Machine Learning Algorithm Method Research Articles

“Thermal-dissolution based carbon enrichment” treatment of biomass: Modeling and kinetic study via combined lumped reaction model and machine learning algorithm

The homogenizing pyrolytic method denoted as the “Thermal-dissolution based carbon enrichment (TDCE)” was effective in converting different biomasses into value-added carbon products. However, the product yields from the TDCE process varied greatly despite the biomasses undergoing similar reactions. Consequently, a generalized kinetic study is essential in determining the different kinetic parameters of each biomass. In this study, the lumped reaction model was coupled with machine learning algorithm (ML-LRM) for the general acquisition of these kinetic parameters. 304 yield data of cellulose under different conditions were accurately predicted by the machine learning algorithm and were further utilized for subsequent kinetic modeling. In addition, the derived conversion pathways were quite reliable, as the one from predicted data was similar to that calculated from experimental data (R2 = 0.95) under experimental conditions, but with higher accuracy (R2 > 0.99). Moreover, a novel five-stage reaction mechanism was proposed to quantitatively and qualitatively describe the TDCE process based on the kinetic studies and the pre-proposed reaction mechanism. Consequently, the ML-LRM method was proven to be effective in elucidating the main reactions at any temperature throughout the TDCE treatment and can therefore provide guidance towards directed regulation and practical application of said products.

Identifying atmospheric pollutant sources using a machine learning dispersion model and Markov chain Monte Carlo methods

Estimating the sources of contaminant or hazard emissions is important for pollution control and safety management. Markov chain Monte Carlo (MCMC), combined with Bayesian inference, was used to identify the source terms of pollutants. However, the efficiency and accuracy of the forward dispersion model greatly impacted the performance of the estimation method. Therefore, a machine learning algorithm (MLA) model with high prediction accuracy and efficiency was proposed and coupled with MCMC method to estimate the source terms. A previously proposed MLA model was used to obtain the expected concentrations in Bayesian estimation. The Delayed Rejection Adaptive Metropolis (DRAM) method was applied to sample particles in order to form Markov chains. To evaluate the performance of the MCMC–MLA method, a Gaussian dispersion model was selected as the forward model. The performances of MCMC–MLA and MCMC–Gaussian models were then compared with release cases in Prairie Grass experiment and the results showed that the MCMC–MLA method converged more rapidly than the MCMC–Gaussian model. Nevertheless, release cases in the Round Hill experiment were also used to test the generalisability of the MCMC–MLA. The results indicated that the performance of MCMC–MLA was better than that of the MCMC–Gaussian model for estimating source terms in estimation accuracy. Hence, the MCMC–MLA method proposed here is potentially a useful tool for identifying emissions source parameters with high accuracy and efficiency, as well as reasonable probability estimates.

Using machine learning algorithms to predict the pressure drop during evaporation of R407C

The calculation of the pressure drop for two-phase flow in evaporation and condensation processes is required by a variety of design practices. In recent years, many correlations were developed in order to determine the pressure drop for two-phase flow. This process needs many experimental tests. Hence, in this study, it is proposed to apply machine learning algorithms (MLAs) to forecast the pressure drop for two-phase flow of R407C. Three methods of MLAs are developed with the purpose of pressure drop prediction in a smooth horizontal copper tube, for 4.5 mm and 8 mm inner diameter. These methods are multilayer feed-forward neural network (MLFFNN), support vector regression (SVR), and group method of data handling (GMDH) type neural network. Mass flux, tube diameter, saturation pressure, and vapor quality of the refrigerant are used as input variables of the models and the target is selected to be the pressure drop of evaporation. The results show that although the developed models can successfully predict the pressure drop of two-phase flow, MLFFNN and GMDH models outperform the SVR model in term of the correlation coefficient close to 1.