Production capacity assessment and carbon reduction of industrial processes based on novel radial basis function integrating multi-dimensional scaling

Abstract

The high energy consuming and carbon emission of complex industrial processes has become a huge challenge all over the world. Therefore, a novel production capacity assessment and carbon reduction modeling technology of industrial processes is presented in this paper by using radial basis function (RBF) neural network integrating multi-dimensional scaling (MDS) (MDS-RBF). The factors affecting the production capacity can be obtained in the low-dimensional space by using the MDS method, which can maintain the same distance in the original multi-dimensional space. Then the main influencing factors are obtained by analyzing the top largest eigenvectors in low-dimensional space. After, the main influencing factors are set as training sets of the RBF to build the production capacity assessment and carbon reduction model. The factors after the MDS process can decouple the original factors effectively, which greatly improving the prediction performance of the RBF. Compared with the RBF, back propagation (BP) neural network, and BP integrating MDS (MDS-BP), the prediction accuracy of the presented model is verified by University of California Irvine datasets (UCI). At last, the introduced modeling technology is applied to the ethylene production processes, which is great helpful to the energy optimization and carbon emission analysis.