Spectrometry-based sensors, because of their dependence on indirect measurements, require the incorporation of a calibration model into their operation. Typically, this calibration model is established on the basis of representative samples that encompass variations in the variable of interest. However, the use of these sensors poses challenges to the robustness of the calibration model. These challenges arise from the inherent variability of external factors, including, but not limited to, temperature fluctuations, alterations in chemical composition or changes in the spectrometer itself.
Resolving these robustness issues requires a strategic approach, which will be outlined in the following discussion, based on available knowledge of the influencing factors. To mitigate the impact of external variations on the calibration model, various correction methods will be elucidated, offering insightful insight into the techniques employed to improve the accuracy and reliability of spectrometry-based sensors.
A more detailed exploration will then be undertaken, focusing on robust modeling methods as a means of fortifying the calibration model against external influences. Particular attention will be paid to methodologies using orthogonal projections and domain adaptation, examining their effectiveness and application in ensuring the stability and accuracy of spectrometry-based sensor measurements.