| Summary: | 碩士 === 國立高雄科技大學 === 電機工程系 === 107 === The procedure for the development of electronic products includes: circuit design, circuit-diagram drawing, circuit simulation, circuit arrangement, printed circuit board manufacturing, and functionality testing. In the circuit-diagram drawing phase, the circuit diagrams for parts required must be obtained from the parts library of the tool used. When designing a circuit diagram for inclusion in the parts library, the positions of part pins in the circuit diagram must be determined. According to the pin type. Currently, the pin types used are mainly determined by parts-diagram design engineers according to their visual inspection, expertise, and experience. This work is both labor and time consuming. Moreover, misjudgment often occurs due to fatigue of the engineers.
This study proposes a method that entails smart categorization of integrated-circuit pins in a specification table. In this method, by considering the sentences presented in a specification table, the features of words used that describe the pin types and the weightings of these descriptions are determined according to bag of words (BoW) and term frequency–inverse document frequency (TF–IDF), and are integrated with an artificial neural network for pin-type recognition training. An artificial neural network was built to learn the pin type to which a sentence belongs. In this study, descriptions of pin types in specification tables for electronic parts were used for learning, training, and test. The empirical results show that this method yielded satisfactory outcomes.
|
|---|
