Speaker
Description
The accurate interpretation of sensor data in prosthetic systems is challenged by nonlinear signal transformations, environmental coupling, and temporal drift within the prosthetic socket environment. Sensor outputs derived from embedded thermistors are influenced by
dynamic interactions between physiological heat generation, mechanical loading, and ambient conditions, resulting in non-stationary measurement behaviour.
This work presents a dynamic calibration framework for nonlinear sensor systems based on inverse modelling and adaptive parameter estimation. A thermodynamically-inspired logarithmic model is employed to reconstruct temperature from sensor voltage ratios, where calibration is governed by a latent parameter estimated through constrained nonlinear optimization. Temporal continuity is enforced via regularization, while Bayesian
updating enables recursive refinement of calibration parameters over time.
To address non-stationarity, calibration is performed across multiple temporal scales,including early-morning equilibrium conditions and sliding time windows, capturing both long-term drift and short-term variability. Additionally, accelerometer-derived activity signals are used to identify low-perturbation states, ensuring calibration occurs under quasi-static physical conditions.
Experimental results demonstrate that the proposed framework improves robustness against sensor drift and enhances stability in reconstructed temperature signals compared to static calibration approaches. The method provides a physically grounded approach for real-time monitoring of prosthetic socket conditions and enables early detection of
deviations associated with tissue stress.
This work contributes to applied physics by addressing nonlinear inverse problems in coupled thermodynamic-biomechanical systems, with implications for adaptive sensing in wearable and biomedical devices.
| Apply for student award at which level: | PhD |
|---|---|
| Consent on use of personal information: Abstract Submission | Yes, I ACCEPT |