The triplecoil inductive debris sensor is the main sensor for the monitoring of metal debris in lubricating oil. But the bubble in the lubricating oil will make the sensor produce interference signal. Aiming at the problem that the traditional method mistakenly detects the bubble interference signal as the metal debris signal, a signal detection method based on continuous wavelet transform and curve fitting is proposed. Taking advantage of the high similarity between the metal debris signal and Gaussian1 wavelet, the continuous wavelet transform (CWT) threshold method is used in wavelet domain to filter interference signal and extract signal waveform, the Gauss Newton method is used to fit the extracted signal waveform, and the Rsquared is used as the standard to judge the fitting results, detect the metal debris signal and remove the interference signal. The actual test on the sensor with an inner diameter of 20 mm shows that it can accurately detect the spherical ferromagnetic debris with a diameter of more than 150 μm and the spherical copper debris with a diameter of more than 250 μm, with an accuracy of 99% and 97% respectively, and can effectively remove the bubble interference signal.