Forest fire modeling and analysis based on K-means clustering algorithm and time series forecasting

Abstract

Since September 2019, forest fire incidents have broken out in several states in southeastern Australia, forming hundreds of fire points. Many fire incidents are rampant, smoky, and harmful. The theme of this paper is how to use specific drones to quickly monitor and eliminate forest fire incidents in Victoria. First, the K-means clustering algorithm is used to divide the fire point dataset into 4 partitions. Next, we model a multi-objective programming algorithm and solve the shortest distance for the drone to traverse all fire sample points in each partition by the ant colony algorithm, and get the lowest cost purchase plan for SSA drones and Radio Repeater drones. And we model a time series forecasting algorithm for predicting the number of fire incidents in the next 10 years and obtain the increasing cost of purchasing drones each year correspondingly.