Optimization study of reflow soldering profile for Surface Mount Technology

As the last step of the Surface Mount Technology (SMT) production line, Reflow Soldering Process determines the ultimate quality of SMT products, the core of which is the thermal profile. Back Propagation Neural Network (BPNN) is proposed to predict the reflow soldering temperature curve and Genetic Algorithm (GA) is adopted to optimize the profile based on Xu´s paper^[1]. Additional momentum method and double adaptive learning rate adjustment method are adopted in BPNN while multi-point cross and non-uniform mutation operators are used in GA to ameliorate the model constructed by Xu ^[1]. The aim is to reduce the `trial and error´ period so as to save the resources and costs. Numerical studies established by Matlab6.5 verify the effectiveness and practicability of this model.