共晶贴片机传送系统故障预示研究

Fault Prediction of Eutectic Mounter Transport System

针对共晶贴片机工作中因传送系统故障所造成的管座引脚折弯或者漏贴现象,提出了基于主成分分析(PCA)与KL散度(KLD)相结合的潜在故障预测法。由于贴片机振动信号小波时频图成分复杂,故利用PCA对其降维,并引入轮廓似然率保证识别灵敏度。通过KLD对各主成分轴上潜分数分布差异进行衡量,找出共晶贴片机传动导轨出现阻滞的故障早期征兆。该方法对制定设备维护保养周期及提高芯片封装良品率具有极其重要的意义。

During the long working hours of eutectic mounter,the transport system′s malfunction leads to socket pin bending or paste leakage.A potential fault prediction method based on principal component analysis（PCA）and Kullback-Leibler divergence（KLD）is proposed.Because of complexity of the eutectic mounter vibration signal wavelet time-frequency graph,PCA is used to reduce the dimensionality of the data.The profile likelihood is used to ensure the sensitivity of recognition.KLD is used to measure the difference of the latent score distribution of each principal component axis.An early sign of the transport guide rail failure is found.This method is of great significance to formulate equipment maintenance period and improve the yield rate of chip packaging.