Small sample analysis method is proposed to find out the relationship between measurement distance and position error. The grey absolute correlation degree is first used to analyze the correlations of position measurement error. A grey forecasting model is then used to forecast data and expand the small sample size to a relatively larger one. The multivariate multi-order polynomial fitting is made with data. Finally, the accuracy evaluation indices are used to assess the accuracy of curve fitting, which include maximum, minimum and average relative errors. Experimental results show that the difference between maximum and minimum relative errors is in an acceptable range and the accuracy of curve fitting is proved to be superior to 99% which satisfies the engineering requirements.