Logistics Operation Optimization Of Coil Warehouse Based On Reinforcement Learning

With the development of technologies,the enterprises of iron and steel have been relatively mature in reducing costs,improving production processes,increasing sales,and so on.Further excavate the profit space is very limited in the productive process.In this case,many enterprises of iron and steel,especially some large iron and steel groups have focus on the change in logistics pattern,and improve the level of modern logistics management from different angles.Storage of steel coils is an important part of iron and steel enterprise logistics management.The traditional management method of relying on artificial experience has been unable to meet the fast and efficient production requirements and supply rhythm of the enterprise.At present,most of the warehousing management system in the design of the imperfect has exposed more obvious problems.A series of problems as the system does not optimize the function and did not take into account the planned cycle of the coil in the storage state of the impact of the follow-up operation and there is no plan to optimize,result in that the use of the warehouse is not high,the amount of pile is large,and increase a lot of logistics operating costs.In this paper,we study the logistics optimization of coil warehouse,and combine some of the methods of reinforcement learning in the field of artificial intelligence research in recent years to apply to the optimization of logistics operation of steel coils.The main research work is as follows:1)This thesis summarizes the current situation of the optimization of the coil library operation,the shortest path problem of the crane and the current situation of the stacking method,and analyzes the shortcomings of the existing research;This thesis introduces the theoretical basis of the reinforcement learning algorithm used in this paper,and summarizes the research status.2)A kind of dynamic programming model is established for the optimization of the coil logistics operation,and a heuristic algorithm considering three priority strategies is designed according to different objectives.3)A kind of reinforcement learning algorithm is constructed for the optimization of coil logistics operation.The heuristic method is improved by using the Rollout algorithm.4)A Monte Carlo tree search algorithm based on the Rollout algorithm is proposed to solve the optimization problem of the coil logistics operation.5)A number of simulation examples are cited,and the above models and algorithms are validated to verify the correctness of the model and the effectiveness of the algorithm.