Removing drift from noisy accelerometer data

72 次查看（过去 30 天）

Jack Upton 2020-4-13

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/517621-removing-drift-from-noisy-accelerometer-data

评论： Emily Keys 2023-4-8

Hi all,

I am using the sensors within my phone to generate a CSV file of accelerations in 3-axis (x,y,z). I have now imported the data to matlab using the CSVread funtion and have began processing the data.

I have applied a filter to reduce some of the noise from the signal however upon integration the signal still drifts. The code I am using is shown below, for simplicitys sake I am only showing data from one axis. Any help is appreciated

clear; close; clc;
D=csvread('test20m3.csv');
t=D(:,1);                                                                           %Define time
XAccRaw=D(:,5);                                                                     %Define X acceleration
XAcc=XAccRaw*9.81;                                                                  %Convert to m/s^2
d=designfilt('lowpassfir','filterorder',10,'CutOffFrequency',10,'SampleRate',100);  %Lowpass FIR filter
AX=filtfilt(d,XAcc);                                                                %Apply filter to data
VX=cumtrapz(t,AX);                                                                  %Integrate acceleration to get velocity
SX=cumtrapz(t,VX);                                                                  %Integrate velocity to get displacement
figure(1);
plot(t,SX);
xlabel(Time (s));
ylabel(Displacement (m))

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

回答（2 个）

Image Analyst 2020-4-14

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/517621-removing-drift-from-noisy-accelerometer-data#answer_426030

在 MATLAB Online 中打开

I think this works great. It's well commented, practically self documenting but if you have any questions, ask.

% Initialization steps.
clc;    % Clear the command window.
fprintf('Beginning to run %s.m ...\n', mfilename);
close all;  % Close all figures (except those of imtool.)
clear;  % Erase all existing variables. Or clearvars if you want.
workspace;  % Make sure the workspace panel is showing.
format long g;
format compact;
% Load data and plot it.
hFig1 = figure;
sa = load('acceleration.mat')
accel = sa.acceleration1;
st = load('time.mat')
t = st.time1;
plot(t, accel, 'b.-', 'MarkerSize', 9);
grid on;
hold on;
fontSize = 20;
xlabel('Time', 'FontSize', fontSize);
ylabel('Acceleration', 'FontSize', fontSize);
title('Original Signal', 'FontSize', fontSize);
hFig1.WindowState = 'maximized'; % Maximize the figure window.
% Draw a line at y=0
yline(0, 'LineWidth', 2);
% A moving trend is influenced by the huge outliers, so get rid of those first.
% Find outliers
outlierIndexes = isoutlier(accel);
plot(t(outlierIndexes), accel(outlierIndexes), 'ro', 'MarkerSize', 15);
% Extract the good data.
tGood = t(~outlierIndexes);
accelGood = accel(~outlierIndexes);
% plot(t(~outlierIndexes), accel(~outlierIndexes), 'mo', 'MarkerSize', 10); % Plot circles around the good data.
% Do a Savitzky-Golay filter (moving quadratic).
windowWidth = 51; % Smaller for tighter following of original data, bigger for smoother curve.
smoothedy = sgolayfilt(accelGood, 2, windowWidth);
hold on;
plot(tGood, smoothedy, 'r-', 'LineWidth', 2);
legend('Original Signal', 'X axis', 'Outliers', 'Smoothed Signal');
% Now it looks pretty reasonable since we didn't include the outliers.
% But smoothedy has fewer points so if we're to subtract it from the original
% we have to fill in the missing points.
smoothedy = interp1(tGood, smoothedy, t);
% Now subtract the smoothed signal to get the variation
signal = accel - smoothedy;
% Plot it.
hFig2 = figure;
plot(t, signal, 'b.-', 'MarkerSize', 9);
grid on;
hold on;
title('Corrected Signal', 'FontSize', fontSize);
xlabel('Time', 'FontSize', fontSize);
ylabel('Acceleration', 'FontSize', fontSize);
hFig2.Units = 'normalized';
hFig2.Position = [.2, .2, .5, .5]; % Size the figure window.
% Draw a line at y=0
yline(0, 'LineWidth', 2);

10 个评论
显示 8更早的评论隐藏 8更早的评论

Image Analyst 2023-4-7

Same reply (not sure why you ignored it). Start a new discussion thread and attach your data and code, especially the code to turn acceleration into displacement.

Emily Keys 2023-4-8

Hi , I have done so except the data that I used is too large to upload to even in zip file format. Apologies again. https://uk.mathworks.com/matlabcentral/answers/1943334-fixing-post-load-for-acceleration-signal-with-savitzky-golay-filter

请先登录，再进行评论。

Ameer Hamza 2020-4-13

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/517621-removing-drift-from-noisy-accelerometer-data#answer_425839

If by drift, you mean the signal continually move away from the actual value then you can try to use detrend() function: https://www.mathworks.com/help/matlab/ref/detrend.html

12 个评论
显示 10更早的评论隐藏 10更早的评论

Jack Upton 2020-4-18

I thought that would be the case as thats what I have read elsewhere,Ive tried to implement it previously using GPS as well as my accelerometer results but I have had no luck

Ameer Hamza 2020-4-18

Yes, it can be a bit complicated because you need to estimate the model of your system. You can try to follow some example on Kalman filter in MATLAB: https://www.mathworks.com/discovery/kalman-filter.html

https://www.mathworks.com/help/control/examples/state-estimation-using-time-varying-kalman-filter.html

请先登录，再进行评论。

请先登录，再回答此问题。