A novel Inverse Feedback Control technique is proposed for the design of optimal open loop control system in semi-batch free radical copolymerization process in which on-line polymer property sensors are absent. The open loop polymer quality control problem is converted to the closed loop controller design problem in which "fictitious" on-line polymer property sensors are assumed to be available. The time varying trajectories of reactor manipulated variables in such "fictitious" feedback control system are found to be equivalent to the optimal trajectorios of control variables or set point programs required in the original open loop control system. The proposed technique can also be easily applied to the design of multivariable open loop systems for nonisothermal semi-batch copolymerization processes. The validity and the applicability of the proposed technique based on "knowledge learning" principle are illustrated through the numerical examples of semi-batch copolymerization process control.