etc., in the traffic network can be determined. The present adaptive fuzzy agents need to be designed by VisVAP which is a VAP programming language. VisVAP enhances the use of free-defined signal control logics by using the VAP language to offer a more comfortable tool for creating and editing the program logics as the flow charts. Moreover, the control efficiency in a multiple intersections traffic network can be verified by applying the VISSIM and the inner functions of VisVAP to build the communication channel and coordination mechanism for the adaptive fuzzy agents without constructing a mathematical model for traffic systems.