Research On Hybrid Intelligence Algorithm And Its Application In Mission Planning

With the development of science and technology,unmanned systems are widely used in military,intelligent manufacturing and other fields,the core of which is mission planning technology.Faced with the actual needs of increasingly complex scenes and more reliance on efficient collaboration,such as UAV control,satellite collaborative search and other fields,traditional mission planning methods such as paradigm solver and decision theory methods can no longer meet the requirements.At present,widely used mission planning is based on two important technologies: path planning and task assignment.We abstracts them in this thesis into traveling salesman problem model and knapsack problem model,designs a hybrid intelligent algorithm and solves these two kinds of NP-hard problems.The specific research contents are as follows:1.This thesis introduces the background and significance of mission planning,as well as the research background and research status of two important technologies of mission planning in this thesis.Among them,the discount {0-1} knapsack problem model with known point path planning is a very important and universal model in mission planning.Then,it focuses on several intelligent algorithms involved in this thesis,including genetic algorithm and grey wolf algorithm,and briefly introduces some mechanisms of other algorithms introduced in this thesis,including the invasive weed algorithm.2.An improved parthenogenetic algorithm(IPGA)is proposed.To solve the path planning model of multi-starting point,closed-loop and multi-traveling salesman problem,a stable and easy-to-realize IPGA is used.However,in the IPGA,individuals only exchange information with their parents,which will lead to the loss of local information.In this thesis,a hybrid intelligent algorithm is designed by introducing the breeding mechanism of invasive weed algorithm,and the important local information is reserved by the way that individuals produce seeds for elite selection.The algorithm uses a two-stage coding based on breakpoint set,which reduces The way of population evolution is used to make the offspring of the population have better diversity;The complexity and convergence of the algorithm are analyzed and proved.Through numerical simulation experiments on standard data sets,the advantages of the improved algorithm are highlighted by comparing different intelligent algorithms.Finally,the universality of the improved algorithm is verified by experiments on practical examples.3.An adaptive grey wolf optimization algorithm based on improved differential evolution operator is proposed.Aiming at the discount {0-1} knapsack problem in task assignment problem,grey wolf optimization algorithm,a swarm intelligence algorithm with few parameters and easy realization,is used,and the grey wolf algorithm whose search parameters change adaptively according to the number of iterations is introduced.By improving differential evolution operator,it is applied to the iteration of grey wolf algorithm,and a hybrid intelligent algorithm with different evolution operators according to the parameter range is designed,which strengthens the global search ability of the algorithm.Using a repair operator based on greedy strategy,considering the factors of value density and value itself;the complexity and convergence of the improved algorithm are analyzed.The excellent solution performance of this algorithm is proved by the function optimization test and the comparison of algorithms in application domain data sets.The transformation function used in discretization is analyzed and discussed,and the universality of the algorithm in this thesis is verified by numerical simulation.