Nonlinear Calibration of Thermocouple Sensor Using Least Squares Support Vector Regression
Calibration is an important operation in the instrumentation industry for determining the relationship between the output (s) (or response) of a measuring instrument and the value of the input (s). This paper proposes a nonlinear calibration method based on least squares support vector regression (LS-SVR) with the output voltage of thermocouple sensor as input and the measured temperature output to eliminate the nonlinear errors in detection process. Firstly, the nonlinear calibrator, expressed by power series, was established based on the principle of inverse model. And then the parameters of the calibrator were identified by LS-SVR. Through this calibrator, the desired linear characteristics of thermocouple sensor could be obtained. Finally, platinum-rhodium 30– platinum-rhodium 6-thermocouple (B-type) was taken as an example, and experimental results show that the proposed calibration method is efficient in the temperature range from 400°C to 1800°C. And the method has an advantage of analytical expression of the calibration model.
Y. J. Yu "Nonlinear Calibration of Thermocouple Sensor Using Least Squares Support Vector Regression", Advanced Materials Research, Vols. 443-444, pp. 302-308, 2012