基于Rough集和神经网络的烧结过程异常诊断研究  

STUDY ON ABNORMITY DIAGNOSIS IN SINTERING PROCESS BASED ON ROUGH SET THEORY AND NEURAL NETWORK

作　　者：张小平[1] 张继生[1] 王杰[1] 历君 

机构地区：[1]鞍山科技大学计算机科学与工程学院 [2]北台烧结厂

出　　处：《烧结球团》2005年第4期24-26,共3页Sintering and Pelletizing

摘　　要：为了及时、准确诊断烧结过程的异常状况并及时消除异常,本文将Rough集和神经网络相结合,建立了烧结过程异常状况智能诊断系统。基本思想是首先利用Rough集对知识库进行约简,然后利用神经网络对约简后的知识进行分层融合。该系统具有简化样本、适应性强和不易陷入局部最小点等特点,能有效处理异常中的噪声或不相容的信息。In order to diagnosis abnormity condition in sintering process accurately in time, in this paper, we build an intelligent diagnosis system for sintering process combining rough set theory with neural network. It＇s basic idea is to reduce rules in knowledge base using rough firstly, and then to classify and merge the knowledge utilizing neural network. The system has so many characteristics such as： reducing examples, more flexibility and not easy to fall into local minimum point. It can also process the noise or inconsistent information on the abnormity condition.

关 键 词：异常 诊断 ROUGH集 神经网络 烧结过程 诊断研究 智能诊断系统 基本思想 分层融合 有效处理 

分 类 号：TF046.4[冶金工程—冶金物理化学] TP18[自动化与计算机技术—控制理论与控制工程]