Applying machine learning techniques to predict laminar burning velocity for ammonia/hydrogen/air mixtures
Applying machine learning techniques to predict laminar burning velocity for ammonia/hydrogen/air mixtures
复制标题
DOI：
10.1016/j.egyai.2023.100270
复制
发表时间：
2023-05
期刊：
影响因子：
--
通讯作者：
Cihat Emre Ustun;Mohammad Reza Herfatmanesh;A. Medina;A. Paykani
中科院分区：
文献类型：
--
作者：
Cihat Emre Ustun;Mohammad Reza Herfatmanesh;A. Medina;A. Paykani
由于目前的脱碳趋势，氨在内燃机中的利用引起了广泛的兴趣，因为氨是一种零碳燃料，具有与烃不同的燃烧特性。层流燃烧速度（LBV）是燃料的一个基本特性，对燃烧过程有重要影响，在各种混合燃料、压力和流动条件下精确计算和测量层流燃烧速度是一个耗时、复杂的过程。目前研究的主要目标是使用混合机器学习（ML）方法预测NH3/H2/空气混合物的LBV，该方法基于训练数据集，该训练数据集由实验LBV值和通过详细的动力学模型从数值模拟获得的附加数据组成。初始ML模型的训练数据收集从现有的实验LBV在文献中的NH3/H2/空气混合物。然后，使用一维（1D）模拟生成合成数据，以减少数据不均匀性并提高ML模型的准确性。总共测试了24种不同的ML算法，以找到实验数据集和混合数据集的最佳模型。结果表明，高斯过程回归（GPR）和神经网络（NNs）都可以用来预测NH3/H2/空气混合物的LBV与合理的精度。混合ML模型的决定系数为R2 = 0.998。最后，优化混合ML模型超参数以实现R 2= 0.999的决定系数。还发现，ML可以加快LBV计算从9500到27000倍相比，一维模拟与减少的机制。
Ammonia utilisation in internal combustion engines has attracted wide interest due to the current trend toward decarbonisation, as ammonia is a zero-carbon fuel with different combustion properties to hydrocarbons. The laminar burning velocity (LBV) is a fundamental property of fuels with a significant effect on the combustion processes and accurate calculations and measurements of the LBV over a wide range of fuel blends, pressures and flow conditions is a time-consuming, complicated procedure. The main goal of the current study is to predict the LBV of NH 3/H 2/air mixtures using a hybrid machine learning (ML) approach based on a training dataset consisting of both the experimental LBV values and additional data obtained from numerical simulations with a detailed kinetic model. Initial ML model training data is collected from existing experimental LBV in the literature for NH 3/H 2/air mixtures. Then, synthetic data is generated using one-dimensional (1D) simulations to reduce data inhomogeneity and increase accuracy of the ML model. In total, 24 different ML algorithms are tested to find the best model both for the experimental and the hybrid dataset. The results suggest that both Gaussian Process Regression (GPR) and Neural Networks (NNs) can be utilised to predict LBV of NH 3/H 2/air mixtures with reasonable accuracy. The hybrid ML model achieved a coefficient of determination of R 2= 0.998. Finally, hybrid ML model hyperparameters are optimised to achieve a coefficient of determination of R 2= 0.999. It was also found that ML can speed up LBV computation from 9500 to 27000 times compared to 1D simulations with a reduced mechanism.