摘要
软测量技术是通过数学模型来估计工程上难以检测的变量值。由于神经网络方法能够描述高度非线性的输入输出关系，因此，基于神经网络的软测量技术已经成为很有吸引力的研究领域，它将辅助变量作为神经网络的输入，将主导变量作为其输出，通过训练网络来实现主导变量在线估计。对基于神经网络的软测量技术进行了综述并详细介绍了神经网络软仪表的结构和方法，给出了神经网络软仪表的系统开发框架，讨论了它在过程控制中的应用，对其发展作了简要的展望。
Soft-sensing techniques provide estimate of engineering variables which are difficult to measure by mathematical model. Because of the ability of neural network to represent highly nonlinear input-output relationships, the soft-sensing techniques based on neural network have attracted much attention, which choose the secondary variables as inputs to the neural network and primary variables as outputs, then train it for on-line estimation. This paper summarizes the soft-sensing techniques based on neural network and introduces the neural network soft-sensor′s structures and methods in detail, the systematically developing frame is given and the applications in process control are discussed,too. The future of development is briefly presented.
出处
《抚顺石油学院学报》
EI
1998年第3期60-64,77,共6页
Journal of Fushun Petroleum Institute
