混合格位傳遞模式之基因模糊邏輯號誌控制

Abstract

在現今台灣都會地區內，隨著工商業活動的頻繁、所得的提高促使汽機車持有率不斷的遽升，然而在原有的都市道路幾何設計之下，道路難以再增加其道路容量，故此現象所造成的車輛擁擠、停等延滯等，將使得駕駛人所花費的時間成本難以估計，除此之外，油耗、噪音…等環境污染問題更日益嚴重。因此，交通管理策略之一的交通號誌控制，在此狀況下就顯的非常重要。一個有效率的交通號誌控制系統，不僅可以解決因道路容量不足而形成的壅塞，更可以減少油料的浪費、二氧化碳氣體的排放，甚至提升道路交通安全。 交通號誌控制理論面上大致可分為離線（off-line）控制與線上（on-line）控制，離線控制主要是事先經由調查員調查各時段之車流資料特性，然後經由號誌時制軟體運算求得適當之號誌時制並放入控制器內加以執行運作。惟此控制方式難以因應瞬息萬變之車流量變動並即時更改時制計畫，其中定時號誌控制（pretimed signal control）就屬此類；至於線上控制方式也就是考慮動態的觀念，利用偵測器（detector）即時的自動傳回各種車流資料，再經由軟體運算後得到一新時制計畫，然後迅速回傳予路口控制器加以執行，因此可以即時反映交通車流狀況。 觸動式號誌控制、動態號誌控制以及適應性號誌控制均屬於線上控制的範疇，其中，適應性號誌控制由於具有彈性、適用性及最佳化等特性，因此，在近年來被廣泛使用。一些較著名的適應性號誌控制如SCATS、SCOOT及OPAC等，然上述模式均以數學式來研定其控制變數之門檻值，並據以作為控制邏輯之核心，但由於交通狀況充滿著不確定性，倘若以明確值來訂定門檻，恐怕會造成控制績效的不彰。基此，近年來有一些研究透過模糊邏輯控制器(fuzzy logic controller, FLC)來改善此問題。透過偵測器所收集之交通資料，可以應用FLC來決定號誌時相及時制計畫。然在FLC控制系統中，雖然推論引擎及解模糊化方法都有一貫的理論依據可遵循，但對於邏輯規則及隸屬函數必須由訪談專家加以主觀設定，導致其應用性大受影響。故近年來漸有相關研究利用人工智慧方法，例如基因演算法（Genetic algorithms, GAs）之樣本學習方法建構FLC，以克服主觀設定偏頗之問題。然而，透過基因演算法同時或連續學習邏輯規則及調整隸屬需要耗費相當多的時間以及產生部分不合理的學習結果。 為了避免上述的缺點，本研究根據反覆GFLC (Chiou and Lan, 2005)，提出一逐步迴歸的模糊邏輯控制器(Stepwise Genetic Fuzzy Logic Controller, SGFLC)，來學習邏輯規則及調整隸屬函數。逐步演算法選擇邏輯規則及隸屬函數之概念與逐步迴歸相似，在既有規則庫中，一次僅選擇一個能使適合度值改善最大的規則，直到入選的規則均無法改善適合度值則停止，則規則庫內之所有規則即為最佳規則。 為了要發展一套以GFLC為基礎的號誌控制系統，一個有效率且能正確描述車流行為的模擬軟體是必要的。有部分研究透過微觀模擬軟體為模擬平台，並藉以訂定一最佳化的號誌控制策略，然而這樣的微觀模擬軟體並不適合用來評估基因演算法的模式訓練。基此，本研究採用中觀範疇的混合格位傳遞模式(MCTM)，除了可以有效率的學習SGFLC外，更能正確的描述都市地區道路車輛混合之交通行為。 本研究考慮汽車及機車在綠燈時段之交通量(TF)及紅燈時段之等候長度(QL)為狀態變數，綠燈延長時間為控制變數，總車輛延滯(TVD)為評估指標，研擬一SGFLC最佳號誌控制方式。為了證明本研究提出模式之績效，在獨立路口部分，比較2種定時號誌及3種適應性號誌控制，結果顯示SGFLC模式之績效最好，此外，當交通流量變化較大時，SGFLC模式之控制績效也較其他模式為佳。在連續路口部分，不論在何種連鎖策略下，SGFLC之控制績效也比其他模式好，證明本研究所提出之模式具有效率、強健及可應用之特性。 另外，在路廊的號誌控制部分，眾所皆知的是當連鎖的路口越多，其控制績效就會受到影響，因此，有部分研究試圖透過分群方式，來獲得最佳之連鎖號誌控制的數目，而此分群的概念對於如何將本研究所構建之模式擴展應用到整個路網的控制是非常重要的，因此在一個廊道中，究竟有多少路口應該被連鎖，也是另一個值得被研究的議題。本研究結合SGFLC與GAs方法，針對一長路廊應連鎖的路口數進行分群，實驗結果顯示，本研究提出之混合模式確實可以有效增加路廊總通過車輛數。

