Te-Jen Su, Yi-Feng Chen, Jui-Chuan Cheng, Chien-Liang Chiu
The wafer dicing saw is a critical process of the semiconductor IC assembly manufacturing, where chipping and cracking can damage internal circuits and result in device failure. This article applies back propagation neural networks to learn and analyze data transmitted from the machine sensors for predictive purposes. Experimental results show that neural network processing provides prediction accuracy of 75%. The results can be monitored during real-time machine operation. If predictive values exceed a certain specification, an alarm can be triggered to prevent product loss, thus increasing production efficiency.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados