Open-loop system identification of an AC motor-driven arterial flow generation platform: vessel-specific models and a parsimonious common ARX model
Accurate reproduction of arterial flow waveforms is important for in vitro cardiovascular experiments and controlled biomedical flow platforms. This study characterizes the open-loop dynamics of an AC motor-inverter-centrifugal-pump platform using five experimentally recorded input-output datasets representing femoral, carotid, iliac, coronary, and aortic waveform conditions. The inverter command was the input, the magnetic flow-meter voltage was the output, and all original recordings were sampled at 100 Hz. Vessel-specific FIR, ARX, and ridge-regularized ARX models were screened separately. A common model was selected using nested chronological validation and assessed on untouched test segments using one-step prediction, recursive free-run simulation, and diagnostic analyses. Vessel-specific one-step validation fits ranged from 80.33% to 92.80%. The common ARX(5,4,80) model achieved mean one-step and free-run fits of 95.82% and 80.04%, respectively. All poles were inside the unit circle, but significant residual autocorrelation showed that the measured response was not fully explained. The common model provides a compact nominal representation of the tested platform. Its comparison with vessel-specific models is descriptive because it uses more training data, and cycle consistency within a recording does not establish between-session experimental reproducibility.
Authors
- Hüseyin Yıldız (ORCID: https://orcid.org/0000-0002-0575-3904)
- Herman Sedef (ORCID: https://orcid.org/0000-0001-7852-5825)
- Serdar Yılmaz (ORCID: https://orcid.org/0000-0002-4316-3737)
- Canan Akay (ORCID: https://orcid.org/0000-0003-2781-8710)
- Feyzi Kaysi (ORCID: https://orcid.org/0000-0001-6681-4574)
Institutions
- Yıldız Technical University (TR)
- Istanbul University-Cerrahpaşa (TR)
Publication Details
- Journal
- Medical Engineering & Physics
- Published
- 2026-10-29
- DOI
- https://doi.org/10.1088/1873-4030/aebb2f
- Primary Topic
- Control Systems and Identification
- Type
- article
- Field-Weighted Citation Impact
- 0.00