What are options to filter signal during simulation ?

I have two signals .

Yellow line is expected and blue line is actual response .

I want to filer blue line output signal to match yellow line as possible as I can.

The frequency of yellow line oscilations is 0.886. The perid mesured P-P is 1.128Sec.

I tried with LPF but didnt see any difference. The filtering is required to be done during simulation not after simulation.

[Umar]

Hi @AK,

To filter the blue line output signal to match the yellow line, you can implement a low-pass filter (LPF) using MATLAB's built-in functions. Since you mentioned that the LPF did not yield the desired results, it may be beneficial to adjust the filter parameters or use a different filtering approach, such as a moving average filter.Here’s a simple example using a moving average filter:

% Simulation parameters

fs = 10; % Sampling frequency (Hz)

t = 0:1/fs:10; % Time vector (10 seconds)

f_yellow = 0.886; % Frequency of yellow line (Hz)

% Generate yellow line (expected signal)

yellow_line = sin(2 * pi * f_yellow * t);

% Generate blue line (actual signal with noise)

blue_line = yellow_line + 0.1 * randn(size(t)); % Adding noise

% Apply a moving average filter

windowSize = 5; % Size of the moving average window

filtered_blue_line = movmean(blue_line, windowSize);

Note: for more information on this function, please refer to

https://www.mathworks.com/help/matlab/ref/movmean.html

% Plotting the results

figure;

plot(t, yellow_line, 'y', 'LineWidth', 2); hold on;

plot(t, blue_line, 'b', 'LineWidth', 1);

plot(t, filtered_blue_line, 'r', 'LineWidth', 1);

legend('Yellow Line (Expected)', 'Blue Line (Actual)', 'Filtered Blue Line');

title('Signal Filtering Example');

xlabel('Time (s)');

ylabel('Amplitude');

grid on;

Please see attached plot.

So, in this example code provided, the moving average filter smooths the blue line, helping it to better align with the yellow line. Adjust the windowSize parameter to optimize the filtering effect based on your specific data characteristics. Let me know if you have any further questions.

[AK]

Tried to use LPF block with 10Hz freq. But no luck.

To elaborate more Blue line is simulation output of my model, Yellow line is expected output or target output. I dont need to generate both signals from m scripting. Both are outputs of my simulink model. I expect to do somthing like this.

[Mathieu NOE]

@AK

I have some serious doubts that you can find out a "real time" LPF that smooth enough your signal without introducing too much time delay

the more you lower the corner frequency and / or increase the LPF, the more you add group delay , so you may end up with a more or less better looking signal , but significantly time delayed . If this is not a problem , any sharp LPF could be a solution to your problem, but at some point too much filtering will also affect even the general shape of your signal

[AK]

I agree with you. Now I also think that its not possible to do real time filtering.Currently its not a issue because the signla fluctuations are steady and folliwng with target signal.

Thanks for the support.

[Umar]

Hi @AK,

As highlighted in @Mathieu NOE’s comments, real-time filtering poses challenges due to the potential introduction of delays. The balance between signal clarity and temporal fidelity is essential. If the application requires real-time processing, it may be necessary to compromise on the level of noise reduction to maintain signal responsiveness. One viable option at this point that I could think of is implementation of FIR (Finite Impulse Response) filter which can be designed to have linear phase characteristics, thus minimizing phase distortion and time delays while providing effective noise reduction. However, here is a refined version of my code that could be tested with varying parameters to find the optimal filter for your specific signals:

% Simulation parameters
fs = 10; % Sampling frequency (Hz)
t = 0:1/fs:10; % Time vector (10 seconds)
f_yellow = 0.886; % Frequency of yellow line (Hz)

% Generate yellow line (expected signal)
yellow_line = sin(2 * pi * f_yellow * t);

% Generate blue line (actual signal with noise)
blue_line = yellow_line + 0.1 * randn(size(t)); % Adding noise

% Apply a moving average filter
windowSize = 5; % Adjust this size based on signal characteristics
filtered_blue_line = movmean(blue_line, windowSize);

% Plotting the results
figure;
plot(t, yellow_line, 'y', 'LineWidth', 2); hold on;
plot(t, blue_line, 'b', 'LineWidth', 1);
plot(t, filtered_blue_line, 'r', 'LineWidth', 1);
legend('Yellow Line (Expected)', 'Blue Line (Actual)', 'Filtered Blue   Line');
title('Signal Filtering Example');
xlabel('Time (s)');
ylabel('Amplitude');
grid on;

I will also add additional comments by analyzing your signals in the frequency domain using the Fast Fourier Transform (FFT). This can provide insights into the frequency components of both signals and help you determine the optimal filtering approach. Now, if the characteristics of the noise change over time, adaptive filtering techniques may be beneficial because these filters adjust their parameters in real-time based on the input signal characteristics, providing more responsive filtering. So, after filtering, it’s crucial to analyze the output signal for time delays and distortions. Cross-correlation between the filtered blue line and the yellow line can help quantify the alignment and timing issues.

Hope this helps. Please let us know if you have any further questions.