This study explores the potential of the advanced selective state spaces model(SSSM)in modeling complicated process industries system and proposes the process industry state identification model(PISIM)for controlled p...This study explores the potential of the advanced selective state spaces model(SSSM)in modeling complicated process industries system and proposes the process industry state identification model(PISIM)for controlled prediction of flotation cell pulp level.As a neural system identification model,the PISIM inherits two advantages of the SSSM to address the challenges in identifying flotation systems,including modeling the impact of frequent upstream fluctuations on system states,complex nonlinear physicochemical processes,and long-term dependencies.The first advantage is the ability to capture long-range dependencies,thereby boosting its long-term predictive accuracy.The second lies in the model structure adhering to scaling laws,enabling ongoing enhancements in performance as datasets expand.PISIM is evaluated using a real industrial dataset from a flotation plant at a copper mine in Zambia,with the results demonstrating its theoretical advantages.In a 4.5-hour pulp level prediction task,PISIM outperforms the baseline model by more than 31.34%.Furthermore,a flotation process control simulation experimental system based on PISIM is developed and deployed in a flotation plant in Zambia,assisting engineers in evaluating and optimizing setpoint strategies,ensuring stable production and improving production efficiency.展开更多
基金supported by the National Natural Science Foundation of China(62506031,62332017,U22A2022)National Science and Technology Major Project of the Ministry of Science and Technology of China(2024ZD0608100).
文摘This study explores the potential of the advanced selective state spaces model(SSSM)in modeling complicated process industries system and proposes the process industry state identification model(PISIM)for controlled prediction of flotation cell pulp level.As a neural system identification model,the PISIM inherits two advantages of the SSSM to address the challenges in identifying flotation systems,including modeling the impact of frequent upstream fluctuations on system states,complex nonlinear physicochemical processes,and long-term dependencies.The first advantage is the ability to capture long-range dependencies,thereby boosting its long-term predictive accuracy.The second lies in the model structure adhering to scaling laws,enabling ongoing enhancements in performance as datasets expand.PISIM is evaluated using a real industrial dataset from a flotation plant at a copper mine in Zambia,with the results demonstrating its theoretical advantages.In a 4.5-hour pulp level prediction task,PISIM outperforms the baseline model by more than 31.34%.Furthermore,a flotation process control simulation experimental system based on PISIM is developed and deployed in a flotation plant in Zambia,assisting engineers in evaluating and optimizing setpoint strategies,ensuring stable production and improving production efficiency.
