find the start point of abrupt change in signal matlab

8 次查看（过去 30 天）

显示更早的评论

Demet 2022-8-19

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1781975-find-the-start-point-of-abrupt-change-in-signal-matlab

评论： Demet 2022-8-22

采纳的回答： Chunru

Hello,

I have a signal as y = [x y sh oa ahdv ...] and looks like as below:

I need to find the the point where signal increased abruptly and tried diff(), findchangepts(), ischange(), etc but any of them did not help me. I have 100 sample looks like this data so I am looking for a function that can do this job automatically.

I would be very grateful if anyone could help me.

Thanks.

Demet

2 个评论
显示无隐藏无

Chunru 2022-8-19

Attach your data.

Demet 2022-8-19

triggerInput.mat

Hello,

Here is a data sample attached.

Thanks.

请先登录，再进行评论。

请先登录，再回答此问题。

采纳的回答

Chunru 2022-8-20

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1781975-find-the-start-point-of-abrupt-change-in-signal-matlab#answer_1029675

编辑：Chunru 2022-8-22

在 MATLAB Online 中打开

triggerInput.mat

load triggerInput.mat

plot(trigger)

trigger = trigger(1:2e4); % select the 1st part

%xlim([1 1e4])

%xlim([1500 2500])

% low pass the signal

b = fir1(128, 0.001);

y = filtfilt(b, 1, trigger);

ipt = findchangepts(y,'MaxNumChanges',2)

yd = diff(y);

idx = find(yd>3e-3, 1, 'first')

figure

hold on

plot(y)

plot(trigger, 'r:');

xline(ipt, 'g');

grid on; box on

xlim([1 2e4])

fprintf('The change point at %d\n', ipt(2))

Update based on your new explanation:

b = fir1(128, 0.001, 'high');

y = filtfilt(b, 1, trigger);

ipt = findchangepts(y,'Statistic', 'rms', 'MaxNumChanges',2)

ipt = 1×2

1733 7865

figure

hold on

plot(y)

plot(trigger, 'r:');

xline(ipt, 'g');

grid on; box on

xlim([1 2e4])

3 个评论
显示 1更早的评论隐藏 1更早的评论

Chunru 2022-8-22

See the update

Demet 2022-8-22

Hello,

I have checked the update and it works almost well. I have checked for each data and I decide to continue with this program as the maximum error is 30 for the sampling frequency is 512 (30/512=0.57 s).

Thank you!

请先登录，再进行评论。