Deep correlation and precise prediction between static features of froth images and clean coal ash content in coal flotation: An investigation based on deep learning and maximum likelihood estimation

Abstract

This research combines deep learning with sophisticated likelihood analysis to achieve precise detection of clean coal ash content in the domain of froth flotation through image acquisition and processing. The utilization of maximum likelihood estimation method establishes a robust correlation between static features of froth images and clean coal ash content, while the application of the BFGS algorithm efficiently searches for optimal parameter estimates. Furthermore, a deep neural network model is constructed using Keras to accommodate multi-feature and hybrid data inputs. The model's performance is rigorously evaluated using metrics such as mean squared error, mean absolute error, and root mean squared error. Experimental results demonstrate remarkable accuracy, with the mean squared error, mean absolute error, and root mean squared error on the test set being 0.003475 %, 0.044740 %, and 0.044585 %, respectively.