Study On Modeling And Optimization For Stochastic Flexible Scheduling By Co-cooperation Evolutionary Algorithm

Stochastic Flexible Scheduling problem is one of the most important and core problem in manufacturing system.The objective is to maximize the robustness and efficiency of production through the resources allocation with high-efficiency under the condition of constraint.In stochastic flexible scheduling problem,each operation can be processed on difference resources,so,it can satisfy more flexibility.Hence,stochastic flexible scheduling problem is more in line with the needs of actual production and attracts more attentions based on the characteristic mentioned before.In this paper,we focus on using the meta-heuristic algorithms to solve flexible scheduling problem.In the previous work,we found that particle swarm optimization(PSO)can get better performance than other classic heuristic algorithms when solving scheduling problem.And the grouping mechanism is suitable for large scale problems.So,we choose PSO as the basic algorithm of our proposed algorithm and then combined with grouping mechanism and parameter self-adjust mechanism.Firstly,we changed scheduling problem from discrete problem to the continuous problem by using real number to encode the problem,and this can reduce complexity of scheduling problem.Then we use grouping strategy which place different variables into different groups,and evaluate groups separately.In this way,we can reduce the influence between independent variables,and highlight the dependency of correlated variable.It can help us obtain a better solution.And for the grouping size,we also use group size self-adjust mechanism,which means that we will keep current grouping size in next generation if it gets better solution in current generation,and choose another size from a pre-given set if it does not find a better solution.Then for the choice of evolution formulas and parameters,we use self-adjust mechanism to reduce the influence of human choice.In the process of iteration,we adjust the parameters based on whether they make contributions to individuals when they find better position in the previous iteration.Finally,we did a list of experiments to prove that,our proposed algorithms can have better performance of searching optimal solutions and have strong optimization ability and stability than original evolutionary algorithms.