Traffic signal timing optimisation based on genetic algorithm approach, including drivers' routing

Department of Civil and Environmental Engineering, Imperial College, Exhibition Road, SW7 2BU London, UK
Transportation Research Part B Methodological (Impact Factor: 2.95). 05/2004; 38(4):329-342. DOI: 10.1016/S0191-2615(03)00015-8
Source: RePEc


The genetic algorithm approach to solve traffic signal control and traffic assignment problem is used to tackle the optimisation of signal timings with stochastic user equilibrium link flows. Signal timing is defined by the common network cycle time, the green time for each signal stage, and the offsets between the junctions. The system performance index is defined as the sum of a weighted linear combination of delay and number of stops per unit time for all traffic streams, which is evaluated by the traffic model of TRANSYT [User guide to TRANSYT, version 8, TRRL Report LR888, Transport and Road Research Laboratory, Crowthorne, 1980]. Stochastic user equilibrium assignment is formulated as an equivalent minimisation problem and solved by way of the Path Flow Estimator (PFE). The objective function adopted is the network performance index (PI) and its use for the Genetic Algorithm (GA) is the inversion of the network PI, called the fitness function. By integrating the genetic algorithms, traffic assignment and traffic control, the GATRANSPFE (Genetic Algorithm, TRANSYT and the PFE), solves the equilibrium network design problem. The performance of the GATRANSPFE is illustrated and compared with mutually consistent (MC) solution using numerical example. The computation results show that the GA approach is efficient and much simpler than previous heuristic algorithm. Furthermore, results from the test road network have shown that the values of the performance index were significantly improved relative to the MC.

231 Reads
  • Source
    • "Lee [10] applied Genetic Algorithm (GA) to individual signalized intersection. Ceylan & Bell [6] suggested GA approach to solve traffic signal control and traffic assignment problem for optimization of signal timings. ACO implementation of the problem has been tried by Basken [3][4] using a variant of ACO known as ACORSES where heuristics has been used to reduce the search space of the potential solution space. "
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents a Hybrid Microscopic Discrete Evolutionary Model to reduce the waiting time of vehicles at traffic signals within the urban transportation system. Signal optimization reduces the wait time of vehicles as well as the mobility within the traffic system. In effect, it helps to achieve green environment and reduces the fuel consumption. The hybrid model is framed on the basis of 2-layers: “traffic signals” and “stochastic user equilibrium” respectively. The upper layer optimizes the wait time at traffic signal using evolutionary computation techniques (ACO, GA and a Hybrid of ACO and GA) whereas the lower layer deals with the actual runtime scenario of the traffic network. A comparative analysis being performed over the aforementioned techniques shows that the Hybrid Microscopic Discrete Evolutionary Model performs better.
  • Source
    • "However, incorporating many binary variables into the optimization frameworks used to study large-scale urban networks (such as Improta and Cantarella (1984) and Lo (1999a,b)) results in complex mixed integer mathematical programs (MIMPs) that are difficult to solve exactly. Instead, heuristic optimization methods must be used (Foy et al., 1992; Chiu and Chand, 1993; Ceylan and Bell, 2004; Murat and Gedizlioglu, 2005), but these provide inexact and suboptimal solutions. Even when exact methods are available to solve MIMPs, they are typically time intensive and computational expensive. "
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper extends the continuum signalized intersection model exhaustively studied in Han et al. (2014) to more accurately account for three realistic complications: signal offsets, queue spillbacks, and complex signal phasing schemes. The model extensions are derived theoretically based on signal cycle, green split, and offset, and are shown to approximate well traffic operations at signalized intersections treated using the traditional (and more realistic) on-and-off model. We propose a generalized continuum signal model, which explicitly handles complex vehicle spillback patterns on signalized networks with provable error estimates. Under mild conditions, the errors are small and bounded by fixed values that do not grow with time. Overall, this represents a significant improvement over the original continuum model, which had errors that grew quickly with time in the presence of any queue spillbacks and for which errors were not explicitly derived for different offset cases. Thus, the new model is able to more accurately approximate traffic dynamics in large networks with multiple signals under more realistic conditions. We also qualitatively describe how this new model can be applied to several realistic intersection configurations that might be encountered in typical urban networks. These include intersections with multiple entry and exit links, complex signal phasing, all-red times, and the presence of dedicated turning lanes. Numerical tests of the models show remarkable consistency with the on-and-off model, as expected from the theory, with the added benefit of significant computational savings and higher signal control resolution when using the continuum model.
    Transportation Research Part B Methodological 05/2015; 77:213-239. DOI:10.1016/j.trb.2015.03.005 · 2.95 Impact Factor
  • Source
    • "Different types of multi-objective GA were also implemented by other researchers to optimize signal timings, even though the most popular one to date was NSGA. Ceylan and Bell (2004) defined the objective function as the sum of a weighted linear combination of delay and number of stops and by integrating the genetic algorithms, traffic assignment and traffic control solved the equilibrium network design problem simpler than with heuristic algorithms. Girianna and Benekohal (2004) formulated the model proposed as a dynamic optimization problem with the objective of maximizing the total number of vehicles released by the network and penalizing it by queue accumulation along the arterials and used genetic algorithms to find the near optimal signal timing. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Two-dimensional multi-objective optimizations have been used for decades for the problems in traffic engineering although only few times so far in the optimization of signal timings. While the other engineering and science disciplines have utilized visualization of 3-dimensional Pareto fronts in the optimization studies, we have not seen many of those concepts applied to traffic signal optimization problems. To bridge the gap in the existing knowledge this study presents a methodology where 3-dimensional Pareto Fronts of signal timings, which are expressed through mobility, (surrogate) safety, and environmental factors, are optimized by use of an evolutionary algorithm. The study uses a segment of 5 signalized intersections in West Valley City, Utah, to test signal timings which provide a balance between mobility, safety and environment. In addition, a set of previous developed signal timing scenarios, including some of the Connected Vehicle technologies such as GLOSA, were conducted to evaluate the quality of the 3-dimensional Pareto front solutions. The results show success of 3-dimensinal Pareto fronts moving towards optimality. The resulting signal timing plans do not show large differences between themselves but all improve on the signal timings from the field, significantly. The commonly used optimization of standard single-objective functions shows robust solutions. The new set of Connected Vehicle technologies also shows promising benefits, especially in the area of reducing inter-vehicular friction. The resulting timing plans from two optimization sets (constrained and unconstrained) show that environmental and safe signal timings coincide but somewhat contradict mobility. Further research is needed to apply similar concepts on a variety of networks and traffic conditions before generalizing findings.
    Transportation Research Part C Emerging Technologies 03/2015; 55. DOI:10.1016/j.trc.2015.03.013 · 2.82 Impact Factor
Show more