On the robust shortest path problem
Computers and Operations Research
Optimal paths in graphs with stochastic or multidimensional weights
Communications of the ACM
Communications of the ACM
A Mean-Variance Model for Route Guidance in Advanced Traveler Information Systems
Transportation Science
Some experiments in machine learning using vector evaluated genetic algorithms (artificial intelligence, optimization, adaptation, pattern recognition)
Optimal routing in time-varying, stochastic networks: algorithms and implementations
Optimal routing in time-varying, stochastic networks: algorithms and implementations
Genetic algorithms and simulation applied to optimization: the stochastic shortest path model
Genetic algorithms and simulation applied to optimization: the stochastic shortest path model
Muiltiobjective optimization using nondominated sorting in genetic algorithms
Evolutionary Computation
Shortest paths in stochastic networks with correlated link costs
Computers & Mathematics with Applications
A fast and elitist multiobjective genetic algorithm: NSGA-II
IEEE Transactions on Evolutionary Computation
A genetic algorithm for shortest path routing problem and the sizing of populations
IEEE Transactions on Evolutionary Computation
Hi-index | 12.05 |
In this study, we propose a new simulation-based multi-objective genetic algorithm (SMOGA) approach to find a portfolio of reliable nondominant (Pareto) paths, a set of paths that is equally good or better at least in one objective space compared to all other paths, in stochastic networks while considering link travel time uncertainties and correlations among link travel times. Our SMOGA model consists of a Monte Carlo simulation, a genetic algorithm, and a Pareto filter module to find a set of Pareto paths that minimize the travel time budgets required to satisfy multiple requirements of travel time reliability pre-determined by users. For our purposes, an alpha (and beta) reliable path finding problem is first formulated as a variant of Chance Constrained Multi-objective Programming (CCMOP) model. Then the simulation module is used to simulate stochastic networks with correlations among link travel times, and genetic algorithm and Pareto filter module are used to effectively search for Pareto paths that satisfy multiple reliability requirements in combinatorial solution space. Numerical results on the Chicago Sketch network demonstrate that our carefully designed genetic representation (a variable-length chromosome and two ways of generating initial population) and genetic operators (a crossover and a mutation operator) effectively explore solution space and ensure the feasibility and diversity of offspring paths. Further, our graphical representations of Pareto paths on the same network indicate that simplified models that do not consider correlations among link travel time distributions may find Pareto paths with a significant bias in travel time budgets and hence provide travelers sub-optimal paths.