Model predictive control for nonlinear unsymmetrical system which exhibits gains and/or dynamics depending on the input or output direction has been studied.Theoretical results for these systems especially for its control aspects are very limited compared with their common occurrence in industry.Due to the differences between the positive/negative model predictive control structure for asymmetric system and the traditional predictive control structure

there may be model mismatch which would lead to bad control effect.In order to handle this problem

input feedback control algorithm is used to revise the positive/negative model predictive controller.The actual control action that applied to the object is joined into the calculation of the next step control law of the positive/negative controllers through the feedback form in order to eliminate the effect.Then the controllability of the positive/negative model predictive control system is analyzed in the unconstrained condition.Ultimately the control performance between the revised positive/negative model predictive control and the traditional positive/negative model predictive control was compared through the simulation of pH control.