An Autonomous Tracking Method Of Plume In The Three-dimensional Space Using A Four-axis Aircraft

In view of the problem that few people have studied the three-dimensional space smoke plume self-tracking and self-localization of smoke plume source,this paper proposes two methods: simulated annealing algorithm combined with fuzzy c-means clustering method and cuckoo search algorithm combined with fuzzy c-means method.The simulated annealing algorithm or the cuckoo search algorithm is used to generate the position information of the autonomous flight of the four-axis aircraft,which avoids the blindness of the concentration of the plume collected by the quadcopter;the position information generated and the corresponding plume concentration information at the position constitute the four-dimensional eigenvectors.The eigenvectors are clustered by fuzzy C-means clustering algorithm to obtain the three-dimensional smoke plume concentration distribution area,and provide data information for the simulated annealing algorithm or the cuckoo search algorithm iterative position information.This paper mainly studies four aspects: parameters optimization of FCM clustering algorithm,simulated annealing algorithm combined with FCM clustering algorithm,identification method of plume distribution and cuckoo search algorithm combined with FCM clustering algorithm.In order to simulate the four-axis vehicle to track smoke plume autonomously,the kinematics model of the four-axis vehicle is constructed.The relationship between the four rotors' speed and the total lift,pitch,roll and yaw moments of the four-axis vehicle is established.A simulation platform for the four-axis vehicle to track smoke plume is built under Simulink environment.In order to optimize the parameters of FCM clustering algorithm,two kinds of clustering evaluation indexes are adopted,and the parameters of FCM clustering algorithm are optimized through a number of experiments.In order to verify the rationality of the proposed method,a hardware acquisition module was designed.The plume source was simulated in the laboratory and a three-dimensional space coordinate system was established.380 four-dimensional eigenvector data consisting of x,y and Z coordinates and corresponding plume concentration were collected in the laboratory.Fifty of 380 four-dimensional eigenvector data were selected as the initial data points of FCM clustering algorithm.Five experiments were performed to verify the rationality of the proposed two methods.The results show that the proposed method can only When the 50-dimensional eigenvector data is used as the initial data,it can be autonomously located near the plume source,which provides a reference value for the three-dimensional space plume autonomous tracking and plume source location.