Study On Flexible Scheduling Of Knowledge Workers Based On Learning Ability

In the era of knowledge economy, scientifically assigning knowledge workerstasks and arranging them are the key activities for knowledge enterprises to make fulluse of knowledge resources and achieve their world-class competitiveness. Due to thespecial nature of knowledge workers, as well as the complexity of the tasks and thedynamic of the scheduling environment, the scheduling problem of knowledge workersis more difficult than the traditional staff scheduling.This paper grasps the distinctive features of knowledge workers, such asknowledgeable, multi-skilled, strong ability to learn, attention to match with their tasksand team members. Based on traditional staff scheduling theory, this paper treats thestaff's long-term skills development as an enterprise strategy and conducts thefollowing study:Firstly, on assumption that learning ability of knowledge workers is certain, amodel of certain learning ability is proposed. Constrained by the matching degree ofknowledge workers and their tasks or team members, aiming at minimizing the totalscheduling cost and maximizing the growth of knowledge workers'skill scores, takingstaff rest and transfers between different tasks, task delay into consider, this paperdevelops a multi-stage flexible scheduling model of multi-skilled knowledge workersin view of certain learning ability. On the other hand, a model of risk learning ability isbuilt on assumption that learning ability of knowledge workers is random. Then on thebasis of Markov random sequential theory, another model of multi-stage flexiblescheduling of multi-skilled knowledge workers in view of random learning ability isbuilt.Secondly, an analysis of advantages and disadvantages of traditional GeneticAlgorithm is made. To solve the problem premature convergence or slow convergence,this paper proposes Adaptive Genetic Algorithm, overall considering convergence anddiversity, by dynamically adjusting the perform probability of genetic operators(crossover and mutation) according to fitness, this algorithm protect the bestindividuals to get access to the next generation, it also speed up the elimination ofindividuals of low fitness. This approach can effectively improve the populationdiversity and search efficiency to avoid premature convergence.Finally, Adaptive Genetic Algorithm is used to seek the optimal schedulingpolicies of the model based on certain learning ability. The results show the excellent convergence property of Adaptive Genetic Algorithm and prove the effectiveness ofthe model.