基于RBF神经网络的钢构件质量追溯系统研究

Research on the System of Traceability of Quality of Steel Component Based on Radial Basis Function(RBF)Neural Network

提出一种基于RBF神经网络的数据挖掘方法,将RBF神经网络应用于数据挖掘的分类和预测中,解决钢构件过程中的性能预测问题。其中用黄金分割法确定基于RBF神经网络的隐层节点数,减少该算法的计算复杂度,最终将其应用于某钢铁企业质量控制系统。构建对钢构件质量检测的数据挖掘及质量追溯平台,该平台是基于RBF神经网络的数据挖掘技术的。实际应用证明,产品的质量合格率可达到96.27%,符合国家相关的标准和技术指标。

To solve the performance prediction problem in the steel production process,this paper presentsed an approach which is based on RBF neural network data mining method and uses RBF neural network in classification and prediction of data mining.The hidden layer nodes of the RBF neural network were determined by the golden section method to reduce the computational complexity of the algorithm,which were applied to a steel enterprise quality control system.Finally,aplatform of data mining and quality retrospective,which is based on RBF neural network data mining technology,was constructed in product quality testing in steel companies.Practical application shows that the qualified rate of products can reach96.27%,in line with national standards and technical specifications.