Article ID Journal Published Year Pages File Type
1139678 Mathematics and Computers in Simulation 2012 8 Pages PDF
Abstract

Soft sensing can be used in the case of watching the variables which are difficult or unable to be measured, or can be measured only with a high cost and significant delays. The key problem for soft sensing is that the method of system identification should meet the accuracy requirement of the real system. An improved T-S fuzzy neural network based on declination compensation is proposed in this paper, which increases the accuracy of system identification by constructing networks of declination compensation. The input of the samples is regarded as the input of the corrected network, the system declinations are regarded as the output samples of the corrected network, the output variables can be compensated by the output of this corrected system dynamically. The testing in catalytic cracking processes demonstrates that the improved T-S fuzzy neural network achieves better results in soft sensing compared with the original network.

Related Topics
Physical Sciences and Engineering Engineering Control and Systems Engineering
Authors
,