A time nodes optimization method is presented for the solution of optimal control problems (OCP) using control vector parameterization (CVP). The control variable in each time interval is approached by a function of time nodes parameters and other decision parameters. The infinite dimensional optimal control problems were approximated into nonlinear programming (NLP) problems, the gradient of which can be analytically obtained. The NLP problems were solved and resulted in non-uniform time grid. Two cases of optimal control problem were solved by the proposed methodology. One is OCP of an ODE system and the superior results are given comparing with the fixed time grid method. Another is optimal injection strategy determination of polymer flooding, which is a sort of OCP constrained to satisfy PDEs. The efficiency is illustrated for solving this kind of problem by the given method
[1]
J. Burns,et al.
A PDE Sensitivity Equation Method for Optimal Aerodynamic Design
,
1997
.
[2]
W. Ramirez,et al.
Optimal injection policies for enhanced oil recovery: Part 2--Surfactant flooding
,
1984
.
[3]
Mangesh D. Kapadi,et al.
Optimal control of fed-batch fermentation involving multiple feeds using Differential Evolution
,
2004
.
[4]
V. Vassiliadis,et al.
Restricted second order information for the solution of optimal control problems using control vector parameterization
,
2002
.
[5]
W. F. Ramirez,et al.
Optimal Injection Policies for Enhanced Oil Recovery
,
1984
.
[6]
L. Petzold,et al.
Software and algorithms for sensitivity analysis of large-scale differential algebraic systems
,
2000
.