The author describes a procedure for reducing an optimal control problem for the improvement of a controllable process described by a nonlinear discrete system to an analogous problem for a controllable system close to a linear system in its properties, which has the same nonlinear cost functional. Using discrete versions of known methods for an approximate solution to linear optimal control problems, it is possible to employ the result described in the paper to construct various algorithms for improving controllable processes described by nonlinear discrete systems. This method is suitable for constructing improvement algorithms for discrete systems of a general kind. An example is given to illustrate the basic result.