Research On Production Scheduling Rule Extraction And Application System Based On Data Mining Technology

Good production scheduling rules are important prerequisite for ensuring orderly and efficient production.The current manufacturing technology is developing in intelligent manufacturing based on knowledge and information,which puts forward further requirements on scheduling rules,such as application scope,response speed,intelligence,etc.An important solution is the heuristic scheduling method based on data.The method extracts the scheduling rules from massive data produced in the manufacturing process based on data mining technology.The rules are then further generalized and optimized.Finally,the scheduling rules which are applicable to the enterprise are obtained.According to the current states of the production and the existing management system,the focus of this paper is to study the part of workflows,contents and key technologies of heuristic scheduling based on data mining technology,and meanwhile propose the solution method.According to the objective requirements of furniture enterprises in production scheduling and the characteristics of data mining technology during production scheduling,the system structure and business model based on production scheduling rules extraction and application system of data mining technology have been established.The system can integrate business data from multiple management systems and carry out discretion process and attribute extraction of the data sets.And then scheduling rules are extracted from the data source and applied in the dynamic scheduling of workshop.In order to improve the flexibility of the data acquistion method,reduce the coupling property,and realize the unity of data semantics,a method of multi-source heterogeneous fusion of production scheduling data is proposed based on ontology.First of all,the process and characteristics of discrete manufacturing are analyzed.Based on the extent to which the production scheduling results are affected and the role that the involved objects play in produc-tion,a unified object model —production event information element and production scheduling event information element—is established by using the method of hierarchical division.The global ontology and local ontology are then built individually.Finally,the ontology technology is utilized to realize the semantic unification and data fusion of various management systems.In order to solve the general problems in continuous attribute values of fused real production data,a dynamic discrete algorithm based on single dimension and multi radius clustering algorithm is proposed.First of all,accounting for the characteristics of non-uniform density and roughness of the real production attribute values,a single dimension and multi radius data clustering algorithm is designed,which achieves better results for clustering processing of attribute values.After that,the dynamic adjustment rules are applied to the clustering parameters of the clustering algorithm,and the data discretization objective function is designed to realize the automation of discretization process of attribute values.Finally,the validity and the efficiency of the algorithm are verified by experimental data.In order to solve the problem of dimension disaster caused by excessive dimension of data attribute in production,an attribute extracting algorithm is proposed based on the importance and correlation coefficient.Firstly,according to the characteristics of production data,a metric function used to measure the importance of attribute is established by using fuzzy entropy.A method for searching the associated attributes is then proposed based on the increment of fuzzy entropy.Secondly,the effect of attribute combination on the improvement of accuracy of data mining is investigated.The attribute reconstruction function is established to reduce the number of attributes by using linear combination of attributes.Finally,the experiment results prove that the attribute selection algorithm may not only effectively reduce the attribute dimension,but also improve the accuracy of the data mining algorithm.A heuristic scheduling method based on simulation optimization and data mining is studied in order to realize the real-time scheduling in furniture products manufacturing.Firstly,the assumptions and constraints in the production process of furniture products are analyzed,and the mathematical model of production scheduling is proposed based on production set difference.Secondly,accounting for the problem of insufficient coverage of production data,the simulation environment of production process is established,in which the intelligent algorithm of production scheduling is integrated to generate sufficient and valuable simulation data.In order to ensure the quality of the extracted rules,the production data are optimized before exacting rules by using data mining techniques.Finally,the C4.5 algorithm is used to extract the scheduling rules,and the pruning process is used to obtain the individual schedul-ing rules for each processing unit,which is applicable to the real-time scheduling for production.Based on the results mentioned above,the technical application for furniture manufacturer has been carried out.Firstly,the global ontology and the local ontology of each management system are established.Secondly,the production scheduling rule extraction and application system based on data mining technology is developed.The data related to production and production scheduling in a variety of management systems may be fused in this system to form the unified data of production event information element.Moreover,Data discretization and key attribute extraction for instance data are conducted automatically and effectively in the system.In addition,the system can also combine the simulation system and the production instance data to extract the good knowledge of scheduling rules and carry out the real-time scheduling effectively.The test results show that the system improves the effect of production scheduling,which ultimately enhances the production capacity of the enterprise.