Research And System Design Of Material Pulling Method Based On Deep Learning

With the intensification of market competition and the rapid progress of science and technology,flexible production with multi-varieties and small batches has become an effective means for manufacturing enterprises to respond quickly to complex and changeable markets,but at the same time,this production mode increases the difficulty of production logistics management.Aiming at the problems of lack of information interaction,inefficient distribution and poor timeliness of distribution in traditional material pulling process,an automatic material pulling scheme based on reservation time is proposed,the Long-Short Term Memory(LSTM)prediction model based on material consumption rate is established,and a material pulling system is designed and developed,which improves the accuracy and timeliness of material distribution.Firstly,the related theories of deep learning and material pulling is studied respectively in this dissertation.In the aspect of deep learning,firstly,the related concepts and terminology of deep learning are elaborated in detail,and the neural network which has an important impact on the rapid development of deep learning is summarized.In the aspect of material pulling,at first,the material classification method is studied,and then the material pulling mode commonly used in mechanical manufacturing industry is summarized.Finally,the traditional material pulling process in manufacturing workshop is introduced and its serious problems are analyzed.Secondly,an automatic material pulling scheme based on reservation time is proposed,and the material pulling process is designed and the characteristics of the scheme are described in detail.Then the generation of material distribution plan based on distribution point,which is the most critical part of the material pulling scheme,is explained,the calculation rules of material distribution plan are described,and the calculation process of material distribution plan is explained.The generation of material distribution plan is analyzed by an example.Then,the LSTM prediction model based on material consumption rate is constructed,the structure of the model is designed,the training methods and steps are introduced,the simulation experiment of the model is completed,and the prediction accuracy of LSTM prediction model is compared with that of back propagation(BP)neural network prediction model.The experimental results show that the prediction effect of the LSTM prediction model based on material consumption rate model is better.Finally,based on the actual production situation of the engine assembly workshop of a company,a material pulling system is designed and developed for the company.The system takes just-in-time material distribution as the guiding ideology.Through this system,accurate material distribution can be achieved,material distribution rate and response rate can be improved,production logistics can be balanced,production activities can be carried out in an orderly manner,and production capital occupancy can be balanced.