On-line traffic signal control typically feeds the real-time traffic data, collected by the sensors, into a build-in controller to produce the timing plans. Thus, it can provide signal-timing plans in response to real-time traffic conditions. Because of its flexibility, applicability and optimality, adaptive signal control tends to be the mainstream of signal controls nowadays. The well-known adaptive signal controllers employ mathematical equations or models to determine “crisp” threshold values as the cores of control mechanism; thus, the control performance could be negatively affected by the uncertainty of traffic conditions. Since a fuzzy control system has excellent performance in data mapping as well as in treating ambiguous or vague judgment, many works have employed fuzzy set theory to develop fuzzy logic controllers (FLC). In FLC systems, both inference engine and defuzzification have been consistently used in previous literature; however, methods for formulating the rule base (logic rules) and data base (membership functions) are subjectively preset, not optimally solved. Employing GAs to construct an FLC system with learning process from examples, hereafter termed as genetic fuzzy logic controller (GFLC), can not only avoid the bias caused by subjective settings of logic rules or membership functions but also greatly enhance the control performance. However, to simultaneously or sequentially learn of logic rules and membership functions may require a rather lengthy chromosome and large search space, resulting into poor performance, a long convergence time and unreasonable learning results (i.e. conflicting or redundant logic rules, irrational shapes of membership functions). To avoid abovementioned shortcomings, based on the iterative GFLC (Chiou and Lan, 2005), this study proposes a stepwise genetic fuzzy logic controller (SGFLC) to learn both logic rules and membership functions. At each learning process, the proposed algorithm selects one logic rule which can best contribute to the overall performance controlled by previously selected logic rules combined with this selected rule. Such a selection procedure will be repeated until no other rule can ever improve the control performance. Therefore, the incumbent combination of logic rules is the near optimal learning results. In order to develop a SGFLC-based signal control requires an efficient traffic simulation model to replicate traffic behaviors and to determine the performance of the control logic. Many studies use microscopic traffic simulation software to simulate the urban signal control and implement the optimized signal policy. However, such simulation software is rather time consuming, making it better for evaluating the control performance for a given signal control model but not suitable for the evolution of genetic generations for model training. For the learning efficiency of SGFLC and the capability in capturing traffic behaviors of Asian urban streets where mixed traffic of cars and motorcycles are prevailing, the mixed traffic cell transmission model (MCTM) is introduced to replicate the traffic behaviors. This study considers traffic flows and queue lengths of cars and motorcycles as the state variables and extension of green time as the control variable, towards the minimization of total vehicle delays. To investigate the control performance of the proposed SGFLC model, comparisons of two pre-timed timing plans and three adaptive signal timing models are conducted at an isolated intersection. Results show our proposed SGFLC model performs the best. Moreover, as traffic flows vary more noticeably, the SGFLC model performs even better than any other models. In the case of a 3-intersection arterial under four coordinated signal systems i.e., simultaneous, progressive, alternate and independent, both experimental example and field case study show that the proposed SGFLC model can perform better than any adaptive control models, suggesting that the proposed SGFLC signal control model is efficient, robust and applicable. Moreover, it is well-known that the control performance of signal coordination would be greatly degraded as the number of coordinated intersections increases. Thus, this study also combines SGFLC with GAs for optimizing the number of coordinated intersections along a long corridor. The experimental example shown that the proposed hybrid model can increase total throughput along the corridor through an optimal coordinated intersections.