MRDPGA: a multiple restart dynamic population genetic algorithm for scheduling road traffic

Abstract A genetic algorithm is a biologically inspired stochastic approach to finding solutions optimization problems. However, unlike its deterministic counterpart, it cannot guarantee globally optimal solution since may be trapped within local optimum of the search space. Most researchers have focused on proposing new techniques for various parameters algorithms. That mutation, crossover, or selection algorithm. This research proposes modification standard algorithm, which serve as framework that can integrate any these their contribution final The multiple restart dynamic population (MRDPGA) proposed in this was used training adaptive neuro-fuzzy inference system (ANFIS) scheduling road vehicular traffic flows. results ANFIS models based different clustering methods showed MRDPGA-based controller performed better with mean square error (MSE) 0.299 and root (RMSE) 0.547 phase; MSE 0.272 RMSE 0.521 testing phase. Using controllers flow scheduling, MRDPGA-trained terms average waiting time (AWT) minimization total arrived vehicles (TAV). best-performing achieved 50.40% AWT 21.44% TAV improvement. Analyzing using one-tailed t -test paired two-sample means MRDPGA had significant impact controllers. Particularly FCM controller, where ( p = 0.0038) 0.0003) at 95% confidence level. Thus, algorithms are recommended further assessment other problems ascertain performance those problem domains.