Micro-electro-mechanical system (MEMS) accelerometers have great potentials for applications in aerospace, autonomous driving and consumer electronics. However, their working principles are mostly based on capacitive and resistance types, which cannot be easily and effectively used for wireless and passive sensing, while surface acoustic wave (SAW) technology is the key solution for this problem. Due to complex acoustic-electric-mechanical coupling during SAW accelerators’ operations, currently, there needs an accurate, reliable, and efficient design and simulation platform to improve their structure and performance. In this work, we proposed an accurate, reliable, and efficient modeling platform to optimize designs of SAW accelerometers, using a double-ended cantilever beam structure as an example. This model integrated the elastic acoustic effect and the coupled wave equations under both the mechanical and electrical loading using the finite element analysis, and effectively obtained acceleration-frequency responses of the SAW accelerators. We have systematically simulated effects of thickness of piezoelectric film, wavelength, and structural parameters of cantilever beams, and the simulation results are well consistent with the theoretical ones. Finally, using the developed model, we designed a high-G SAW accelerometer (up to 20000 g) with a high sensitivity (−41.8 Hz g−1) and excellent linearity (0.9999), and another one with a high sensitivity (3.02 KHz g−1) and a good linearity (0.9999) over a 100 g acceleration range.