Title :
Implementation of nonlinear inferential sensors
Author :
Neelakantan, Ramesh
Author_Institution :
Aspen Technol. Inc., Pittsburgh, PA, USA
Abstract :
In industrial applications of inferential sensing, the most popular technology deployed is neural network technology. However, the industry needs more than one modeling option for building inferential sensors. Statistical tools such as partial least squares (PLS) must be provided to the user as options. Use of fuzzy functions and a hybrid of PLS and neural network are also becoming popular because of the robustness in predictions. First principle models are also used where data is not reliable. Successful implementation of an inferential sensor project requires a range of choices in modeling tools and good project engineering. This paper focuses on empirical model based inferential sensor and the implementation steps to accomplish a successful inferential sensing project
Keywords :
data acquisition; fuzzy set theory; least squares approximations; neural nets; parameter estimation; principal component analysis; fuzzy functions; inferential sensors; neural network; partial least squares; principal component analysis; Artificial neural networks; Cost benefit analysis; Extrapolation; Fuzzy neural networks; Input variables; Laboratories; Least squares methods; Mathematical model; Neural networks; Predictive models;
Conference_Titel :
American Control Conference, 1999. Proceedings of the 1999
Conference_Location :
San Diego, CA
Print_ISBN :
0-7803-4990-3
DOI :
10.1109/ACC.1999.782338
