Applying data mining methodology to establish an intelligent decision system for PCBA process

This paper aims to consider the practical production environment of electronics manufacturing industry firms, and the large quantities of information collected on machine processes, testing data and production reports, while simultaneously taking into account the properties of the processing environment, in conducting analysis to obtain valuable information.

This research constructs a prediction model of the circuit board assembly process yield. A decision tree is used to extract the key attributes. The authors also integrate association rules to determine the relevance of key attributes of undesirable phenomena.

The results assure the successful application of the methodology by reconfirming the rules for solder skip and short circuit occurrence and their causes.

Measures for improvement are recommended, production parameters determined and debugging suggestions made to improve the process yield when the new process is implemented.