The pyrolysis behaviors, reaction kinetics, thermodynamic parameters, and reaction mechanism models of tobacco straw (TOS) were determined using a three-parallel Gaussian reaction model. The apparent activation energies for the pyrolysis of pseudo cellulose (P1-HE), hemicellulose (P2-CE) and lignin (P3-LI) pyrolysis were determined using the Ozawa-Flynn-Wall, Kissinger-Akahira-Sunose, Starink, Tang, and Distributed Activation Energy Model model-free methods. The average apparent activation energies for P1-HE, P2-CE, and P3-LI were 153.67, 178.53, and 190.45 kJ/mol, respectively. The pyrolysis process was well described by the dimensional, order, diffusional, and power law reaction mechanism models. Twelve types of pyrolysis products, including benzenes, acids, phenols, furans, indoles, pyridines, hydrocarbons, pyrroles, ketones, alcohols, quinolines, and aldehydes, were identified as the main pyrolysis products. Combining Raman and SEM analyses, pyrolyzed biochar at 800 °C was suitable as a catalyst or catalyst carrier. An artificial neural network (ANN) was utilized to predict the thermal degradation behaviors of TOS pyrolysis, and the hyperparameters of the activation function were optimized. The best ANN model structure was ANN (5 *11 *1). This study provided theoretical and practical guidance for further rational utilization of TOS waste.

使用三平行高斯反应模型确定了烟草秸秆 (TOS) 的热解行为、反应动力学、热力学参数和反应机理模型。使用 Ozawa-Flynn-Wall、Kissinger-Akahira-Sunose、Starink、Tang、分布式活化能模型无模型方法。P1-HE、P2-CE 和 P3-LI 的平均表观活化能分别为 153.67、178.53 和 190.45 kJ/mol。热解过程通过维度、有序、扩散和幂律反应机制模型得到很好的描述。十二类裂解产物，包括苯类、酸类、酚类、呋喃类、吲哚类、吡啶类、烃类、吡咯类、酮类、醇类、喹啉和醛被确定为主要的热解产物。结合拉曼和 SEM 分析，在 800 °C 下热解的生物炭适合作为催化剂或催化剂载体。利用人工神经网络 (ANN) 预测 TOS 热解的热降解行为，并优化激活函数的超参数。最好的ANN模型结构是ANN(5 *11 *1)。该研究为进一步合理利用TOS废弃物提供了理论和实践指导。并且优化了激活函数的超参数。最好的ANN模型结构是ANN(5 *11 *1)。该研究为进一步合理利用TOS废弃物提供了理论和实践指导。并且优化了激活函数的超参数。最好的ANN模型结构是ANN(5 *11 *1)。该研究为进一步合理利用TOS废弃物提供了理论和实践指导。