.png){kind=icon}
.png){kind=icon}
Return
Multiple Distribution Adaptive Collaborative Transfer Self-Learning Control Based on Joint Perception
.png){kind=icon}
PRE
AI
DOI：10.1109/TIE.2025.3634441.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In complex industrial processes, due to the variety of coupled factors, industrial processes often operate in a state of frequent fluctuations. Precise control of dynamic operating conditions is critical to the optimization of industrial process operations. Predictive control is an important method for controlling industrial processes under constraints. However, if operating conditions are not immediately detected and an accurate model is not quickly established, the control strategy may become mismatched with operating conditions, leading to a decrease in control performance. To achieve precise control under dynamic operating conditions, this article proposes multiple distribution adaptive collaborative transfer self-learning control method. This method enables real-time detection of operating condition changes under multifactor coupling and rapid identification of prediction model, thereby ensuring precise control across operating conditions. Specifically, a joint perception strategy based on distribution shift and model drift is proposed to enable timely detection of operating condition changes. Then, a distribution difference metrics based on fuzzification is proposed, which accurately identifies the potential causes of operating condition changes and provides direction for operating condition transfer. Finally, to achieve rapid identification of the model under new operating conditions, a multiple distribution adaptive collaborative transfer method is proposed, ensuring precise control across operating conditions. To verify the superiority of the proposed method, experiments are conducted on the Hammerstein system and industrial systems. Experimental results show that the proposed method achieves better control performance across operating conditions.
Keywords:
Collaborative transfer
distribution difference
joint perception
multifactor coupling
operating condition
Journal
IF:
7.2
Papers:
1.8W
Citations:
9.8W