Surface temperature estimation in determined multi-wavelength pyrometry systems.

Area of Science:

• Thermodynamics
• Optical Engineering
• Statistical Modeling

Background:

• Multi-wavelength pyrometry relies on Wien's law for temperature estimation.
• Modeling surface spectral emissivity with polynomial functions presents accuracy challenges.
• Simple linear models can yield unacceptably low temperature estimates in pyrometry.

Purpose of the Study:

• To develop and evaluate a multi-wavelength pyrometry model.
• To investigate the impact of emissivity functions on temperature estimation accuracy.
• To apply ridge regression for improving temperature estimation in multi-wavelength pyrometry.

Main Methods:

• Derived a multi-wavelength pyrometry model using Wien's law.
• Modeled surface spectral emissivity as a polynomial function of wavelength.
• Simulated temperature outputs using various emissivity functions and computed estimates via a simple linear model.
• Applied ridge regression to address ill-posed systems and enhance accuracy.

Main Results:

• Simple linear models showed variable accuracy depending on emissivity functions.
• Ridge regression substantially increased temperature estimation accuracy, particularly for higher-order systems.
• Estimation errors decreased by approximately 52% with 3 to 20 channels.
• Errors reduced by over 65% using ridge regression (20 channels) compared to the linear model (2 channels).

Conclusions:

• Ridge regression offers a robust solution for improving multi-wavelength pyrometry accuracy.
• The effectiveness of ridge regression increases with the number of pyrometer channels.
• This statistical technique holds significant potential for precise high-temperature measurements.