Boosted Genetic Algorithm Using Machine Learning for Traffic Control Optimization
Abstract : Traffic control optimization is a challenging task for various traffic centers around the world and the majority of existing approaches focus only on developing adaptive methods under normal (recurrent) traffic conditions. Optimizing the control plans when severe incidents occur still remains an open problem, especially when a high number of lanes or entire intersections are affected. This paper aims at tackling this problem and presents a novel methodology for optimizing the traffic signal timings in signalized urban intersections, under non-recurrent traffic incidents. With the purpose of producing fast and reliable decisions, we combine the fast running Machine Learning (ML) algorithms and the reliable Genetic Algorithms (GA) into a single optimization framework. As a benchmark, we first start with deploying a typical GA algorithm by considering the phase duration as the decision variable and the objective function to minimize the total travel time in the network. We fine tune the GA for crossover, mutation, fitness calculation and obtain the optimal parameters. Secondly, we train various machine learning regression models to predict the total travel time of the studied traffic network, and select the best performing regressor which we further hyper-tune to find the optimal training parameters. Lastly, we propose a new algorithm BGA-ML combining the GA algorithm and the extreme-gradient decision-tree, which is the best performing regressor, together in a single optimization framework. Comparison and results show that the new BGA-ML is much faster than the original GA algorithm and can be successfully applied under non-recurrent incident conditions.
? The traffic intersection is used to illustrate problems where optimal scheduling for drive order is needed, thus the results produced by the genetic algorithm depend on the present situation in the intersection.
? The control of the intersection is assumed to be directly from this hardware implementation, and being totally independent from any outside control system.
? The traffic intersection, which is not a real existing intersection, is independent from adjacent intersections.
? All simulation programs acts as controller programs for the intersection.
The proposed model is applied in the response and clearance phases. To simplify the case study, we assume that the incident was previously detected, verified and the duration of the incident clearance was predicted. In addition, the severity of the incident is also reported as an indication of the number of lanes affected. Last but not least, the incident affected area is determined using previous studies. Recently, Pan et al. [29] studied the spatial-temporal impact of traffic incidents based on archived data using advanced sensors and came up with the incident impacted area and the delay occurrence prediction in a road network.
? It uses a number of artificial individuals looking through a complex search space by using functions of selection, crossover and mutation.
? The purpose to use GA is searching and finding optimal or good enough solution.
? This solution will hide in a big search space to look through. Is no guaranty to find any exact solutions when using a GA.
? Some result can even be far from optimal when GA gets stuck in so called local optimum in the search space.
The biggest advantage of this approach is reducing the time that genetic algorithms spend in creating the initial and subsequent populations, and to learn from previous iterations in the past which were the best choices that meet the optimization criteria, instead of always starting from random and new combinations which need intensive simulations to be run multiple times.
|