Global optimization techniques for modeling, control, and, estimation of large-scale systems

Many engineering applications can be modeled as mathematical programs. Efficient methods are required to solve these optimization problems. Optimization has a wide range of applications in Chemical Engineering, including production scheduling, process synthesis, design, and control.; Very few real systems are linear. Model nonlinearities often give rise to nonconvexities which may lead to multiple local solutions. Conventional nonlinear programming techniques have only been very successful in determining local solutions which can be suboptimal. Determination of global optima for nonlinear nonconvex problems has been an important topic of optimization research during the last several decades.; Complex nonlinear problems can prove to be computationally demanding and can require a large amount of time to be solved. Deterministic global optimization algorithms like Branch-and-Bound and outer approximation algorithms typically depend on generation of relaxations to nonlinear problems. Convergence speeds of global algorithms are often slow because of poor relaxations using existing methods. A MILP based piecewise linear relaxation technique is developed in this work for generating tighter relaxations. Generated MILP relaxations are tighter compared to those developed by traditional methods and hence are also used in Optimization-based variable range reduction technique. A rigorous decomposition based MINLP algorithm is also developed for solving factorable MINLP problems. Additionally, a Matlab based toolbox called GLOBO and several parallel global optimization techniques have been developed for solving different types of optimization problems.; Developed global optimization algorithms have been successfully applied for model identification of regulated biological systems. Additionally, dynamic metabolic pathway models have been used in MINLP formulations for species yield optimization.