摘要
目的:1.从辐射能利用角度出发,探究一维固定床煤粉富氧燃烧的辐射能流特性,为固体燃料燃烧能量分质分级转化应用提供参考;2.对比研究半经验模型与人工神经网络模型这两种建模方法,为人工神经网络模型在后续研究中的应用提供参考。创新点:1.提出燃烧光热能量分级转化的概念,为燃烧光热能量分质分级转化系统提供研究基础;2.从辐射能量利用的角度研究煤粉燃烧的辐射能流特性;3.不局限于实验报告,基于实验数据探究2种建模方法,揭示神经网络模型的优势。方法:1.在一维管式炉反应器上进行实验,探究不同燃烧条件下煤粉富氧燃烧的辐射能流特征;2.基于辐射传热理论,通过半经验模型描述煤粉在固定床中燃烧的辐射能流;3.训练神经网络模型来描述实验结果,通过对比2种方法来揭示神经网络模型在预测结果方面的优势。结论:1.固定床煤燃烧过程中的挥发分及煤烟会降低辐射能;可利用低挥发分燃料以及增大氧浓度来提高火焰辐射能比例。2.较高的燃烧温度是提升燃烧辐射能比例最重要的因素;实践中可以通过采用高热值燃料以及烟气回热等方法来提高燃烧温度。3.多联产半焦燃烧辐射能比例高于原煤;可通过煤热解多联产技术与半焦燃烧光热能量分级利用相结合的方式构成新的煤炭高效清洁利用系统。4.人工神经网络不但可以对实验结果进行建模,还能够很好地预测未知工况结果,因此值得在更多的后续研究中使用。
This paper describes the radiative energy flux characteristics of fixed-bed oxy-coal combustion for the purpose of guiding the quality-splitting conversion of combustion energy.An experiment was performed in a tube furnace at a temperature range of 800–1200℃ in O2/N2 and O2/CO2 atmospheres,and the radiative intensity was measured.It was found that an increase in oxygen concentration and temperature could increase the radiative intensity more than 1.5 to 2 fold during combustion,and the radiative energy flux was higher for semi-coke than coal by about 16%–27%.The radiative energy results could be described by a semi-empirical model and an artificial neural network(ANN) model.The results showed that the errors of the ANN were less than 0.01%,and demonstrated the superiority of the ANN.This study provides guidance for subsequent research on quality-splitting conversion of combustion energy.
基金
Project supported by the China Scholarship Council(No.201806320236)
the Zhejiang University Academic Award for Outstanding Doctoral Candidates(No.2018071)
the Ningxia Provincial Key Research and Development Program of China(No.2018BCE01004)
关键词
辐射能流
固定床
富氧燃烧
人工神经网络
能量分质分级转化
Radiative energy flux
Fixed-bed
Oxy-coal combustion
Artificial neural network(ANN)
Energy quality-splitting conversion