This paper discusses a high precision rational spectral collocation method for singularly perturbed problems.Using the rational spectral collocation method with sinh transformation, the Chebyshev nodes are encrypted at the boundary layer, and the higher precision can be achieved with fewer nodes. Then, in order to obtain the width of boundary layer in the sinh transformation, it designs an unconstrained nonlinear optimization problem with the least error as the objective function, and gives the differential...