基于Rough集和神经网络的烧结过程异常诊断研究

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。

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

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.