The water-lubricated thrust bearings of the marine Rim-Driven Thruster (RDT) are usually composed of polymer composites, which are prone to serious wear under harsh working conditions. Ultrasonic is an excellent non-destructive monitoring technology, but polymer materials are characterized by viscoelasticity, heterogeneity, and large acoustic attenuation, making it challenging to extract ultrasonic echo signals. Therefore, this paper proposes a wear monitoring method based on an improved ultrasonic reflection coefficient amplitude spectrum, introducing an ultrasonic transfer function to mathematically resolve the attenuation coefficients in the propagation model, and using a differential evolutionary algorithm to invert the thickness, sound velocity, density, and attenuation coefficients of polymer bearings simultaneously. The effects of different influencing factors on the ultrasonic signal and identification results are explored, including probe contact pressure, surface roughness and surface inclination; In order to remove the influence of water film overlap on the identification of polymer bearing thickness, a water film separation method based on time-frequency domain ultrasonic reconfiguration is proposed to construct the ultrasonic theoretical time domain waveform of the bearing. An ultrasonic wear measurement system is built to record the ultrasonic echoes of PEEK test blocks with different pressures to the probe, surface roughness, and tilt angles. The experiment is carried out on the water film calibration test rig, using a high-precision spiral micrometer to simulate the mixing of water films of different thicknesses with polymer test block and collecting ultrasonic signals. The results show that the method has a thickness identification error of approximately 0.5% for PEEK specimens with different probe contact pressure, roughness, and tilt angle, and the reconfiguration thickness identification error is less than 1% for the full range of water film overlap.