【求助】请教一个拟合的问题 (转载)

This is just like identifying a static nonlinear system. Just check any system identification book about the Least Mean Square algorithm, this is the most simple algorithm to identify the a b c. Since the output is linear in a b c, you can also use Recursive Least Square algorithm, Kalman Filter. Or, more general one, Recursive Prediction Error method. Coding is very easy, use the original equation to get a reference signal, then assumed that a b c is not known, give them a initial value, then calculate the output, compare the output with reference signal to obtain the error signal, then use LMS algorithm, (a b c)=(a b c)+u*error*(x x^2 x^3). u is step size, less than 1. After several iteration, you can get a b c.