Meaning
Near-infrared spectroscopy on polymer pellets and powders is frequently affected by variations in particle size and packing density that alter the optical path length. This physical variation is minimized by the standard normal variate transformation, which normalizes each individual spectrum by subtracting its mean and dividing by its standard deviation. It reduces the multiplicative scaling effects caused by light scattering without losing the chemical information contained in the absorption bands.
The method is used in continuous monitoring systems for raw material verification and blend analysis.
Scattering Correction
Light traveling through a bed of polymer pellets is scattered in different directions depending on the shape and size of each pellet. This scattering increases the path length of the light, making the absorption peaks appear artificially high or low. By applying the standard normal variate algorithm, the spectral data are adjusted to remove these physical scattering influences.
This adjustment allows the sensor to measure the true chemical composition of the pellets, even when their size distribution varies between batches.
Calibration Optimization
Development of chemometric calibration models is simplified when physical variation is removed from the training data. Standard normal variate preprocessing reduces the complexity of the calibration equations, making the models more stable over long periods and across different instruments. This stability is essential for transferring calibrations from a development spectrometer in the laboratory to a production sensor on the compounding line.
Calculation Step
Calculation of the normalized spectrum is done independently for each scan, requiring no reference standards or master spectra. Each data point is centered and scaled based entirely on the statistical properties of that single spectrum, which streamlines the computation process. This independence makes it highly suitable for real-time monitoring of rapid extrusion runs.
