论文信息 - PRIMJENA SOFTVERSKOG ANALIZATORA U UNAPREĐENJU KVALITETE PROIZVODA I VOĐENJA PROCESA HIDRODESULFURIZACIJE

PRIMJENA SOFTVERSKOG ANALIZATORA U UNAPREĐENJU KVALITETE PROIZVODA I VOĐENJA PROCESA HIDRODESULFURIZACIJE

Today, we are witnessing the necessity to produce high quality fuels in our refineries. To meet the market needs, refining processes must be equipped with a satisfactory number of process analyzers, which is not always easy, because of their price. In the past ten years, there has been a significant number of software sensors present in the process industry. Their application is based on process model built from process data in order to continuously, on-line, predict physical and chemical properties or product content and improve process monitoring and control. In this paper, practical application of software sensor based on neural networks was explored on the actual case of sulfur content in the hydrodesulphurization unit product, LGO. The research results shall be applied in practice. I. Lukec, D. Lukec, D. Klaric Softverski analizator... goriva i maziva, 45, 6 : 353-368, 2006. 36

Ivana Lukec | Darko Lukec | Darko Klarić

[1] A. Barsamian,et al. Inferential property predictors using neural networks , 1998 .

[2] W. Zhong,et al. Improve nonlinear soft sensing modeling by combining multiple models : FMM-based methods are shown to give better results , 2000 .