Online Energy Consumption Optimization in Electrolytic Copper Foil Electrodeposition via Incremental Soft Actor-Critic

Electrolytic copper foil (ECF) is a fundamental material for the new energy and electronics industries and faces significant challenges due to the high energy consumption (EC) of its electrodeposition. Existing EC optimization methods for ECF electrodeposition are limited to offline operation, yielding optimal process parameter combinations under static conditions. Consequently, these methods exhibit poor dynamic adaptability and are susceptible to performance degradation caused by concept drift during long-term machine operation. To address these limitations, this article proposes an incremental soft actor-critic (I-SAC) method for online EC optimization in ECF electrodeposition. Specifically: 1) an SAC learner and optimizer are constructed to train an actor policy network and determine optimal process parameters, respectively; 2) an incremental module is designed to continuously update the EC model within the objective function and actor policy network within SAC learner using new samples. This enables online knowledge learning and periodic updating of the actor policy network, thereby enhancing the method’s dynamic adaptability. Finally, the proposed method is validated using a dataset collected from industrial sites. Comparative evaluations against other optimization algorithms and reinforcement learning methods demonstrate its superior performance.