Abstract
The expensive facilities of semiconductor run for 24 hour a day in order to keep high utility rate. If any accidents such as power failure happen, it takes lots of time to set up and fix the facilities for reoperation, and results in significant negative effects on yield and production. In most companies, the important machines are protected by generators. Generators keep machines operated continually and reduce loss when power fails. However, the control of generator system is complex, and thus only experts could control generators well. Most staffs cannot make the best decision immediately for lacking experiences or practice. In this research, an expert system was developed including training function and emergency searching function in W factory. Then, an experiment was conducted to verity the effect of the generator training system. Subjects were 22 students in industrial engineering at NTHU. None of the students had experiences about generators before. These subjects were randomly assigned to one of the training conditions. One group used generator training system for training, and the other group received conventional training. The accuracy of test after training and training time were used to evaluate subject’s performance. From the results of the experiment, one could see that there was no significant different between generator training system and conventional training on training time, but generator training system had better performance in accuracy of test. The main, contributions and limitation of this research were discussed. For the future study, the similar design concept can be applied to other complex machines in the factory to improve performance on emergency operation.