An improved quasi-sequential approach to large-scale dynamic process optimization

We present an improved quasi-sequential approach to large-scale dynamic process optimization. This new approach incorporates the quasi-sequential approach with the interior point method to handle inequality constraints. In this way the eventual optimization problem to be solved becomes a NLP without constraints. Mathematical derivations and computation schemes are developed. We first take a two-dimensional constrained optimization problem as an example; the result is compared with the simultaneous and quasi-sequential approach in terms of path solution with a graphical interpretation. A highly nonlinear reactor optimal control problem is also taken to demonstrate the effectiveness of this approach.