Soft calibration technique with SVM for intelligent temperature measurement using thermocouple

An adaptive soft calibration circuit using the technique of Support Vector Machine (SVM) is reported in this paper for temperature measurement using thermocouple. The objectives of the present work are: (i) to extend linearity range of measurement to 100% of full scale input range, (ii) to make measurement technique adaptive to variations in temperature coefficients (different type of thermocouple). SVM model is trained to map output voltage of data conversion unit to produce linear characteristics with input temperature and independent of variations in temperature coefficients all within certain range. Further, the proposed technique is tested with simulated data. Fulfilment of objectives set was tested by simulation results, and the results shows a root mean square of percentage error in terms of 0.34%.