毫米波电路引线楔焊质量智能预测技术研究

Intelligent prediction technology research of wire wedge bonding quality for millimeter wave IC

针对毫米波电路引线楔形焊接工艺优化难题,提出将一种带惩罚函数项的改进BP(Back Propagation,反向传播)神经网络算法用于引线楔形焊接质量智能预测中。通过试验分析了影响楔形焊接质量的关键工艺参数,提取了楔形焊接质量评价指标,基于改进的BP神经网络,建立了引线楔焊质量智能预测模型。研究结果表明,所提出的改进BP神经网络算法合理,且能有效预测工艺参数对引线楔焊质量的影响规律。

Aiming at the process optimization problem of wire wedge bonding for millimeter wave IC, a improved BP （Back propagation） neural network algorithm with penalty function used in the intelligent prediction of wire wedge bonding quality was presented. The key process parameters influencing on the wedge bonding quality were analyzed by experimentation, the evaluation index of quality for wire wedge bonding was established. Based on the improved BP neural network, the intelligent predictable model of wire wedge bonding quality was established. The research results show that the improved BP neural network algorithm is rational, and it can effectively predict the influence disciplinarian of process parameters on wedge bonding quality.