Modeling of Limestone Capture Performance During CO2 Looping CyclesBased on BP Neuron Network

摘要

In this work, the limestone sample, with particle size distribution of 38-180 u m, was subjected to 20 calcination/ carbonation cycles in TGA under different reaction conditions so as to providing the basis of neural network. These results indicated that there was a decay tendency of Calcium-based sorbent activity along CO2 looping cycles, moreover, the calcination parameters, such as temperature, duration and atmosphere, were concerned as the influences of carbonation kinetics. Moreover, Artificial Neural Network was proposed as an approach for modeling and mathematical description of the Calcium-based sorbent CO2 looping cycle depending on TGA experimental data. Moreover, BP neural network optimized by LevenbergMarquardt equation, with 6-42-1 topography structure, has been proved to be available of acquiring the carbonation kinetics, even the reaction conditions were set up as extreme ones such as high calcination temperature and prolonged calciantion duration.