摘要
提出一种高炉径向煤气流模式分布的混合神经网络模式识别方法,针对从“样本特征模式空间”到“样本类别空间”具有较强非线性的模式识别问题,提出了一种由自组织竞争子神经网络(ASCSNN)和基于径向基函数子神经网络(RBFSNN)串联而成的混合神经网络.其中,ASCSNN用于对输人的特征模式空间进行自组织分类,以使识别更加稳定;RBFSNN对样本特征模式进行识别.根据工业现场采集数据进行实验,验证了该方法的可行性.
A hybrid neural network is proposed for the radial gas distribution of blast furnace.According to the patterns with nonlinear from the inputs to outputs, the proposed network consists ofself-organizing competing sub-neural network (ASCSNN) and the radial basis function sub-neuralnetwork (RBFSNN). The ASCSNN is used to dassify the patterns of inputs, and the RBFSNN is usedto identify the pattern of inputs. The tests on the collected real time data show that the proposedmethod is feasible.
出处
《沈阳工业大学学报》
EI
CAS
1999年第3期265-268,共4页
Journal of Shenyang University of Technology
关键词
模式识别
混合神经网络
高炉
径向煤气流
neural network
competing learning
supervied learning
pattern distribution