Intelligent Prediction and Optimization of Extraction Process Parameters for Paper-Making Reconstituted Tobacco

To study the impact of extraction process parameters on Baume and solid content of extraction solution in paper-process reconstituted tobacco production, artificial neural network and ensemble learning methods are utilized to build the prediction models for the extraction process in the paper. The prediction models describe the influencing factors such as amount of adding water, extraction water temperature, feed mixing time, squeezing time, centrifuge frequency, squeezing dryness and screw pump frequency on Baume and solid content of extraction solution. It is found that the ensemble learning model is better than the artificial neural network model by the comparison of the prediction results of these two models. The influencing parameters of the solid content and Baume are optimized by genetic algorithm based on the ensemble learning model to improve the product quality of the extraction solution. The experimental results show that the proposed models are helpful to improve the product quality of the paper-making reconstituted tobacco.