Prediction of product yields using fusion model from Co-pyrolysis of biomass and coal

Abstract

This study aimed to establish a self-corrective machine learning model base on co-pyrolysis data of biomass and coal. Proximate and ultimate analysis of raw materials were chosen as input parameters. Radial basis function (RBF), support vector machine (SVM), and random forest (RF) were used to build the base regression models for the fusion (FU) model. 96 sets of the experimental data were applied to train and test the base models. A learning weight were then determined by the predicted performance of base models. Based on the learning weight method, FU model spontaneously regulated and controlled the weight of base models to output the predicted result of co-pyrolysis products. The coefficient of determination (R2) was more than 0.99 and the root-mean-squared error (RMSE) was lower than 0.88%. The results suggested that FU model was more accurately adequate to forecast the yields of co-pyrolysis products than any of the base models.