Stacking approaches for the design of soft sensors using small data set

In this paper a number of approaches to design a soft sensor for an industrial plant in case of small data set are compared. In particular different strategies to aggregate suboptimal models obtained by bootstrapped neural networks and noise injection are considered. An industrial case of study, consisting in the estimation of the T95% of a Thermal Cracking Unit (TCU) of a refinery in Sicily is considered to evaluate the performance of the different approaches.

[1]  Yu-Lun Chiu,et al.  Dynamic modeling of batch polymerization reactors via the hybrid neural-network rate-function approach , 2007 .

[2]  W. Byrne,et al.  Generalization and maximum likelihood from small data sets , 1993, Neural Networks for Signal Processing III - Proceedings of the 1993 IEEE-SP Workshop.

[3]  S. Graziani,et al.  Selection of regressors using correlation analysis to design a Virtual Instrument for an SRU of a refinery , 2007, 2007 Mediterranean Conference on Control & Automation.

[4]  Thomas E. Marlin,et al.  Multivariate statistical monitoring of process operating performance , 1991 .

[5]  S. Graziani,et al.  Comparing regressors selection methods for the Soft Sensor design of a Sulfur Recovery Unit , 2006, 2006 14th Mediterranean Conference on Control and Automation.

[6]  Luigi Fortuna,et al.  SOFT ANALYSERS FOR A SULFUR RECOVERY UNIT , 2002 .

[7]  Luigi Fortuna,et al.  Soft Sensors for Monitoring and Control of Industrial Processes (Advances in Industrial Control) , 2006 .

[8]  Jules Thibault,et al.  Process modeling with neural networks using small experimental datasets , 1999 .

[9]  Jie Zhang.  Sequential training of bootstrap aggregated neural networks for nonlinear systems modelling , 2002, Proceedings of the 2002 American Control Conference (IEEE Cat. No.CH37301).

[10]  S. Graziani,et al.  Development of a Soft Sensor for a Thermal Cracking Unit using a small experimental data set , 2007, 2007 IEEE International Symposium on Intelligent Signal Processing.

[11]  K. Mondal,et al.  Neural network approximation of iron oxide reduction process , 2005 .

[12]  J. Macgregor,et al.  Development of inferential process models using PLS , 1994 .

[13]  S. Graziani,et al.  Virtual instruments in refineries , 2005, IEEE Instrumentation & Measurement Magazine.

[14]  Chun Chen,et al.  Modeling and optimization for the secondary reaction of FCC gasoline based on the fuzzy neural network and genetic algorithm , 2007 .

[15]  T. Poggio,et al.  Networks and the best approximation property , 1990, Biological Cybernetics.

[16]  Feng Zhang,et al.  Bayesian neural networks for nonlinear multivariate manufacturing process monitoring , 2005, Proceedings. 2005 IEEE International Joint Conference on Neural Networks, 2005..

[17]  Ching-Chih Tsai,et al.  Generalized predictive control using recurrent fuzzy neural networks for industrial processes , 2007 .

[18]  S. Graziani,et al.  Virtual Instruments for the what-if analysis of a process for pollution minimization in an industrial application , 2006, 2006 14th Mediterranean Conference on Control and Automation.

[19]  Luigi Fortuna,et al.  SOFT ANALYSERS FOR A SULFUR RECOVERY UNIT , 2002 .

[20]  Haichao Zhu,et al.  A New Method to Assist Small Data Set Neural Network Learning , 2006, Sixth International Conference on Intelligent Systems Design and Applications.

[21]  B. Kowalski,et al.  Partial least-squares regression: a tutorial , 1986 .

[22]  Michael J. Piovoso,et al.  The Use of Multivariate Statistics in Process Control , 2010 .

[23]  Sohrab Khanmohammadi,et al.  Fault diagnosis competitive neural network with prioritized modification rule of connection weights , 2000, Artif. Intell. Eng..

[24]  Chonghun Han,et al.  A nonlinear soft sensor based on multivariate smoothing procedure for quality estimation in distillation columns , 2000 .

[25]  David H. Wolpert,et al.  Stacked generalization , 1992, Neural Networks.

[26]  Jouko Lampinen,et al.  Bayesian neural networks for industrial applications , 1999, SMCia/99 Proceedings of the 1999 IEEE Midnight - Sun Workshop on Soft Computing Methods in Industrial Applications (Cat. No.99EX269).