Low-Level Hybridization of Scatter Search and Particle Filter for Dynamic TSP Solving

This work presents the application of the Scatter Search Particle Filter (SSPF) algorithm to solve the Dynamic Travelling Salesman Problem (DTSP). SSPF combines sequential estimation and combinatorial optimization methods to efficiently address dynamic optimization problems. SSPF obtains high quality solutions at each time step by taking advantage of the best solutions obtained in the previous ones. To demonstrate the performance of the proposed algorithm, we conduct experiments using two different benchmarks. The first one was generated for us and contains instances sized 25, 50, 75 and 100-cities and the second one are dynamic versions of TSPLIB benchmarks. Experimental results have shown that the performance of SSPF for the DTSP is significantly better than other population based metaheuristics, such as Evolutionary Algorithms or Scatter Search. Our proposal appreciably reduces the execution time without affecting the quality of the obtained results. 13.