Matheuristic approaches for multi-visit drone routing problem to prevent forest fires

Abstract

Forest fires draw more attention as the impact of elements that threaten nature increases, such as the thinning of the ozone layer, and global warming. The prevention of forest fires is extremely important for the protection of natural life, and the provision of a healthy world to future generations. Some of the methods used for the prevention of forest fires are observation towers, unmanned aerial vehicles, images taken from satellites, and detectors. Noticing the fire as soon as it starts and intervening in the fire prevents the fire from spreading and causing major and negative consequences. The degree of fire sensitivity of forest areas may vary depending on factors such as the climate of the region, topographic structure, humidity ratio, vegetation, tree species, and density. Observing regions with high fire sensitivity more frequently than regions with low fire sensitivity will prevent the spread of fires faster by detecting them. Different from the literature, in this study, the degree of sensitivity of fire-sensitive areas are taken into account. Visit frequency is determined according to the degree of fire sensitivity. Due to the complexity of the constraints, the mathematical model can not reach the optimal solution in a short time. To solve larger problem decomposition based matheuristic approaches are proposed. Matheuristic algorithms are compared using different parameters for samples of different sizes.