一种新的软测量方法及其工业应用

软测量是一门新兴的工业技术，它通过数学模型计算得到工程上难以检测的变量值．本文提出了一种基于模糊神经网络的软测量新方法，运用模糊集合论的知识对非线性对象进行局部模型的划分，并采用趋化性网络进行建模和模型校正．经在某炼油厂精密精馏塔的推断控制系统中的应用，表明该方法能够比较精确地建立过程模型，有效地进行模型的在线校正，而且具有灵活、简明的特点．

A New Soft Sensor and Its Industrial Application

The soft sensor is a new industrial technique, which estimates the difficulty to measure variables in engineering through the calaulation of mathematical models. In this paper, a new kind of fuzzy neural net-based soft sensor is proposed, the local models of the nonlinear plants are classified by the fuzzy set theory, and the models are built and corrected through the chemotaxis networks. This soft sensor has been applied to the inferential control system of a high purity distillation column. The results show that the method proposed can build the process models accurately, correct the models on line effectively and has the pro perly of flexibility and simplicity.