Modeling nonlinear dynamics of circulating fluidized beds using neural networks--颗粒学报PARTICUOLOGY

Chen, W., Tsutsumi, A., Lin, H., & Otawara, K. (2005). Modeling nonlinear dynamics of circulating fluidized beds using neural networks. China Particuology, 3(1), 84-89. https://doi.org/10.1016/S1672-2515(07)60172-9

Modeling nonlinear dynamics of circulating fluidized beds using neural networks

Wei Chen ^a, Atsushi Tsutsumi ^a *, Haiyan Lin ^a, Kentaro Otawara ^b

a Department of Chemical System Engineering, The University of Tokyo, Tokyo 113–8656, Japan
b Project Management Division, Kureha Technol. Eng. Co., Iwaki 974–8232, Japan

10.1016/S1672-2515(07)60172-9

Volume 3, Issues 1–2, April 2005, Pages 84-89

Received 22 March 2005, Accepted 29 March 2005, Available online 14 December 2007.

E-mail: tsutsumi@chemsys.t.u-tokyo.ac.jp

Highlights

Abstract

In the present work, artificial neural networks (ANNs) were proposed to model nonlinear dynamic behaviors of local voidage fluctuations induced by highly turbulent interactions between the gas and solid phases in circulating fluidized beds. The fluctuations of local voidage were measured by using an optical transmittance probe at various axial and radial positions in a circulating fluidized bed with a riser of 0.10 m in inner diameter and 10 m in height. The ANNs trained with experimental time series were applied to make short-term and long-term predictions of dynamic characteristics in the circulating fluidized bed. An early stop approach was adopted to enhance the long-term prediction capability of ANNs. The performance of the trained ANN was evaluated in terms of time-averaged characteristics, power spectra, cycle number and short-term predictability analysis of time series measured and predicted by the model.

Graphical abstract

Keywords

circulating fluidized beds (CFBs); hydrodynamics; nonlinear dynamics; ANNs; modeling

文章导读