Modeling And Process Analysis Of Garment Customization Production System Based On Knowledge Graph

The traditional garment production model is changing to customized garment production rapidly.Orders are converted from few varieties and large batches to multiple varieties and small batches.Under this transformation,the garment production line is required to be highly flexible,and the line resources should be allocated to respond quickly.However,the long garment production process and the variable process parameters caused by individualized needs pose a huge challenge to custom garment production,process optimization and resource allocation rationalization.Aiming at the problem of resource allocation in custom garment production,this paper conducts three researches: system modeling,dynamic process data fusion,and intelligent agent for process optimization.The research work in this paper includes the following:（1）A customized production process model of garment for cognitive manufacturing is established.The paper establishes a multi-level mathematical model from the unit and system levels,defines the analytical problem,optimization objectives,and constraints of the custom garment production process.In addition,a cognitive manufacturing system model for customized garment is proposed.A multi-factor information model based on Sys ML（Systems Modeling Language）has been constructed from the perspective of production processes,use cases,requirements,state machines and control.（2）A dynamic data fusion method of garment customization production process based on knowledge graph is designed.According to the characteristics of garment customization,three types of ontology modeling methods of process,resource and feature are designed based on the knowledge modeling method of ontology.For the characteristics of small batch size,long process and high data dispersion,a knowledge graph construction method based on bi-directional fusion for garment custom production is proposed.The paper establishes a KGF（Knowledge Graph Facet）model with order as the unit,which can build a complete knowledge graph of garment custom production.Moreover,knowledge representation,generation,inference and extension for custom garment production are studied,and a method for fusing multimodal data in the custom production process is proposed.（3）An intelligent agent approach for garment customization process optimization is studied.The paper models perceptual and cognitive agents,and designs a strategy for implementing perceptual and cognitive agents with local and global process optimization.A multi-factor interaction between perceptual and cognitive agents is realized based on data flow,knowledge flow and control flow.A process optimization model and algorithm based on dual-loop reinforcement agent（DLRA）are proposed to realize multi-level optimization of production process.Finally,the proposed method is applied to the custom production process of a suit to verify the feasibility and effectiveness.Compared to the traditional solution,the DLRA algorithm develops the most superior solution with 60.82%,37.74%,and 36.73% increase in completion time,cost,and average equipment utilization respectively,and 39.58%decrease in average equipment waiting time.