自组织神经模糊系统及其在辛烷值软测量中的应用

Self-organizing Neural Fuzzy System and Its Application to Modeling Octane Number

本文提出一种基于神经网络的自组织模糊系统 ,该系统由前向多层神经网络构成 ,分两阶段学习 :竞争学习阶段、监督学习阶段。该系统最大的特点是根据输入数据的分布灵活地划分模糊集合 ;根据同一聚类空间输出数据的分布确定模糊规则数 ,而不只是由输出数据的平均值确定。最后将该系统用于某炼油厂辛烷值的软测量建模 ,实验结果表明 ,该系统具有结构简单、学习速度快、建模精度高、泛化能力强等优点 ,优于

In this paper,a novel self organizing neural fuzzy system is proposed for nonlinear soft sensing modeling of chemical process.It is a feed forward multilayer neural network and learns in two main phases,namely competitive learning phase and supervised learning phase.It partitions the input space according to the distribution of the training data and uses the average,the maximum and the minimum point of desired output of all points belonging to same clustering space to determine the center of consequent of fuzzy rule not only the average point.The proposed systems is applied to a practical application of modeling octane number of continuous reforming plant.Experiment results show that it possesses better generalization ability and simple model structure and is better than UOP model.