| Title |
Characteristics of Optimization Methods for the Design of Groundwater Bioremediation |
| Abstract |
The characteristics of optimization methods are inspected by applying those to groundwater in situ bioremediation. The methods used in the comparison are Evolutionary Algorithms (Binary-Coded Genetic Algorithm (BIGA), Real-Coded Genetic Algorithm (RGA), Derandomized Evolution Strategy (DES)), Direct Search Methods (Nelder-Mead Simplex (NSLX), Modified Simplex (MSLX), Parallel Directive Search (PDS), and Derivative-Based Methods (Implicit Filtering for Constrained Optimization (IFFCO), Successive Approximation Linear Quadratic Regulator (SALQR)). For time-invariant control (9 decision variables), MSLX, IFFCO, DES and SALQR show preciseness in solutions. In case of time-varying control (54 and 72 decision variables), all the directive search methods and IFFCO converge to local optima, and produce inaccurate solutions. SALQR, an optimal control algorithm, shows high efficiency, but also converges to local optima to yield uns table solutions for some cases of time-varying control problems. RGA is more efficient and accurate than BIGA, and more robust than the deterministic search methods. Among the methods, DES is a promising method to be easily applied to other system problems because of its efficiency and accuracy. |