摘要：Ant colony optimization is a set of heuristic optimization techniques that
emulate real ant’s colony foraging behavior to find the shortest path between
its nest and a food source. Those techniques have been widely used in order to
solve least-cost-path problems based on vector’s representations. This study
presents an adaptation of the traditional Max-Min Ant System algorithm to solve
least-cost-path problems on the grid or raster structure, usually used in
Geographical Information Systems. The algorithm finds, generally, the optimal
path given a cost-of-passage surface in raster format, the path start and end
points and a function that defines the incremental cost-of-passages between two
neighboring cells. Five hypothetical tests with increasing complexity are made
aiming to assess the model performance, including two of optimal routes
identification for linear engineering structures, like canals or roads. Although
real cost functions were not used, the results were coherent and showed the
algorithm’s capabilities. The algorithm was able to find multiple solutions in a
problem with multiple optimal paths. In other tests the algorithm was also able
to identify complex paths that would define, for example, trajectories of
irrigation channels or roads in mountainous zones. The algorithm was programmed
in Visual Fortran language allowing partial results presentation on the computer
screen.
