MATLAB
Computational Analysis

MathWorks Validated Proficiency

Production-grade workflows require certified tool mastery. These certifications validate the core competencies applied across the projects in this portfolio.

Vibration & Acoustic Pipelines

Production-grade signal analysis pipelines for vibration and acoustic data — from raw sensor output to actionable features for condition monitoring and system identification.

% Vibration signal processing pipeline
function features = extract_vibration_features(signal, fs)
    % Preprocessing
    signal = detrend(signal);
    signal = bandpass(signal, [10 5000], fs);

    % Time domain
    rms = sqrt(mean(signal.^2));
    crest = max(abs(signal)) / rms;
    kurt = kurtosis(signal);

    % Frequency domain
    [pxx, f] = pwelch(signal, hanning(1024), 512, 2048, fs);
    [peak_f, peak_p] = findpeaks(pxx, f, 'MinPeakProminence', 0.1*max(pxx));

    % Envelope analysis for bearing faults
    env = abs(hilbert(signal));
    [env_pxx, env_f] = pwelch(env, hanning(1024), 512, 2048, fs);

    features = struct('rms', rms, 'crest', crest, 'kurtosis', kurt, ...
                      'peak_freq', peak_f, 'peak_power', peak_p, ...
                      'envelope_spectrum', struct('f', env_f, 'pxx', env_pxx));
end

Grey-Box & Black-Box Modeling

Mechanical systems modeled from input-output data for surrogate validation, reduced-order modeling, and controller synthesis. Three model structures cover the spectrum from physically interpretable to purely data-driven.

% N4SID Subspace Identification - Operational Modal Analysis
function [A, C, K, x0] = identify_modal_params(u, y, fs, nx)
    % u: input (force), y: output (acceleration), fs: sample rate, nx: state order

    % Create iddata object
    data = iddata(y, u, 1/fs);
    data = detrend(data);

    % N4SID with focus on simulation
    opt = n4sidOptions('Focus', 'simulation', ...
                       'Display', 'on', ...
                       'EstimateCovariance', true);
    sys = n4sid(data, nx, opt);

    % Extract modal parameters
    [fn, damp, mode_shape] = modal(sys, fs);
    MAC = modalsys(mode_shape);

    % Stabilization diagram for order selection
    [stab, ~] = stabplot(sys);
    disp('Stabilization diagram generated for order validation.');
end

Hardware-in-the-Loop Integration

Real-time capable models deployed to Speedgoat hardware for hardware-in-the-loop (HIL) testing of embedded controllers. The plant model runs at microsecond timesteps while the controller executes on the target hardware.

From Signal to Decision

• Feature Extraction: Automated bearing fault diagnosis from raw vibration — 94% classification accuracy on test rig data.
• Model Reduction: 500 DOF FEA model → 12-state subspace model — 40× speedup for HIL with <3% frequency error.
• Controller Validation: HIL testing caught 3 timing bugs in embedded controller before hardware integration.