Abstract
We present a genetic algorithm approach to fitting curves with multiple parameters to data points. Using data from three functions with two to ten coefficients and one to ten variables with linear, polynomial and transcendental functional forms, we varied the number of data points, the range searched over, the distance metric, the distribution for sampling, and the parameters of the genetic algorithm to examine the robustness of the approach. We developed a sequential evolution mechanism to overcome premature convergence to local minima when some terms dominate others in magnitude. Comparison with previously published work is favourable with regard to both goodness of fit and computational effort.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
