Research On Key Technologies Of UAV Group Cooperation And Intelligent Control

Recently,unmanned aerial vehicle(UAV)has been widely used in military,industrial and agricultural industries.With the combination of UAVs and the task scenarios more and more closely,some drawbacks of single UAV are gradually exposed,such as flight endurance,autonomous decision-making.Therefore,a multi UAV system has gradually become a research hotspot.At present,despite the rapid development of a multi UAV system,there are still many technical difficulties waiting to be solved for the UAV group collaboration.For example,how to develop a cooperative communication network for information sharing,to provide a stable and efficient communication foundation for situation collection and task command transmission;how to integrate environmental information and assess the environmental situation efficiently and provide situation information for collaborative tasks;how to select collaborative tasks based on situation information and arrange each UAV to complete specific tasks;how to perform intelligent control for the UAV mission execution to improve the task execution efficiency.To address these problems mentioned above,this thesis conducted relevant research works to achieve an efficient communication network,the situation assessment,collaborative mission planning,and intelligent control method in the Multi UAV system.The main research contents and innovations of this thesis are as follows:1)To achieve stable communication in the Multi UAV system with a weak connectivity network,this thesis investigates the UAV cooperative network architecture and related theories.Based on the fuzzy recognition and decision-making method,this thesis developed an adaptive cooperative communication method for the UAV system to improve the anti-damage ability of the UAV system.This thesis uses a fuzzy recognition to integrate the communication attribute of UAV nodes and evaluates the importance of node attribute in the process of data transmission.Analytical Hierarchy Process is used to assign a weight to each communication feature of the UAV node,and to determine membership functions of different levels for quantization vector.Then,the fuzzy relationship between the communication attributes and the nodes is transformed into a specific transmission priority value,to build a continuous and reliable UAV data sharing dynamic self-organizing network model.Finally,to improve the data transmission efficiency of the Multi UAV system in a complex network,an information forwarding mechanism based on fuzzy decision-making is proposed to optimize the data transmission between multiple UAVs.Experiments show that the proposed method can improve the stability of information sharing and the efficiency of communication.2)To meet the requirement of situation assessment,this thesis proposed a method of situation fusion and assessment based on a Fuzzy deep neural network for the Multi UAV system.Firstly,the global area is divided into the local situation area of each UAV according to the actual detection range of each UAV.Then,in the local situation area,the collected local scene data is quantified based on time series and used as the input for the improved deep neural network.Secondly,the adaptive momentum and elastic stochastic gradient descent are introduced into the training process of the neural network to improve the performance,and then the local situation assessment results of each UAV are obtained.Finally,the fuzzy logic is used to effectively fuse the local situation assessment results of each UAV,and the fuzzy logic relationship of a global situation assessment is built.The results of the global situation assessment are more suitable for uncertain mission scenarios.Experimental results show that this method can effectively improve the capability of Multi UAV system situation fusion and assessment.3)In the Multi UAV system,the basic group collaborative theories and hierarchical decision-making methods are investigated.To address the problems of low efficiency of the conventional group decision-making model,a hierarchical mission planning method based on the Fuzzy ant colony algorithm for the Multi UAV system is proposed by combining the collaborative mission allocation and path planning.Firstly,based on the clustering algorithm,the initial target groups are clustered into new target groups and target sub-groups.The number of UAVs and the cost of collaborative task execution are reduced by reducing the number of initial target groups.Secondly,based on the proposed situation fusion and assessment method,the situation of the target group is evaluated,and different types of UAVs are selected for task assignment according to the results of situation assessment and the relative position relationship between UAVs and target groups.Fuzzy ant colony algorithm is used to solve the first level mission planning among target subgroups in the target group so that the cost of each UAV executing tasks among the target subgroups is the lowest.Then,the fuzzy ant colony algorithm is used to perform two-level mission planning for the targets in a single target sub-group,to optimize the mission planning in the target sub-group.Finally,the efficiency of Multi UAV system collaborative task execution is improved by hierarchical collaborative mission planning.The experimental results show that the method is effective.4)To address the motion control problem of UAV in the cooperative tasks,this thesis proposed a Visual Servo Intelligent Control Method based on Adaptive Nonlinear Visual Feedback.Firstly,a hybrid image perception method is developed,and it completes the rapid extraction of the feature points.This method can improve the visual perception capacity of UAV in the moving process,and provide visual feature data for motion control.Meanwhile,the visual servo control method based on ensemble learning is used to optimize the sample data intelligently.This method can reduce the noise interference of the visual image and avoid computing the pseudo-inverse matrix directly to reduce the computational complexity.Finally,to address the slow convergence rate of the conventional methods,an adaptive nonlinear feedback control method is proposed,which estimates the value function and action in the sampling state through iteration,and completes the dynamic solution of the visual feedback nonlinear controller by combining with the UAV dynamic model,to achieve the efficient end-toend motion control of UAV using visual perception.Experiments show that this method can improve the motion control capacity of UAVs.