Optimizing the Segment Value of Welch Algorithm by Data Fitting Technique for Double Pulse Welding Signal

Welch algorithm in signal processing is a method used for monitoring the stability of welding current in DC welding function during Gas Metal Arc Welding (GMAW) process. The principal advantage of this method is reducing the number of computations and the required core storage. This method involves sectioning a signal, taking modified Periodograms of each section, averaging these modified Periodograms Feature Extraction from the sensitivity zone was used, but segment value affected the Periodogram modifications such as data detention, running time, and error of raw and prediction data. This study proposes the error function to predict error value between raw data and prediction data from the model in order to select a segment value suitable for Welch algorithm in the analysis of welding current signal from double pulse welding function. The result showed that segment value (S=9) was suitable for Welch algorithm to analyse double pulse welding signal as it reduced the error rate 3.614dB.