Task Allocation Of Multi-AUV System Based On Multi-objective Optimization

In this paper, we just take the topography survey of multi-area for example to focuson the study of task allocation of multi-AUV system based on multi-objectiveoptimization. A multi-objective optimization algorithm based on ant system was proposedto apply to the practical application of task allocation of multi-AUV system and solve aproblem of the cooperative technology of multi-AUV system.Based on the traditional ant colony algorithm, we proposed a multi-objectiveoriented optimization algorithm (multiple ant colony system optimization algorithm—MACSOA). Firstly, establish an ant colony for each objective and the amount of the antsin each ant colony is equal. Secondly, assign a target for each ant colony and keep it allthe same. In each iteration cycle, each ant colony optimizes the target that assigned for itin accordance with the single-objective ant colony optimization to create a feasiblesolution, and then find the non-dominated solutions by comparing all the solutions foundby the ants of the corresponding ant colony which are stored in the Pareto set. Meanwhile,the solutions that which are dominated by the others should be punished as a principle.Finally, calculate the deviation between the solutions of the Pareto set and the idealsolution supported by the weight coefficient, and export the solution which has thesmallest deviation as the final output.When being applied to solve the task allocation problem based on the multi-AUVsystem, the MACSOA should be improved. In the step of feasible solution constructing,this paper proposed a competitive mechanism between the multi-AUV based on thetransition probability of ant system algorithm to get a feasible solution. For the taskallocation of isomorphic system, we have to consider the problem of energy limit of eachAUV, and as to the heterogeneous systems, the task allocation should also consider that ifthe operation ability of the AUV is match with the required ability of the task. Theprocessing method is studied in this paper.Finally, we perform a simulating experiment of the multi-objective optimizationoriented task allocation problem of multi-AUV system based on Qt platform, and the taskallocation problem of the isomorphic system and the heterogeneous system that proposedin this paper is simulated and the simulation results is provided.