In order to explore the transmission characteristics of acoustic emission (AE) signals in industrial robot body, the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) method is applied to decompose the AE waveform, and the characteristics of AE signals are analyzed by time-frequency domain analysis. The simulative AE signal generated by pencil-lead breaking was initially acquired using the DS5 acoustic emission instrument with a 3 MHz sampling frequency. Furthermore, the AE signal was analyzed by CEEMDAN to obtain various layers of the Intrinsic Mode Functions (IMF) component, followed by calculation of weight, correlation coefficient, and variance contribution ratio of the original signal. The reconstruction was then utilized to reconstruct the aforementioned AE signal waveform, and the FFT analysis was introduced to analyze the frequency domain characteristics of the reconstructed signal. The findings indicate that the AE signal generated by interior and exterior surface of the body ranges from 144 kHz to 170 kHz and 122 kHz to 155 kHz, respectively. Those results suggest that the surface coating of the industrial robot has an un-negligible impact on the transmission of the AE signal.