基于改进型神经网络的软测量模型的建立与应用

Establishing and Application of Soft-sensing Model Based on Improved Neural Network

使用改进的RBF(径向基函数)神经网络(即小波神经网络)来建模,并采用改进算法确定其隐含层节点的各个参数,然后以一个充填搅拌生产过程中的料浆浓度为研究对象,建立其软测量模型,并通过仿真实验验证了该方案的可行性。该方案充分利用了软测量的优点及神经网络的强分类能力,为沙浆浓度的测量提出了优良的方法,从而可以保证整个充填系统的可靠运行。该测量模型应用范围很广,可以对工业生产中的一些重要的工艺参数如矿浆、煤浆、混沙浆、纸浆等进行精确测量。

This paper introduces a improved RBF Neural Network （RBFNN） for modeling with a improved arithmetic to fix on the parameter of latent node and apply it to a sand-slurry concentration measurement. The feasibility is proved by emulational experiment. It puts forward excellent way for the sand-slurry concentration measurement taking full advantage of soft-sensing method and neural network＇s classificatory ability. It can guarantee the reliable run of filling system . This soft-sensing model can measure lots of technical parameter, such as mineral-slurry, coal-slurry, sand-slurry ,paper-slurry.