The researchers analyzed ultrasonic signals to detect welding defects by studying their chaotic features. They used fractal dimension and Lyapunov exponent to quantify the chaotic behavior of the signals. By examining the trajectory changes in the signals, they found that different defects showed varying levels of chaoticity. The study identified unique features of defects in the weld zone based on these differences. The analysis of fractal dimension and Lyapunov exponent from the signals revealed chaotic behavior in cases of porosity and incomplete penetration. This method improved the recognition of welding defects using ultrasonic signals.