Contact-type vibration sensors (accelerometers) are primarily used to assess the condition of moving machines. Their susceptibility to electromagnetic interference, the need to attach them to each monitored element, and the impossibility of mounting them on rotating machine parts limit the capabilities and portability of vibration analysis. To improve the portability of defect detection and protect the sensor signal from pulsed electromagnetic interference present in production environments that mask signals indicating defects, a method for contactless diagnostics of moving machines has been proposed. The method is modeled using a dual-channel microwave interferometric displacement sensor designed to assess the technical condition of gears based on their vibration. Direct conversion of vibration into displacement is achieved through quadrature processing of the sensor's output signal. Diagnostics involves searching for signs of defects in the received signal in the form of specific frequencies inherent to a particular defect. These frequencies are characteristic of damaged gear teeth, shaft misalignment, and damage to associated bearings. Synchronous time averaging, signal envelope calculation, spectral analysis, and digital filtering were used to identify defect frequencies. To improve the signal-to-noise ratio, preprocessing of the signal was performed, highlighting the spectral excess, and the bandpass filter parameters required for re-filtering were determined. The onset and localization of degradation of a specific gear component at an early stage were determined by identifying characteristic signal harmonics associated with the corresponding defect. The obtained results are relevant for mobile, contactless monitoring of defects based on vibration of machines with moving and rotating parts. In an industrial environment, the proposed method is less expensive and more reliable than the laser method and is insensitive to sound interference, which is typical for the acoustic method.