Research On UAV Forest Fire Detection System With Particle Swarm Genetic Algorithm

As a tool with high mobility and low human cost,UAVs have been widely used in forest fire prevention field for forest fire detection,but there are low detection efficiency and low intelligence in the early stage of forest fires,and people often cannot find the fire in time.In this thesis,a forest fire information model with realistic simulation of forest fire spread and smoke diffusion is constructed,and the inertia weight linear reduction,fusion algorithm improvement and forest wind feature optimization are optimized in the UAV cluster detection algorithm,and an improved particle swarm genetic algorithm with fast iteration,population diversity and forest wind feature is proposed to improve the search of forest fire by UAV cluster.The improved particle swarm genetic algorithm with fast iteration,population diversity and forest wind characteristics is proposed to improve the search efficiency and stability of forest fires by UAV cluster.The experiments show that the UAV cluster with the improved particle swarm genetic algorithm has better searchability and convergence,and can effectively shorten the search time for forest fires.The main research elements of this thesis are the following three points.（1）Simulation modeling of forest fire information model.Two types of models based on meta-cellular automata for forest fire spread simulation system and Gaussian smoke plume model for fire smoke dispersion simulation system are constructed to simulate the spread of forest fire and smoke dispersion in real scenarios,which provides an experimental environment and theoretical basis for the UAV dispatching system.（2）An improved particle swarm genetic algorithm is designed for the UAV monitoring and scheduling system.This study mainly uses smoke information as the target signal detection function,improves the inertia weight reduction mechanism of the particle swarm optimization algorithm,and enhances the search speed and accuracy;combines the genetic algorithm with strong local search ability and improves the hybridization method and mutation method to improve the local search ability of the particle swarm optimization algorithm;improves the algorithm speed iteration formula,adds forest wind characteristics,so that the UAV cluster has the characteristics of tending to the fire source.The research conducted comparison experiments on the three algorithm improvement schemes of the particle swarm optimization algorithm in the linear reduction of inertia weights,fusion algorithm and forest wind characteristics,and proved the effectiveness of the algorithm improvement schemes.Finally,the improved particle swarm genetic algorithm is proposed and effectively used in the UAV scheduling system for forest fire detection.（3）Experimental research on UAV cluster fire point search simulation was conducted.Based on the simulation of forest fire information model,the cluster distribution,search time and error distance of UAV clusters applying each algorithm in forest fire point search were calculated separately in this study.It is proved that the average distance between the fire point identified by the improved particle swarm genetic algorithm and the nearest actual fire point is 97.15 m.At the same time,the improved particle swarm genetic algorithm improves the search speed by 91.34%,340.89% and 52.21% compared with the particle swarm optimization algorithm based on linear decreasing weight,artificial fish swarm algorithm and particle swarm algorithm based on dynamic adaptive inertia weight,respectively.of the search speed,which can be effectively applied in the forest fire detection task.This study provides a new application for UAV cluster application in forest firefighting,and proves the effectiveness and efficiency of the improved particle swarm genetic algorithm UAV forest fire detection system through experiments,which provides effective support for forest fire detection and can effectively prevent the spread of forest fires.