“…101 In the literature, data-driven methods also have been published 102 to predict the heating value of coal using artificial nervous network 103 [25][26][27], fuzzy inference system [28], multiple regression method 104 [29], etc. Some of them need input the composition of the coal 105 through the proximate or ultimate analysis into the models 106 [25,28]; the others take several highly relevant process variables 107 as input, such as mass flow rates of the main steam and the reheat 108 steam, coal feeder speed, total air amount, etc. [26,27].…”