The Research And Application Of Location Optimization Algorithm Based On Reinforcement Learning In Logistics Pull System

In recent years, the rise of the Automation Technology and production line promote the development of the manufacturing. In order to get a good development space, the enterprise trying to seek the way to control the production cost and to improve the production efficiency.As an important part of the production logistics, the allocation of the storage location of automatic storage and retrieval system has remarkable influence to the efficiency and energy cost of production line. Using a good allocation strategy can apparently decrease the cost of time and the cost of energy, improve the operating efficiency at the same time.In this paper, we take the situation of a famous automobile factory into consideration. Aim at the problem of low efficiency and high energy cost in current store management, optimal objects was come up with. Consider the character of the allocation of storage location problem model, after analysis different solution, we research and use an algorithm of reinforcement learning based on environment abstract and temporal abstract.Aim at the character that the distribution of the storage location problem is a large scale problem, we wipe off and abstract the environment information, convert the environment information into the classification of the storage location information. Shrink the size of input, speed up the calculation and the convergence process.For it is a global optimum problem, we take SMDP into consideration, abstract the time process, transform the immediately statistics to period statistics. By calculate the result of the action in the period, we adjust the module and avoid it become partly optimization.We use BP neural network to calculate the value function in order to avoid the lack of RAM and training sample, and we train it by the best and the current period evaluation value. If we use query table solution, it may result to a big requirement of RAM and a long training time.At last, build a system based on the research contents, description the process of system design and system implementation, show the optimized effect to the production logistics of the automobile factory.