Aircraft Path Planning Based On Multi-agent Co-evolutionary Algorithm

Aircraft route planning has long been a major research subject for mission planning system, which is the key technology for the aircraft to realize flight safety and successful attack. Traditional route planning algorithm can not satisfy the requirement of intelligence with the development of computer. Therefore, it has been an inevitable trend in the development and application of this field to use intelligent algorithm which has high intelligence and strong adaptability.This paper focuses on the problem of aircraft route planning. The technical characters of route planning were analyzed, combined with the character of multi-agent system that solves problems in a distributed way. The method of aircraft route planning based on multi-agent co-evolution algorithm was proposed with the introduction of co-evolution mechanism. The effectiveness and practicality of this method was verified by the simulation experiment of matlab.The organization of this paper is as follows.Firstly, the mathematical model of aircraft route planning was constructed. The whole frame of aircraft route planning was built. It was decomposed into four parts, which contains expression of flight environment, processing of constraint information, planning of desired target and key algorithm. The mathematical expression of planning environment was proposed with the method of contour space modeling. This method was based on basic characteristic factors of aircraft, such as mobility, terrain information, threat constraints, etc.The constraints can be classified into three kinds,which contains terrain information constraints, threat constraints from radar and air defense firepower, flight mobility constraints.Evaluation function was designed to calculate the price paid for fulfilling the route. This function was based on flight mission, concealment, voyage, flight turning angles and other requirements.Secondly, basic principle of multi-agent co-evolution was analyzed. Agent structure and formal description were researched. The characteristic that multi-agent system solves problem in a distributed way was summarized. Local sensing, competitive cooperation and autonomous learning were introduced to evolutionary algorithm. Multi-agent co-evolution was realized by means of sharing information between multi-agents.After that, specific operation of route planning method based on multi-agent co-evolution was realized. Definition of track agent was given and its evolution behavior was designed. Further realization of moiety regional method was carried out in built two-dimensional route planning space. The whole planning space was parallel divided into D dimensional plane region, based on precision requirement and minimum step constraint of route planning. On this basis, co-evolution mechanism was introduced into algorithm by means of the designing principle of it, and turning angle, voyage and other constraint information were merged into its operators. To be specific, neighborhood competition,neighborhood cross, variation and self-learning operators were designed. Optimal route agent was selected, according to fitness evaluation function. The optimal route that has shortest voyage, lowest consumption and minimum threat was obtained under this method.Finally, the simulation experiment of aircraft route planning based on multi-agent co-evolution was carried out, and the simulation result has verified its effectiveness.