Why this validation matters
Compare reference-speed tracking and torque ripple while testing sensitivity to observer convergence, machine parameters and operating-speed transitions.
Several different numerical or control assumptions can produce visually plausible plots. The validation must establish that the observed trend comes from the intended physical mechanism rather than mesh, solver, controller or boundary-condition artifacts.
First-pass physical and numerical checks
- Test near-zero speed separately because back-EMF-based observability can become weak.
- Recheck pole-pair scaling when converting electrical and mechanical angles.
- Compare identical motor parameters, inverter conditions and load torque across control variants.
- Measurement cross-check: Reference versus measured rotor speed and estimated speed.
- Measurement cross-check: Actual versus estimated rotor electrical angle and estimation error.
Challenge the model under changed inputs
- Case 1: sensored field-oriented control. Establish a reference case, change one physical assumption and compare the same metrics across runs.
- Case 2: sensorless FOC without sliding-mode observer tuning. Establish a reference case, change one physical assumption and compare the same metrics across runs.
- Case 3: SMO-based FOC under parameter mismatch. Establish a reference case, change one physical assumption and compare the same metrics across runs.
Quantities to retain for audit
- Archive reference versus measured rotor speed and estimated speed with units, time window, data export format and configuration identifier.
- Archive actual versus estimated rotor electrical angle and estimation error with units, time window, data export format and configuration identifier.
- Archive d–q currents, phase currents and electromagnetic torque ripple with units, time window, data export format and configuration identifier.
- Archive observer convergence and load-step recovery at low and medium speeds with units, time window, data export format and configuration identifier.
What would invalidate a comparison
Good steady-state speed agreement can hide angle-estimation transient error. Quantify tracking and observer errors separately, then interpret torque oscillation against the current-loop bandwidth.
Do not use mismatched operating conditions or different normalization conventions to claim improved performance. Report the numerical tolerance or controller sampling time where it materially affects the result.
Reporting checklist
Record model fidelity and exclusions; name the data source for each parameter; label whether a result is simulated, measured or calculated from another quantity.
Validate repeatability before expanding parameter sweeps, and preserve a baseline output so later studies can be reproduced.
Related project watch pages
Watch the relevant simulation output first, then use this article as a technical review or experimental planning guide.
Engineering research support
For the underlying MATLAB/Simulink, COMSOL, ANSYS HFSS or power electronics model, verification of solver settings, and documented outputs, contact the MATLAB Projects Code research team. Articles describe engineering validation approaches and are not a substitute for measured validation data.
Contents are for representative purposes; actual content may vary.