Investigation on the pyrolysis process, products characteristics and BP neural network modelling of pine sawdust, cattle dung, kidney bean stalk and bamboo
Investigation on the pyrolysis process, products characteristics and BP neural network modelling of pine sawdust, cattle dung, kidney bean stalk and bamboo
复制标题
DOI：
10.1016/j.psep.2022.04.055
复制
发表时间：
2022-04-30
影响因子：
7.8
通讯作者：
Geng, Liyan
中科院分区：
文献类型：
--
作者：
Li, Jishuo;Yao, Xiwen;Geng, Liyan
为实现废弃物的资源化利用，减轻环境污染，对松木屑(PS)、牛粪(Cd)、菜豆秸秆(KS)和竹子(BA)的热解行为进行了研究。用热重-质谱仪分析了四种物质在热解过程中的质量损失、气相产物演化和动力学参数。根据TG-MS结果，建立了预测不同生物质热解质量损失的BP神经网络模型。更重要的是，利用FTIR和SEMEDX对生物油和生物炭的性质进行了分析，为进一步利用这些热解产物提供了便利。结果表明，PS在800℃时的热裂解失重率最高(87.25%)，较高的D值(1.23E-05)表明PS比其他材料更容易分解。在气体产物方面，PS在热解过程中产生的H-2、C2H6、C(3)H(8)和CO2比其他物质多，而BA在热解过程中产生更多的CH4和H2O。此外，PS生物油中酚类或芳香族化合物的含量最高，制得的PS生物炭表面气孔均匀整齐，证明了PS生物油具有较高的利用价值。最终，所建立的BP神经网络模型实现了令人满意的随温度升高的质量损失预测性能。(C)2022化学工程师学会。爱思唯尔有限公司出版。保留所有权利。
To realize resource utilization of waste and alleviate associated environmental pollution, the pyrolysis behaviour of pine sawdust (PS), cattle dung (CD), kidney bean stalk (KS) and bamboo (BA) was investigated. The mass loss, gaseous product evolution and kinetic parameters of these four materials during pyrolysis were analysed via TG-MS. According to the TG-MS results, a back propagation (BP) neural network model was developed for the mass loss prediction of different biomass pyrolysis. More importantly, FTIR and SEMEDX were used to analyse the characteristics of bio-oil and biochar to facilitate further utilization of these pyrolysis products. The results indicated that PS exhibited the highest mass loss (87.25%) during pyrolysis at 800 C, and the higher D value (1.23E-05) indicated that PS was more easily decomposed than other materials. In terms of gaseous products, PS produced more H-2, C2H6, C(3)H(8 )and CO2 than did the other materials during pyrolysis, while BA produced more CH4 and H2O. In addition, the content of phenols or aromatic compounds in PS bio-oil was the highest, and the surface pores in the obtained PS biochar were uniform and regular, which verified that PS achieved a higher utilization value than that of the other materials considered. Finally, the established BP neural network models realized a satisfactory mass loss prediction performance with increasing temperature.(C) 2022 Institution of Chemical Engineers. Published by Elsevier Ltd. All rights reserved.