.png){kind=icon}
.png){kind=icon}
Return
Machine learning-based multi-objective optimization framework for industrial black nickel electroplating
.png){kind=icon}
PRE
AI
DOI：10.1007/s10845-025-02573-w.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The optimization of industrial processes, such as the industrial black nickel electroplating (IBNE) process, is challenging due to the intricate structure and dynamics involved. Machine learning (ML) methods are proposed to learn from historical data to address this issue. In this study, we propose a novel intelligent process optimization framework based on ML methods to optimize the IBNE process through a sim-to-real approach. The framework consists of a virtual IBNE environment simulator and a deep reinforcement learning-based optimization architecture. The virtual IBNE environment simulator is designed to address three objectives: lightness, uniformity and plating rate, based on a historical plating dataset. Lightness and uniformity are considered in one defect detection problem, where a sample is classified as Pass if it meets both criteria, and Fail otherwise. In addition, a reward function is formulated to evaluate the plating performance of samples, with the penalty term derived by solving a constrained polynomial optimization problem based on constraints extracted from the dataset. A deep deterministic policy gradient (DDPG) algorithm is presented to learn the optimal current density corresponding to different plating conditions, ensuring that the plating process can achieve optimal performance under specific conditions. Finally, we apply the policy learned in the virtual IBNE environment to a real-world application, and results from laboratory experiments validate the effectiveness of the proposed framework.
Keywords:
Black nickel electroplating
Machine learning
Industrial modeling
Deep reinforcement learning
Process optimization
Journal
IF:
7.4
Papers:
3.5K
Citations:
1.1W