Title :
Combined algorithm for time-varying system based on improved genetic algorithm and EWRLS algorithm
Author :
Xue, Yun-can ; Yang, Qi-Wen ; Qian, Ji-Xin
Author_Institution :
Coll. of Comput. & Inf. Eng., Hohai Univ., Changzhou, China
Abstract :
A combined algorithm, based on the modified genetic algorithm and EWRLS algorithm, is presented in this paper. A modified genetic algorithm with dyadic mutation operator is also presented to enhance the response speed of genetic algorithm. The selection criteria of the switching threshold between the GA and EWRLS algorithm is also given by using robust minmax estimation method. This combined algorithm solves the tracking problem of time-varying system with fast parameter changes, which is very difficult to the RLS algorithm. It is not sensitive to the noise. Its good performance is verified by simulation studies.
Keywords :
genetic algorithms; least squares approximations; minimax techniques; recursive estimation; time-varying systems; tracking; EWRLS algorithm; GA; RLS algorithm; dyadic mutation operator; fast parameter changes; genetic algorithm; noise insensitivity; robust minmax estimation method; selection criteria; switching threshold; time-varying system; tracking problem; Educational institutions; Estimation error; Genetic algorithms; Genetic engineering; Genetic mutations; Machine learning; Machine learning algorithms; Parameter estimation; Resonance light scattering; Time varying systems;
Conference_Titel :
Machine Learning and Cybernetics, 2002. Proceedings. 2002 International Conference on
Print_ISBN :
0-7803-7508-4
DOI :
10.1109/ICMLC.2002.1167449
