论文部分内容阅读
Combustion kinetic parameters(i.e.,activation energy and frequency factor) of coal have been proven to relate closely to coal properties;however,the quantitative relationship between them still requires further study.This paper adopts a support vector regression machine(SVR) to generate the models of the non-linear relationship between combustion kinetic parameters and coal quality.Kinetic analyses on the thermo-gravimetry(TG) data of 80 coal samples were performed to prepare training data and testing data for the SVR.The models developed were used in the estimation of the combustion kinetic parameters of ten testing samples.The predicted results showed that the root mean square errors(RMSEs) were 2.571 for the activation energy and 0.565 for the frequency factor in logarithmic form,respectively.TG curves defined by predicted kinetic parameters were fitted to the experimental data with a high degree of precision.
Combustion kinetic parameters (ie, activation energy and frequency factor) of coal have been proven to relate closely to coal properties; however, the quantitative relationship between them still requires further study. This paper adoption a support vector regression machine (SVR) to generate the models of the non-linear relationship between combustion kinetic parameters and coal quality. Kintic analyzes on the thermo-gravimetry (TG) data of 80 coal samples were performed to prepare training data and testing data for the SVR.The models developed were used in the estimation of the combustion kinetic parameters of ten testing samples. The predicted results showed that the root mean square errors (RMSEs) were 2.571 for the activation energy and 0.565 for the frequency factor in logarithmic form, respectively. Graphs defined by predicted kinetic parameters were fitted to the experimental data with a high degree of precision.