Torque measurement is an essential method for assessing whether the output power of a power system meets operational requirements, serving as a critical source of information for equipment operational status and fault pre-warning. Surface Acoustic Waves (SAW) sensors, with their passive and wireless characteristics, are particularly well-suited for torque measurement of rotating shafts, overcoming the challenges in energy and signal transmission encountered with active torque sensors in rotational applications. To achieve high-precision SAW torque sensing, this study focuses on the selection of piezoelectric substrate materials, the design of high-Q resonators, and the development of a high-accuracy frequency demodulation algorithm for SAW sensor echo signals. The designed SAW torque sensor was tested, achieving a load nonlinearity of 0.183%, unload nonlinearity of 0.160%, load repeatability of 0.260%, unload repeatability of 0.103%, and hysteresis of 0.299%, thereby demonstrating high-precision torque sensing capabilities.