.png){kind=icon}
.png){kind=icon}
Return
SGCF: An MPC-Inspired Control Framework Using Sequence Generation Networks for Flotation Process
.png){kind=icon}
PRE
AI
DOI：10.1109/TASE.2025.3599660.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In order to integrate the advantages of Model Predictive Control (MPC) and artificial neural networks, this paper proposes a control framework based on sequence generation (SGCF). The main idea of SGCF is to first generate an ideal subsequent state trajectory conditioned on the current working condition, and then infer feasible control sequences accordingly. To achieve implicit optimization of the performance metric during sequence generation, different learning weights are assigned to labeled samples. A confidence score based on model interaction is further introduced to evaluate the reliability of each inference and assist training. Theoretical analysis is provided to explain how the sequence generation network learns under performance guidance, promoting better trajectories. The effectiveness of SGCF is validated through experiments on inverted pendulum control and lead-zinc rougher flotation reagent control. Multiple baseline methods are considered, including state feedback control (SFC) models and data-driven MPC approaches. Results demonstrate that SGCF achieves advantages in both control performance and cost, while maintaining favorable inference efficiency. Moreover, SGCF exhibits potential for interpretable reasoning. Note to Practitioners–Model predictive control (MPC) is widely used for balancing control performance and cost, but its application to flotation processes faces challenges due to system modeling difficulties and online optimization complexity. On the other hand, conventional neural network control may suffer from overfitting and poor interpretability, making its deployment in industrial settings risky. This paper proposes a control framework that combines MPC principles with sequence generation networks. The approach first generates target adjustment strategies based on current flotation conditions—such as desired froth size or brightness—and then infers the required reagent flow rates to achieve these targets. Beyond providing recommended control actions, the framework also offers users additional feedback on predicted process changes and the confidence of each decision, improving transparency and user trust. Experimental results on both inverted pendulum and flotation reagent control tasks demonstrate that this method can improve both product quality and reagent usage efficiency, while keeping computational requirements practical for real-time deployment. Further research is still needed to explore which types of sequence generation models are most suitable for various scenarios, as well as to develop more user-friendly explanations based on predicted system behavior.
Keywords:
Model predictive control
sequence generation
neural network control
variational autoencoder
flotation process
industrial process control
Journal
IF:
6.4
Papers:
4.9K
Citations:
1.6W