Genetic Algorithms Research Based On Nonlinear Least Squares Estimation

Genetic algorithms(GA), which are stochastic search methods that mimic the metaphor of natural biological evolution, are introduced to do with the problem of Nonlinear Model Parameters Estimation(NMPE) in surveying.At the beginning of this thesis, the necessity of research on nonlinear science and the aim of research on NMPE are stated; Some advantages and disadvantages about present methods are pointed out when the analysis to these methods are made; The possibility of applying GA to NMPE is justified under the consideration of the advantages of GA and the features of NMPE.Then, the primary principles', methods and algorithms of GA are given in detail; And according to the object function of NMPE, the corresponding fitness function of GA is properly designed; With the practical application of NMPE in surveying, binary encoding and floating encoding GA Based On Least Square Estimation(GA-LSE) are constructed separately, and GA-LSE softwares have also been programed and compiled.At last, with the applications of GA-LSE to NMPE in surveying, such as Distances Observed_only Network,Traverse Network, Pseudo Random Noise Positioning and etc, some outcomes meeting the precise commands are obtained. Many reasonable conclusions are drawn and some benifical advice is given in the end.