基于神经网络的软测量技术及应用  

Neural Network Based Soft-sensing Technique and Its Applications

作　　者：吴晓汉[1] 王金春[1] 王树青[1] 王骥程[1] 

机构地区：[1]浙江大学,杭州310027

出　　处：《自动化与仪器仪表》1998年第2期16-19,共4页Automation & Instrumentation

摘　　要：软测量是一门新兴的工业技术，它借助现代估计理论构造模型推断出工程上难以检测的变量。本文提出了基于径向基函数神经网络（RBFNN）的软测量技术，并且结合工艺机理分析和过程数据关联，对其在轻柴油凝固点软测量的应用进行了研究。结果表明，RBFNN的良好的非线性动态建模能力使其在软测量中具有很大的应用潜力。Soft-sensing is a newly-developed industrial technique which by means of modem modelling technique can provide inferences of variabes hard to be measured. In this paper, radial basis function neural networks (RBFNN )based soft-sensing is proposed. It, combining with the analysis of process principles and correlation of the process data, can be applied to the soft-sensing of the light cycles oil freeze point. The result shows that RBFNN has fine capability of nonlinear dynamic modelljng.

关 键 词：测量技术 神经网络 温度 流量 压力 框图 软测量 

分 类 号：TK31[动力工程及工程热物理—热能工程] TP18[自动化与计算机技术—控制理论与控制工程]