signalized stochastic equilibrium transportation networks

Determining the optimum signal settings in a road transportation network is an important issue for providing shorter travel time and lower fuel consumption. In the Stochastic EQuilibrium Network Design (SEQND) context, traffic signal setting problem has been widely formulated as an optimization problem which is addressed with both deterministic and heuristic approaches. While deterministic approaches such as gradient-based methods are preferred, they may not be effective since the problem contains several local optima and the decision space is highly convoluted. Recently, heuristic global search approaches such as Genetic Algorithms (GAs) are utilized to solve the SEQND problem which may be non-convex in nature. Although heuristic approaches are very effective at exploring the search space, they may require relatively long time to find the global optimum solution. Thus, a hybrid approach, which utilizes a local search method that fine-tunes the solution of the global search method, may provide more accurate results for SEQND problem. Thus, this study proposes "Hybrid Harmony Search and Hill Climbing with TRANSYT" (HSHCTRANS) model to solve the SEQND problem. In the HSHCTRANS model, meta-heuristic Harmony Search (HS) algorithm is employed as a global search method while the TRANSYT hill climbing routine is used for fine-tuning. It has been applied to an example signalized road network. The effectiveness of the hybrid HSHCTRANS over the HS and Genetic Algorithm (GA)-based models has been investigated in terms of network performance index (PI). Results showed that the hybrid HSHCTRANS model provided about 11% improvement when it is compared with GA-based model. (c) 2012 Elsevier Ltd. All rights reserved