Genetic Algorithms Applied to the Ride Matching Problem with Time Windows.
Ridematching, Ridesharing, Genetic Algorithms, NSGA-II.
This disertation presents optimization strategies for Ridematching Problem with Time Windows (RMPTW). The Ridematching Problem with Time Windows is an extension of Pickup and Delivery Problem with Time Windows (PDPTW). Which is associated with vehicle routing area. The interest in the subject is by its direct application of model solutions to real world problems. For instance allocation rides in previously known vehicles routes. Genetic algorithms based on the NSGA-II are developed. Two variations of the algorithm are compared with the state of the art. A new deterministic rider insertion method is proposed in a partial route, which avoids the use of mutation operator in the service time of the route points. Experimental tests show that the proposed insertion algorithm is faster and produces better results than the state of the art algorithm. Results are discussed and comparisons are presented using multi-objective quality indicators. Further tests are applied to evaluate statistical significance of the results obtained.