| Production scheduling has always been one of the research focuses of academia and industry.With the continuous in-depth application of information technology in manufacturing enterprises,it has caused a sharp increase in the data accumulated in the actual workshop during the production process.These data includes a large amount of valuable dynamic information closely related to processing jobs,processing machines,processing technologies,and workshop personnel operations.How to effectively combine these dynamic information with production scheduling research has become a new research central issue.This thesis researches the data-driven production scheduling problem with the Job Shop workshop as the object,and proposes an modified genetic algorithm for job scheduling based on data mining and rule scheduling.And a real-time workshop scheduling system based on data is developed considering the features of workshop scheduling data under the Microsoft Visual Stduio Code environment.In order to realize the effective integration of actual data on the job site and the scheduling algorithm,the scheduling framework based on data mining technology is designed in this thesis.It mainly includes three parts: rule mining,static optimization and workshop management.First,select a suitable scheduling sample set and divide it into a training set and a test set in a certain proportion,and the CART decision tree algorithm in the data mining method is used to obtain the knowledge of dispatching rules for jobs processing from offline scheduling data to form a CART tree dispatching rule base.Then the obtained dispatching rule base is applied to the genetic algorithm to optimize and improve it,and a genetic algorithm based on data mining and dispatching rules is obtained.Furthermore,the algorithm is used to achieve real-time scheduling optimization of job shop,and it is also used to guide production workshop scheduling issues.In order to verify the feasibility and effectiveness of the proposed genetic algorithm based on data mining and dispatching rules,this thesis builds a simulation environment under Python jupyter and Microsoft Visual Stduio Code platform.Through the simulation tests of the classic job shop instances,the Gantt chart and the algorithm convergence curve comparison chart of the priority processing order of multiple tasks in the scheduling process are given,and the algorithm is compared with the optimization algorithms in the literature.The simulation results shows,the modified genetic algorithm for job scheduling based on data mining and rule scheduling proposed in this thesis has obvious advantages in optimizing performance and convergence speed. |
