In this study, the gross calorific value (GCV) of coal was accurately and rapidly determined using eight artificial intelligence models based on big data of 2583 observations of coal samples in the Mong Duong underground coal mine (Vietnam). Accordingly, the volatile matter, moisture, and ash were considered as the key variables (inputs) for determining GCV. Seven artificial neural network (ANN) models were developed to estimate GCV as the first stage. Subsequently, the best ANN model (with the highest performance) was selected as the initial ANN model for the optimization process, i.e., ANN 3-12-9-1 model. The particle swarm optimization (PSO) algorithm was applied to perform a global search for the optimal weights/biases of the selected ANN model. This novel procedure is denoted as PSO-ANN. A variety of performance metrics was used to assess the quality of the training process, as well as the models’ performance in the testing dataset. The results revealed that the models developed in this study could determine GCV rapidly and accurately. Of those, the PSO-ANN model provided the highest accuracy in estimating GCV of coal with a root-mean-squared error of 182.476, the correlation coefficient of 0.964, the variance accounted for of 96.411, and mean absolute percentage error of 0.016. Besides, the analyzed and compared results also indicated that the PSO algorithm played a significant role in improving the accuracy of the ANN model. It was introduced as an alternative solution to determine the GCV of coal in practical engineering rapidly.

在这项研究中，基于8个人工智能模型，根据孟都地下煤矿（越南）的2583个煤样观测大数据，准确，快速地确定了煤炭的总发热量（GCV）。因此，挥发性物质，水分和灰分被认为是确定GCV的关键变量（输入）。开发了七个人工神经网络（ANN）模型来估计GCV作为第一阶段。随后，选择最佳的ANN模型（具有最高性能）作为优化过程的初始ANN模型，即ANN 3-12-9-1模型。应用粒子群优化（PSO）算法对所选ANN模型的最佳权重/偏差执行全局搜索。这种新颖的过程称为PSO-ANN。各种性能指标用于评估训练过程的质量以及测试数据集中模型的性能。结果表明，本研究开发的模型可以快速，准确地确定GCV。其中，PSO-ANN模型以182.476的均方根误差，0.964的相关系数，96.411的方差和0.016的平均绝对百分比误差提供了最高的煤GCV估计精度。此外，分析和比较结果还表明，PSO算法在提高ANN模型的准确性方面发挥了重要作用。引入它是在实际工程中快速确定煤炭GCV的替代解决方案。结果表明，本研究开发的模型可以快速，准确地确定GCV。其中，PSO-ANN模型以182.476的均方根误差，0.964的相关系数，96.411的方差和0.016的平均绝对百分比误差提供了最高的煤GCV估计精度。此外，分析和比较结果还表明，PSO算法在提高ANN模型的准确性方面发挥了重要作用。引入它是在实际工程中快速确定煤炭GCV的替代解决方案。结果表明，本研究开发的模型可以快速，准确地确定GCV。其中，PSO-ANN模型以182.476的均方根误差，0.964的相关系数，96.411的方差和0.016的平均绝对百分比误差提供了最高的煤GCV估计精度。此外，分析和比较结果还表明，PSO算法在提高ANN模型的准确性方面发挥了重要作用。引入它作为在实际工程中快速确定煤炭GCV的替代解决方案。相关系数为0.964，方差占96.411，平均绝对百分比误差为0.016。此外，分析和比较结果还表明，PSO算法在提高ANN模型的准确性方面发挥了重要作用。引入它作为在实际工程中快速确定煤炭GCV的替代解决方案。相关系数为0.964，方差占96.411，平均绝对百分比误差为0.016。此外，分析和比较结果还表明，PSO算法在提高ANN模型的准确性方面发挥了重要作用。引入它作为在实际工程中快速确定煤炭GCV的替代解决方案。