Simple linear signal prediction

Question

0 个投票

Hello, I'm trying to build simple linear prediction model. I have no problem to find coefficients(lpc function), but combining them with signal is a bit of a problem - is there a way to calculate them without loops (for/while)? Formula : X(m)= a_1*x(m-1)+a_2*x(m-2)+....+a_k*x(m-k) X - predicted value x - signal value a - prediction coefficient

I've seen that, there is a solution with filters (at leats filter function), but there is no normal explanation so far.

I'm new to this site, and have little experience with matlab, therefore I'm sorry if this question is little bit simple.

In addition, I wonder how close can I predict next value of signal(by using any type of prediction - not only linear)?

Thank you!!

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Greg Heath 2012-11-24

Calculate the crosscorrelation function xcorr, crosscor, or nncorr to see how many lags it takes to drop from 1 to exp(-1). This is the conventional decorrelation time. However, you can use any other reasonable value (e.g., 0.2).

请先登录，再进行评论。

请先登录，再回答此问题。

Follow Question