ABSTRACT

If the real sine function is y A ⋅0 0 0sin( ).ω φx⋅ + The sampling data are ( , ), ( , ) ( , ).x y x y xn ny1 1 2 2 By some useful parameter extraction way, it is able to find the parameters ( , , )* * *ω φ to minimize the difference between yi and yi

*, where y Ai i

* * * *sin( )⋅ ω φx⋅ + , i n1 2, , ,… . Furthermore, if the parameter extraction way is good enough to anti interference, the difference between y Ai 0 0i⋅0 sin( )ω φx⋅ + and yi

* can be minimized. Actually, almost all the actual sampling data are interference. So the anti interference needs are reasonable, even necessary. There are variety ways for measuring the difference, like min | |,*y yi i min | |

=1 and min ( ) .*y yi i

There are some shortcomings for the least square method in sine wave signals parameter extraction: large computing, low precision. The nonlinear optimal algorithms, like Quasi-Newton method and Levebberg-Marquart method, may easy to converge to the local minima. The genetic algorithm can overcome these shortcomings properly.