Off course it is possible. You should keep in mind that the y+ must be sufficiently small (less than 1, usually 0.2-0.3). Moreover, the CFL has to be limited to 0.2-0.3 as well

Adrian Lungu is right，It is very necessary to ensure the grid parameters near the wall such as y+. On this basis, I would like to know how to estimate the required number of grid nodes based on the flow characteristics such as Reynolds number. It can be compared to estimating the grid based on Reynolds number in LES simulation. In theory, the sub-grid scale required for simulation is closely related to the Reynolds number, so the grid distribution can be analyzed based on the flow characteristics.

But the RANS/LES hybrid model is different. It also has the problem of the position of the interface between RANS and LES.What I am interested in is to know it from the perspective of theoretical analysis how to estimate the approximate number of grids required that near wall RANS zone and the core turbulence LES zones based on the basic characteristics of the flow, such as the thickness of the boundary layer and the Reynolds number.

Yes Y+ controls the near wall mesh resolution i.e sufficiently smaller than 1 to capture the velocity gradients. Moreover, the mesh size in LES is limited by the filter size as large cells will not allow the required energy spectrum to be captured. Therefore, for your case, you need to estimate the size of small scale eddies which depends on Integral scale Re and choose the mesh size accordingly. You also need to ensure CFL < 1 for accurate and stable LES calculations.

Documents that analyze LES grids like this: "Grid-point requirements for large eddy simulation: Chapman’s estimates revisited" or "Grid Construction Strategies for Wall-Resolving Large Eddy Simulation and Estimates of the Resulting Number of Grid Points".