Optimization of Key Process Parameters for Surface Mount Technology Production Line Based on Improve2d Proximal Policy Optimization Algorithm

The optimization of key process parameters on the surface mount technology (SMT) production line plays a significant role in improving the product quality qualification rate. Existing research methods involve identifying the mapping relationship between key process parameters and product quality through vast amounts of historical data, and then determining the optimal process parameters through optimization algorithms. However, in actual production environments, product quality is affected by factors that cannot be easily quantified, such as the quality of solder paste and environmental conditions, resulting in continuous changes to the optimal process parameters. Therefore, when optimizing key process parameters, it is necessary to make real-time adjustments based on the actual production conditions of the production line. This paper proposes an improved Proximal Policy Optimization (PPO) algorithm, Firstly, the Actor network and Critic network in the PPO algorithm are replaced with a ID-CNN-GRU-Attention network structure. Secondly, the state space, action space, and reward function are designed based on the actual operating conditions of the SMT production line. Finally, the improved PPO algorithm is evaluated from three aspects: reward return, robustness, and success rate. Compared to the Actor-Critic algorithm, the improved PPO algorithm may have some limitations in terms of robustness, but it shows significant improvements in reward return and accuracy. Specifically, the reward return is increased by 76.22% and the accuracy is improved by 26.16% compared to the Actor-Critic algorithm. When compared to the standard PPO algorithm, the improved PPO algorithm demonstrates an increase in reward return by 29.31%, a reduction in variance by 2.4%, and an improvement in accuracy by 8.66 %.