The chemical percolation devolatilization (CPD) model has been shown to represent the devolatilization process of different coals and heating conditions with good accuracy. However, its use in computational fluid dynamics is limited because of its relatively high computational cost. Here, an Artificial Neural Network (ANN) based model for predicting coal devolatilization kinetics is developed based on a database constructed with the CPD model for a wide range of coals and heating rates. The heating rates and the information of ultimate and proximate analysis are chosen as inputs of the ANN model to consider the effects of coal types and heating conditions on coal devolatilization; the outputs are the kinetic parameters for the two-step kinetic model. The learning, validation, and application results show that the proposed ANN model has a competitive prediction capability on both the total volatile release and release rates when compared with the CPD model, but has obvious computational efficiency advantages. Furthermore, the relative impact of the coal type and heating rate on each kinetic parameter for coal devolatilization is quantitatively evaluated through the Garson equation. It is found that the heating rate has the strongest effect on the pre-exponential factor, while the coal types show significant influence on the activation energy and final yield of the two reactions in the two-step model.

化学渗滤脱挥发分（CPD）模型已被证明可以很好地表示不同煤种和加热条件下的脱挥发分过程。然而，由于其相对较高的计算成本，其在计算流体动力学中的使用受到限制。在此，基于使用人工神经网络（ANN）预测煤炭挥发挥发动力学的模型，该模型是基于使用CPD模型构建的数据库来开发的，该数据库适用于各种煤炭和加热速率。选择加热速率以及最终分析和邻近分析的信息作为ANN模型的输入，以考虑煤类型和加热条件对脱挥发分的影响；输出是两步动力学模型的动力学参数。学习，验证，应用结果表明，与CPD模型相比，所提神经网络模型在总挥发物释放速率和释放速率上均具有竞争性预测能力，但具有明显的计算效率优势。此外，通过Garson方程定量评估了煤类型和加热速率对煤脱挥发分的每个动力学参数的相对影响。发现在两步模型中，升温速率对指数前因子的影响最大，而煤种对两个反应的活化能和最终收率表现出显着影响。通过Garson方程定量评估了煤类型和加热速率对煤脱挥发分的每个动力学参数的相对影响。发现在两步模型中，升温速率对指数前因子的影响最大，而煤种对两个反应的活化能和最终收率表现出显着影响。通过Garson方程定量评估了煤类型和加热速率对煤脱挥发分的每个动力学参数的相对影响。发现在两步模型中，升温速率对指数前因子的影响最大，而煤种对两个反应的活化能和最终收率表现出显着影响。