Mathematical Model
Non-linear relationships between raw sensor signals and physical properties are described using a complex algebraic equation. In textile testing, polynomial regression is used to create calibration curves that translate electronic pulses into fiber diameter measurements in microns. This method allows the software to account for the slight non-linearity found in most optical detectors.
Calibration Curve
Technicians run a series of certified reference materials through the machine to gather data points across a wide range of values. The software then calculates the best-fitting curve that passes through these points. A second or third-order equation often provides the most accurate mapping of the instrument response.
This curve becomes the internal reference for all subsequent production testing.
Data Fit
A curved line improves the accuracy of the machine at the extreme ends of the measurement scale compared to a simple linear model. Very fine or very coarse fibers often produce signals that do not follow a perfectly linear pattern. The regression model ensures that the instrument remains precise across the entire spectrum of animal fibers.
Range Constraint
Extrapolating the curve beyond the values of the reference materials can lead to measurable errors. For this reason, the calibration is only valid within the boundaries established during the initial setup. Reliable mills verify the fit of their regression models frequently using independent check samples.