By an equivalent transformation using the pth power of the objective function and the constraint, a saddle point can be generated for a general class of nonconvex optimization problems. Zero duality gap is thus guaranteed when the primal-dual method is applied to the constructed equivalent form.