冷轧建模中基于粗糙集和神经网络的案例检索

针对一般基于案例属性相似度的案例检索方法在实际应用时存在相似性度量的定义等难点而无法保证两效、准确检索的问题。提出了一种新的案例检索方法。根据粗糙集理论对冷轧负荷分配决策表进行简约以及从简约的决策表数据中发现规则，在此基础上利用发现的规则建立隐层节点数更合理的神经网络模型。以此神经网络模型实现相似案例检索，经在线实验证明，应用该方法可实现快速高效的案例检索。提高了整体智能推理建模的效率和质量。

Case retrieval based on rough set and neural networks for cold rolling modeling

The general retrieval based on similar degree between the case features is difficult to guarantee the rapidity and accuracy of case retrieval for practical application in definition of similarity measurement,therefore a new case retrieval method is developed in which rough set is introduced to reduce the decision table of the cold rolling schedule and to find the rules from the reduced table by which the neural networks model with proper hidden nodes is constructed to retrieve the similar cases.An on-line experiment with satisfying results show that case retrieval with rapidity and accuracy can be realized by adopting the proposed case retrieval model,and the efficiency and quality of case-based modeling can be improved accordingly.