This paper proposes an embedded system, as a major part of intelligent module, to improve the reliability and precision of the hydrostatic bearing, which is known as a key subsystem of precision machine tools. The embedded system integrated with sensory devices has two main functions: virtual metrology and fault diagnosis. By employing virtual metrology, the embedded system can accurately estimate the change of oil-film thickness as well as bearing stiffness based on on-line measured pressure, temperature and flow rate with the sensory devices, which will provide to the controller of machine tool to either compensate the dimensional inaccuracy or initiate preventive maintenance. To fulfill the function of embedded system the equations governing the relationship among the oil-film thickness, pocket pressure and flow rate are derived and employed to train the predictive model of virtual metrology by a large amount of experimental data with Linear Regression method. Comparing the error predictive model with theoretical derivation, the error of the predictive model was reduced less than 5%.