Application of Minimum Spanning Tree in the Solution of Forest Fire Intelligent Drone Deployment

Abstract

Abstract In recent years, devastating wildfires have occurred in all states of Australia, and the wildfires in eastern Victoria have been extremely affected. In order to prevent the disaster from happening again, we built a model to warn the wildfires in Victoria. First, we extracted the Victorian fire data from NASA’s official website. After sorting by credibility and flame temperature, we performed grid division and labeling based on latitude and longitude and the number of fires per unit time. Then, we used the k-means method to cluster the locations of wildfires, and obtained three clusters. Second, we divide wildfires into two categories, wildfires near the cluster center and others not. Among them, 3 regions are near the cluster center, and 4 region is away from the cluster center. For wildfires near the cluster center, remote sensing detection drones are used to fully cover the fire, and use the minimum spanning tree algorithm to solve the position of the repeater drone. For other wildfires, we establish a nonlinear programming equation to find the combination of drones with the lowest rate for the alarm missed by the fire, and get the number of telemetering drones is 65, the number of repeater drones is 49. Third, we map the data of different distances in the three-dimensional topographic map to the two-dimensional plane according to the geographic data of Victoria, and use the Dijkstra algorithm to obtain the approximate distance between the two points in the given point set. Then use the minimum spanning tree to solve, get the number of telemetering drones is 65, the number of repeater drones is 52, it is still close to the solution of model 1 after adding the terrain, indicating that the optimization result is successful.