Performance prediction of SMT production lines based on digital twin technology

In the electronics manufacturing industry, the demand for performance optimization of surface mounted technology (SMT) production lines has been continuously increasing. Traditional prediction methods are often inadequate for handling the complexity arising from multi-equipment collaboration, multi-process coupling, and dynamic interactions among multiple variables. In this study, the performance prediction problem of SMT production lines is investigated based on digital twin theory. Firstly, focusing on SMT production lines, a general influencing factor system is established, including processing equipment quantity, logistics conditions, product characteristics, and storage factors. Subsequently, simulation experiments under different production line configurations are conducted using FlexSim simulation software. Processing rate, idle rate, blocking rate, and throughput are extracted as key indicators for production line performance prediction. The dataset is then expanded through the SMOTE oversampling technique. Finally, a performance pattern classification model is constructed based on the K-means++ algorithm to identify and predict production line performance states under different configuration conditions. The results indicate that performance patterns are closely related to equipment scale, logistics resource allocation, and spatial layout. The proposed approach provides analytical support for configuration optimization and performance improvement of SMT production lines.