| Summary: | 碩士 === 國立雲林科技大學 === 機械工程系碩士班 === 93 === The quality of spot welding could be grouped into destructive test and nondestructive test. Breaking work-pieces to examine the quality of spot welding by visual inspection is the most common destructive test, but it takes a lot of time and cost. The nondestructive test is time- efficient and cheap. However it needs experienced technicians and high facility investment. In order to reduce cost and improve user-interface, the selected features of impedance waveform, which can be computed measured voltage and current signals during the welding process, are used for monitoring the welding quality( Shih-Fu Ling & Lixue Wan 2000，[1]).
In time research, appropriate training procedures are performed to obtain the weights at all neurons for the constructed neural network. The performance of quality monitoring through the well trained neural network is discussed with respect to welding environment, material property and parameters set during welding process. In addition, total energy bound and impedance waveform bound are also used together for monitoring the welding quality. The results are compared with those obtained through neural network.
Keyword：spot welding、ANN、energy bound、impedance bound、quality of welding
|
|---|
