Modified genetic algorithm for nonlinear data reconciliation

Abstract

Nonlinear data reconciliation problem are inherently difficult to solve with conventional optimization methods because of the existence of a multimodal function with differentiated solutions. In this paper, the genetic algorithm (GA) of Wasanapradit [Wasanapradit, T. (2000). Solving nonlinear mixed integer programming using genetic algorithm. Master Thesis, King Mongkut University of Technology Thonburi, Bangkok, Thailand. Available: fengtcs@ku.ac.th] based on modified cross-generational probabilistic survival selection (CPSS) is explored for solving the steady state nonlinear data reconciliation (DR) problem. The DR problem is defined by a redescending estimator as the objective function, which is both a non-convex and discontinuous function. In the GA method, first the appropriate GA parameters are found and then the algorithm must be validated with the problem. The results show that the GA solves the redescending function without the complex calculations required by conventional optimization methods, but the calculation time is longer.