A Two-Phase Federated Optimization Framework for Process Parameter Settings of Electrodeposition

The process parameter settings of electrolytic copper foil (ECF) electrodeposition face challenges due to insufficient data within individual production domains and barriers to cross-domain data sharing. Coordinating federated predictor training with client-side process parameter search remains a practical challenge in cross-domain ECF production. We propose FedO, a two-phase federated optimization framework for process parameter settings in ECF electrodeposition. Our approach coordinates federated EC predictor training with local process parameter search through an aggregation procedure independent of optimizer internals. The framework operates in two phases: local autonomous optimization, where clients generate high-quality initial solutions, followed by federated collaborative optimization, where global aggregation integrates cross-domain knowledge. Additionally, we incorporate adaptive parameter screening, dynamic learning rate decay, and an elite retention strategy to enhance generalization and stability. Experimental results demonstrate the superiority of the proposed framework over existing methods and its compatibility with the six tested optimizers.