Hard Modeling refers to data analysis with mechanistic models (Hard Models). In contrast to statistical models (Soft Models), mechanistic models are based on physico-chemical principles which provide these models with many beneficial properties, ranging from simple interpretability to very good robustness against shifting peaks.
With our analysis software PEAXACT you get easy access to spectral Hard Modeling methods, especially methods Hard Modeling Factor Analysis (HMFA) and Indirect Hard Modeling (IHM) you will find in PEAXACT exclusively. Both methods use intelligent ways to exploit the physical knowledge about spectra being composed of sums of peak-shaped signals.
Some advantages of Hard Modeling methods:
- In contrast to Peak Integration and Peak Fitting, IHM and HMFA both also work well in case of strongly overlapping and shifting peaks. They are particularly suited for MIR, Raman, and NMR spectra.
- The methods are robust and therefore well-suited for the application in process analytics.
- In contrast to statistical models, Hard Models can be calibrated with comparatively few reference samples, i.e. with reduced calibration costs.
- Hard Models can be used for extrapolation, i.e. for the analysis of mixtures which are difficult to calibrate such as reactive mixtures.
- In addition, hard models are easy to interpret which enables a fast detection of errors if analysis results were outside the expected range.
HMFA and IHM have been developed at RWTH Aachen University and are published.