Modeling And Optimization Methods For Multi-UAV Cooperative Reconnaissance Mission Planning Problem

ISR(Intelligence, Surveillance and Reconnaissance) is the main task undertaken by Unmanned Aerial Vehicles(UAV). A team of UAVs with different capabilities cooperate reconnaissance will be the main manner to conduct battlefield reconnaissance in the future. How to make elegant mission plan according to the reconnaissance mission and capabilities of UAVs is one of the focus to exert the multi-UAV cooperative reconnaissance system and improve the system reconnaissance efficiency.Multi-UAV cooperative reconnaissance mission planning problem falls into the category of task allocation and resource scheduling problem in Multi-UAV cooperative system control. It concerns planning the right UAV reconnoitering the right target with the right sensor at the right moment while satisfying the UAV capabilities constraints as well as the reconnaissance targets requirements, so as to increase the efficiency of Multi-UAV cooperative reconnaissance system. Properly modelling and solving is the key to resolve this problem. Based on modelling theory and optimization theory, this dissertation studies the multi-UAV cooperative reconnaissance mission planning problem. The contributions are as follows:(1)Presenting the multi-UAV cooperative reconnaissance mission planning model. Following the thorough analysis on multi-UAV cooperative reconnaissance problem, the factors that should be considered in modelling the problem are summarized, which include the imaging requirements and time window constraints of the targets, and the capabilities of different UAV platforms and the imaging sensors onboard. Based on these, we formulate the factors and there contributions mathematically, and present the multi-UAV cooperative reconnaissance mission planning model(MUCRMPM). Furthermore, the multi-base multi-UAV cooperative reconnaissance problem are analyzed, and the corresponding model MB-MUCRMPM is presented. The multi-UAV cooperative reconnaissance mission planning model formulated in this dissertation has the following advantages. On the one hand, it embodies the essence of multi-UAV cooperative reconnaissance mission. On the other hand, it prevents us from bogging down by resolving the over-complicated model. Compared with the models formulated in previous studies, the model we present is more suitable for military applications since it concerns more specifics of the cooperative reconnaissance mission.(2)Putting forward a new adaptive evolutionary multi-objective optimization method, AEMOM. Multi-UAV cooperative reconnaissance mission planning problem belongs to NP-hard multiple objectives combinatorial optimization problems. Based on the rich research achievements in the field of evolutionary multi-objective optimization(EMO), a new adaptive evolutionary multi-objective optimization method (AEMOM) is presented. By means of formalization and modularization on multi-objective evolutionary algorithms(MOEAs), AEMOM can gives the most suitable MOEA for different multi-objective optimization problems(MOPs). AEMOM is a popular method to solve MOPs instead of a particular algorithm. The core of AEMOM is to formalizing the MOEA. For this, a general evolutionary multi-objective optimization schema(GEMOS) is present, which formalizes and modularizes MOEAs, and separates the problem specific components(PSC) from problem independent components(PIC). Based on GEMOS, AEMOM introduces the orthogonal design method to optimize PIC for a specific MOP, so as to get the profit MOEA for the MOP.(3)Proposing the AEMOM based multi-UAV cooperative reconnaissance mission planning algorithms. There are many constraints in MUCRMPM and MB-MUCRMPM, and these two models are not classical combinatorial optimization problems. Designing proper PSC is the emphasis to utilizing AEMOM in solving them. Combining the specifics of these two models, we presents the appropriate PSC, which assures the feasibility of the individuals and the evolutionary operators. And we propose a heuristic method to construct initial feasible population for evolution, which present the over-slow convergence caused by totally infeasible initial population. In order to verify the performance of the AEMOM based cooperative reconnaissance mission planning algorithm, a set of test instances which reflects different mathematical properties of MUCRMPM and MB-MUCRMPM is constructed via uniform design method based on analysis on the two models. Experiments on different test instances show that the algorithm can solve the problem effectively.(4)Presenting dynamic multi-UAV cooperative reconnaissance mission planning method. To accommodate the multi-UAV cooperative reconnaissance system to the changing mission requirements and battlefield during the execution of cooperative reconnaissance mission, the dynamic multi-UAV cooperative reconnaissance mission planning problem is investigated. Following the summarization on the circumstances which maybe emergent and need to deal with during the cooperative reconnaissance process, the dynamic multi-UAV cooperative reconnaissance mission planning model D-MUCRMPM is formulated. Then a fast heuristic dynamic mission planning algorithm D-MUCRMPA is proposed to solve the problem. Taking the pre-mission plan into account, the complexity of dynamic multi-UAV cooperative reconnaissance mission planning problem is reduced greatly. Simulation results show that D-MUCRMPA can solve D-MUCRMPM effectively and quickly.