基于模糊聚类RBF神经网络的生产指标预报模型  

Production Indices Forecasting Model Based on Fuzzy Cluster RBF Neural Network

作　　者：刘威[1] 李慧莹[1] 柴天佑[1] 

机构地区：[1]东北大学自动化研究中心,沈阳110006

出　　处：《系统仿真学报》2004年第5期1030-1033,共4页Journal of System Simulation

基　　金：国家863计划资助项目(2001AA413510);国家863计划资助项目(2002AA414610)

摘　　要：本文提出一种同图论中可达矩阵和区域划分思想相结合的模糊聚类方法,以此来确定RBF网络的隐含层节点数。并根据某选矿厂经济指标和生产指标之间的关系,建立了生产指标预报模型,并应用于选矿厂MES系统中,应用结果表明所建模型收敛快、预报精度高,使企业能够及时了解生产过程成本动态,优化生产运行管理。A way of fuzzy cluster is provided, which is combined with reachability matrix and the idea of region partition in graph theory, in order to find out the number of the hidden layer nodes of RBF neural network. Based on the relationship between economic indices and production indices, the forecasting model of production indices was set up, and had been applied in MES of ore concentration plant. It has proven that the model is quickly convergent and high accurate. By using the proposed model, enterprises can dynamically acquire the cost information of production process, and optimize production and operation management.

关 键 词：可达矩阵 区域划分 模糊聚类 RBF神经网络 生产指标预报 

分 类 号：TP-9[自动化与计算机技术]