Abstract: To ameliorate input and output constraint problems that exist during the control process of an analytic tower, a type of constrained predictive control algorithm is proposed on the basis of quadratic interpolation. By combining the step dynamic matrix control, the algorithm can change a matrix operation into a numerical value operation. The algorithm ensures incremental optimal solutions that satisfy constraint conditions in the constraint interval by introducing quadratic interpolation to obtain the predictive control value. The results of a field measurement model simulation show the effectiveness of this algorithm.