Moving Average Filter: A doubt

Question

Filippo Grazioli 2016-12-28

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/318423-moving-average-filter-a-doubt

回答： Julian 2017-1-5

Hello,

I need to develop a moving average filter that only considers the time window PRIOR TO a certain instant. In other words, considering the value of my output signal at a certain time t0, it must be the result of the prior time window. Therefore, the averaging time window must not be centred in t0, but end in t0. The application will be implemented real-time and there is not yet a time after t0.

filtCoeff = ones(1, T)/T;  
output = filter(filtCoeff, 1, input);

Do you think these two lines correctly carry out what I am trying to do? If not, how to do it?

Thanks a lot. Filippo

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

dpb 2016-12-28

Well, the implemented as written is simply

output=mean(input);

if input is the last T values of the data feed. If, OTOH, input continues to grow over time, filter will output a vector of results of length(input) each and every time it's called--probably not what you're looking for...

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Julian 2017-1-5

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/318423-moving-average-filter-a-doubt#answer_249498

Yes, the 2 lines you wrote do achieve what you want. The filter function is causal and inclusively ends at t0. The much newer function movmean (since R2016a) centres on t0 by default (but allow you to specify the window extent forward and backward).