How to Do Regression Testing in Simulink: Step-by-Step Guide

To do regression testing in Simulink, use the Simulink Test toolbox to create test cases and harnesses that run your model with known inputs and compare outputs to baseline results. This ensures your model behaves consistently after changes by automatically detecting differences.

Syntax

Regression testing in Simulink mainly involves these components:

• Test Harness: An isolated environment to run tests on a model or subsystem.
• Baseline: A saved set of expected outputs from a previous test run.
• Test Manager: Tool to create, run, and manage test cases.
• verifyEqual or compare functions: Used to compare current outputs with baselines.

Typical workflow syntax:

1. Create test harness: sltest.harness.create(model, subsystem)
2. Run test and capture output: simOut = sim(model)
3. Compare output to baseline: verifyEqual(testCase, simOut.yout, baselineOutput)

matlab

import sltest
model = 'myModel';
subsystem = 'myModel/mySubsystem';

# Create test harness
sltest.harness.create(model, subsystem);

# Run simulation
simOut = sim(model);

# Compare output to baseline (example)
# verifyEqual(testCase, simOut.yout, baselineOutput);

Example

This example shows how to create a test harness, run a simulation, save baseline data, and then run regression testing by comparing new results to the baseline.

matlab

model = 'simpleModel';
subsystem = 'simpleModel/Subsystem1';

% Create test harness
sltest.harness.create(model, subsystem);

% Run simulation and save baseline output
simOut = sim(model);
baselineOutput = simOut.get('yout');
save('baselineData.mat', 'baselineOutput');

% Later: Load baseline and run regression test
load('baselineData.mat', 'baselineOutput');
newSimOut = sim(model);
newOutput = newSimOut.get('yout');

% Compare outputs
if isequal(newOutput, baselineOutput)
    disp('Regression test passed: Outputs match baseline.');
else
    disp('Regression test failed: Outputs differ from baseline.');
end

Output

Regression test passed: Outputs match baseline.

Common Pitfalls

• Not isolating tests: Running tests without a test harness can cause interference from other model parts.
• Ignoring baseline updates: Always update baselines intentionally; accidental changes can mask real issues.
• Comparing raw floating-point data directly: Use tolerance-based comparison to avoid false failures due to minor numerical differences.
• Not automating tests: Manual testing is error-prone; use Simulink Test automation features.

matlab

%% Wrong: Direct equality check without tolerance
if isequal(newOutput, baselineOutput)
    disp('Test passed');
else
    disp('Test failed');
end

%% Right: Use tolerance for numeric comparison
tolerance = 1e-6;
diff = abs(newOutput - baselineOutput);
if all(diff < tolerance)
    disp('Test passed with tolerance');
else
    disp('Test failed');
end

Quick Reference

Key steps for regression testing in Simulink:

• Create a test harness for the model or subsystem.
• Run simulation and save outputs as baseline.
• After changes, rerun tests and compare outputs to baseline.
• Use tolerance-based comparison for numeric data.
• Automate tests with Simulink Test for repeatability.

Key Takeaways

Use test harnesses to isolate and run regression tests in Simulink.

Save baseline outputs to compare against future simulation results.

Use tolerance-based comparisons to handle floating-point differences.

Automate regression tests with Simulink Test for consistent validation.

Update baselines only when model changes are verified and intentional.