Research On A Method Of Task Allocation Under The Condition Of Human Ability And Task Uncertainty

Dynamic task assignment,which makes the right task assigned to the appropriate members in real time could take full advantage of the system resources and improve the performance of the task.In an uncertain system,the arrival time of task is random and uncertain.Besides,the type of the task often varies,and the changing external environment will increase the system's uncertainty.At the same time,people's ability of handling task is also volatile,affected by their own status or external uncertainty.With the development of artificial intelligence,the theory of dynamic task allocation is becoming more and more mature,and its application field is also more extensive.But there is a lack of research on the dynamic task-staff allocation in the uncertain human-machine system.The study of task-staff allocation is mainly focused on external uncertainty,and the fluctuation of personnel capability is ignored which also lead to the uncertainty of human-machine system.Based on the theory of dynamic task assignment in multi-agent system,the MDP model is established to solve the problem of dynamic task-staff assignment.The MDP model enrich the theory of dynamic assignment and solve the problem of dynamic task-staff assignment and the selection of cooperation mode in the real uncertain system,so as to reduce the adverse effects of uncertain events.and improve system efficiency.The thesis is based on the following process:Firstly,the MDP model is established for the uncertain man-machine system based on external uncertain environment and fluctuation of operators' ability in human-machine system,with giving concrete definition and explanation of each parameter in the MDP model,and proving the existence of solution.The improved strategy iterative algorithm is used to solve the model.Secondly,the MATB system is chosen as the application object to solve the problem of task-staff assignment and selection between two cooperation modes.The parameters of the MDP are obtained by appropriate experiment,including mental workload value,processing speed probability and personnel heterogeneity parameter.Finally,MATLAB is used to compare the stochastic allocation strategy and the MDP-based allocation strategy.Result shows the MDP allocation strategy is more effective.This paper also explores the two cooperation modes,concluding that the team of similar ability is better than the one of different abilities in cooperation.