基于案例推理的纺纱质量预测模型研究

Case-based Reasoning Model for Quality Prediction of Spinning Yarn

针对纺织企业的纱线质量控制问题,提出了基于案例推理(CBR)的纱线质量预测模型。利用纺织企业纱线生产中积累的历史数据构建案例库,采用聚类算法为其建立索引结构。在此基础上,使用最近邻法分两步检索出相似案例,并对相似案例进行修改以获得预测值,同时采取主动学习策略保存当前案例,更新案例库。最后使用纺纱生产的实际数据进行仿真,得到具有较高精度的预测结果,验证了模型的有效性。

According to the problem of yarn quality control in a textile enterprise, a case-based reasoning （CBR） model for predicting yarn quality was proposed. The historical data of spinning production accumulated in a textile enterprise were used to build case base, while a clustering algorithm was selected to create indexical structure. Based on the structure, the nearest neighbor method was used to search similar cases, which were modified to become the predictive values. Then this new case was reserved on an active learning strategy and case base was refreshed. Finally, the simulation was implemented by the actual data in spinning production. The more precise predictive results prove the effectiveness of this model.