面向平缝工艺信息融合的知识图谱构建方法

Knowledge graph construction technology for provision of sewing process information

针对缝纫工艺流程链路长,生产要素多样,工艺知识分散化的特点,首先,提出缝纫工艺信息组织建模方法,面向多源异构平缝工艺信息构建本体模型;其次,研究基于面料力学性能的平缝工艺推荐方法,构建面料基础性能、缝制性能知识体系,结合面料层次分类方法和实验参数分析,形成力学性能参数与缝制参数之间的数学模型;在此基础上,以专家经验文本、实验测量数据为原始数据,构建面向平缝工艺信息融合的知识图谱,开发WEB系统展开应用验证。结果表明:所构建的工艺推荐方法实现了对面料平整度和最大缝缩率的预测,知识融合系统实现了缝纫工艺知识的智能搜索和缝制参数的智能推荐,为工艺数据集成、装备故障运维、工艺路线设计、产品质量控制提供决策参考。

Objective The sewing process is characterized by long processing chains,diverse production elements and scattered processing information.Using knowledge graph technology for the management of design,operation and maintenance data generated during the sewing process,this research proposed a knowledge graph construction method for sewing process information management to achieve standardized knowledge representation.Method Modelling methods for the organisation of sewing process information were investigated.The process information generated during the fabric sewing process was classified,and a sewing process knowledge ontology model was established based on the classification results to realise the construction of a knowledge graph.The process recommendation method was established based on the graph.Experiments were carried out on fabric structure,fabric mechanical parameters and fabric sewing process to establish a knowledge system and to analyse the mechanical properties of fabrics before and after sewing.Based on the analysis,a regression model of fabric mechanical properties and sewing flatness and a theoretical model of fabric sewing shrinkage were established.An ontology model of the sewing parameter knowledge system was created for sewing parameter recommendation based on knowledge graph.Results According to the requirements of sewing process corpus and knowledge graph,a process recommendation method based on knowledge graph was established by combining the characteristics of industry knowledge structure and knowledge management requirements.The developed ontology and knowledge graph contains a total of 2865 entities and 52 relations,with wide knowledge coverage and strong generalization,facilitating the standardized representation of unstructured knowledge.The relationship between mechanical parameters and sewing parameters were modelled for common fabrics in the flat sewing process,the flatness of the sewn fabric and the maximum sewing shrinkage were predicted and recommendations for sewing parameters,bonding parameters and processing instructions for the corresponding fabrics were achieved.The technical architecture for intelligent recommendation of sewing parameters was established.The knowledge system was interconnected with other sewing process knowledge and enabled integration of process information.Conclusion The established knowledge graph is characterized by strong integration and interconnection of sewing process knowledge,which enables data integration and facilitates the maintenance and expansion of knowledge at a later stage.The research provides a useful supplementary case for process information management paths in the sewing industry,showing that knowledge graph technology has good application prospects in the sewing industry and has a certain reference value.