摘要
软测量技术在化工生产过程中具有较好的应用前景,适合于监测测量成本高、难于或无法实际测量的过程变量。将改进的T-S模糊神经网络模型引入到软测量建模中,通过偏差校正网络对系统输出量进行动态补偿,可比传统T-S模糊神经网络模型获得更好的系统辨识效果,通过实际测试,软测量结果的均方误差可降低约70%左右。改进的T-S模糊神经网络中由于增加了偏差补偿系统,因此软测量精度获得提高。
Soft sensing has a better application prospect in the measuring of chemical industry process, it is suitable to use in the case of watching the variables which are difficult or unable to be measured, or some of them can be measured, but it implements with a high cost. An improved T-S fuzzy neural network had been introduced to the modeling of soft sensing, and the system output can be compensated by the output of the deviation corrected network dynamically. It achieved a better result than the traditional T-S fuzzy neural network. The MSD of soft sensing can be decreased 70% by testing in practice. The improved T-S fuzzy neural network can achieve a better performance by using the additional deviation compensation system in the model structure.
出处
《电子测量与仪器学报》
CSCD
2010年第6期585-589,共5页
Journal of Electronic Measurement and Instrumentation
