A non uniform control vector parameterization approach for optimal control problems

Key words:  optimal control  chemical process  non uniform control vector parameterization  sequential quadratic programming (SQP)  fixed final time

Abstract:

The traditional uniform control vector parameterization method can not obtain accurate optimal control policy when solving optimal control problems with fixed final time. For this reason, a non uniform control vector parameterization approach was proposed. Firstly, the given time interval was divided into several time stages of varying lengths which were defined as new optimization parameters. Secondly, the original optimal control problem was transformed into a uniform parameterization problem by introducing a time scaling factor. Finally, the adjoint equations were solved to obtain the gradients by constructing the Hamilton functions of objective and constraint functions. The numerical solution was obtained by sequential quadratic programming method. Simulation results on two optimal control problems of chemical process show the effectiveness of the proposed method.