Design of Deformation Monitoring Networks Using Penalty Function-based Genetic Algorithm

This paper suggests a new method for designing an optimum deformation monitoring network using genetic algorithms. Optimization using genetic algorithms needs neither linearization nor differentiation of the object function or the constraint equations. As genetic algorithm uses simple mathematical computations it is easy to implement. In constraint problems one can use penalty functions to redefine the problems as unconstraint ones. This paper uses genetic algorithm to find out the optimum location for stations of a deformation monitoring network as well as the optimum observation weight (FOD and SOD problems). The network is designed in a way that the variancecovariance of the estimated parameters optimally approximates the criterion matrix. The paper reviews different components of the genetic algorithm and shows its efficiency using numerical example.