一种基于陶瓷工艺知识图谱的可推理数字化设计服务平台架构研究

Research on the Architecture of a Reasoning-Enabled Digital Design Service Platform Based on Ceramic Process Knowledge Graph

摘要: 近年来，中小陶瓷企业对产品设计创新能力的重视程度不断提升，但受限于设计资源获取渠道有限、工艺知识积累方式传统等因素，其设计能力和响应速度难以跟上日益增长的个性化消费需求。本文提出一种基于陶瓷工艺知识图谱的可推理数字化设计服务平台架构(简称：RDSPA)，面向中小陶瓷企业提供低成本、低门槛的设计知识服务。该平台的核心是一个面向陶瓷设计领域的工艺知识图谱，通过异质图结构实现工艺知识的语义化存储，通过图注意力网络实现跨类别推理检索，并集成Beta-spline参数化建模与改进Harris角点检测，分别支撑器型连续形态表达与图像特征自动提取。模块级实验验证了改进Harris角点检测与Beta-spline曲线拟合两项核心技术的有效性，改进算法在检测精度与拟合精度上均优于经典方法。在原型图谱上开展的路径检索测试验证了推理模块在跨类别关联发现中的有效性。以工艺知识图谱和可推理服务为核心的平台化设计模式，能够有效提升中小陶瓷企业的设计能力和订单快速响应能力，是弥合其设计能力与市场需求之间差距的一条可行路径。

Abstract: In recent years, small and medium-sized ceramic enterprises have placed increasing emphasis on product design innovation capabilities. However, constrained by limited access to design resources and traditional approaches to process knowledge accumulation, their design capabilities and responsiveness have struggled to keep pace with the growing demand for personalized consumer products. This paper proposes a reasoning-enabled digital design service platform architecture based on a ceramic process knowledge graph, referred to as RDSPA, which provides low-cost, low-threshold design knowledge services for small and medium-sized ceramic enterprises. The core of the platform is a process knowledge graph oriented to the ceramic design domain, which employs a heterogeneous graph structure for semantic storage of process knowledge, utilizes a graph attention network for cross-category reasoning and retrieval, and integrates Beta-spline parametric modeling and improved Harris corner detection to support continuous form representation of vessel types and automatic image feature extraction, respectively. Modular experiments have validated the effectiveness of the two core technologies—improved Harris corner detection and Beta-spline curve fitting—demonstrating that the improved algorithms outperform classical methods in both detection accuracy and fitting precision. A path retrieval test conducted on the prototype knowledge graph has validated the effectiveness of the reasoning module in discovering cross-category associations. The platform design model, centered on the process knowledge graph and reasoning-enabled services, can effectively enhance the design capabilities and order responsiveness of small and medium-sized ceramic enterprises, offering a viable pathway to bridge the gap between their design capacity and market demands.