Wavelet Transformation for Hand-Motion Signal Analysis of TIG Welder Performance

The welder performance is critically importance to obtain the weld quality. The objective of this research is to use the data from the Myo hand motion sensor to evaluate the welder performance in tungsten inert gas (TIG) welding process. The data of the acceleration and gyroscope from the Myo hand motion sensor have been studied to assess the performance of the welder. In this study, the Euclidean Norm has been used to combine the three axes of accelerometer and gyroscope. The three-axis signal of accelerometer and gyroscope are combined respectively and a relation was found. This enables to differentiate the normal and abnormal of hand motion by time-domain analysis. Moreover, using the continuous wavelet transform (CWT) detection is clearer. It able to classify the abnormal in welding process which is caused by the changing of the movement cycle which is hard to detect in raw data. By using the quantitative method, the data of accelerometer and gyroscope have been transferred to create a scoring system for welder performance evaluation. In the result, the data 2 has 0.61 score, and data 3 has 0.92 score based on the calculation of hand motion signal. Hence, the data from hand motion sensor is able to use to classify the quality or grade of welder. However, this method enables us to classify the quality of welder while the welder uses the same type of cup-walking pattern because the difference of cup-walking types causes the difference in optimal time speed of welding which is the result of inaccurate of welder score performance.