Parameter Estimation of Exponentially Damped Sinusoids Based on GA-LS Method

Dong Zheng,Peng Cheng,Wang Yong
DOI: https://doi.org/10.3969/j.issn.1000-0755.2010.10.005
2010-01-01
Abstract:A parameter estimation method for exponentially damped sinusoids based on GA-LS is proposed here.According to the features of system model the parameters to be estimated can be divided system into two parts,the linear form part and the nonlinear form part.The genetic algorithm and least-squares method are combined to estimate the parameters of the exponentially damped sinusoids.After the observation and analysis of signal character in frequency domain,the parameter range of various nonlinear parts is preliminarily determined using genetic algorithm(GA) to search in parameter space,and then using the optimum of current generation to determine linear parameters by least squares(LS).Through reducing the dimensions of the search space,and considering the constraint relationship between linear part parameters,the efficiency and the precision of this kind of parameter estimation are improved.Lots of simulation show that this algorithm is superior to standard genetic algorithm in convergence rate and has a high precision.The application in processing shock signal can basically satisfy the practical requirements.
What problem does this paper attempt to address?