Deterministic construction heuristics for the time-dependent travelling salesman problem
The time-dependent travelling salesman problem (TD-TSP) extends the TSP by considering the travel time between two locations as time-dependent. As the TD-TSP is NP-hard, it is difficult to find an optimal solution for large-sized instances. In this case, construction heuristics provide fast and easy to implement solutions. Such solutions can be used in real-world situations where optimality is not necessary. Furthermore, they can be used for improvement heuristics as initial solutions. Although construction heuristics are commonly used for the TSP, only few of them are adapted to the TD-TSP. This thesis adapts and evaluates five of the most prominent construction algorithms of the TSP to the TD-TSP: Cheapest Insertion, Christofides, First Fit, Nearest Neighbour and Savings. An attempt was made to create a new benchmark derived from taxi rides in New York City. In order to compare the performances of the considered heuristics, numerical experiments are performed on one artificial and two real-life based time-dependent benchmarks. The results of the comparison are validated on TSP instances from the literature. The outcomes show that the modified Savings algorithm outperforms all other approaches significantly.