The ultrasonic echo signal of wear debris is influenced by many matters, thereby noises are existed and selection of decomposition level has great influence on the effect of noise denoising. Therefore, a kind of algorithm to select the optimal decomposition level based on particle swarm optimization (PSO) method was put forward. The obtained optimal decomposition level was substituted in the dualtree complex wavelet field, and a method was used which combined the asymptotic semisoft threshold function with adaptive threshold selection. Dualtree complex wavelet transform (DTCWT) threshold was utilized to denoise for the noisy ultrasonic echo signal. According to the parameters of signaltonoise ratio (SNR) and root mean square error (RMSE), the result of denoising was evaluated. Simulated and experimental results show that the SNR is the highest which is obtained by the decomposition level selected by PSO, the adaptive denoising method for ultrasonic echo signal of wear debris in oil has obvious effect on signal denoising, and it obviously improves the SNR and reduces the RMSE as well as restore the waveform features of the signal, which lays a good foundation for the further feature extraction and intelligent recognition.