The new method presented in this paper shows an effective way of solving an estimation problem, the estimated values are nearer to their theoretical ones than in an adjustment with the method of least squares. Considering the semi parametric adjustment models, firstly, the estimators of the parameters and the nonparametric are derived by using a kernel function and the least square method. Secondly, we discuss the approach of bandwidth choice in the semi parametric adjustment models. Finally, a simulated adjustment problem is constructed to explain the method.