Software based data driven approaches for K type thermocouple linearization

Authors

  • Prashant Dattatraya Sarawade Government Polytechnic, Karad, Satara, Maharashtra-415124

Keywords:

‘K‟ type thermocouple,Linearization,ANN, BPN.

Abstract

Thermocouple based temperature measurement system although widely used in industries has to overcome nonlinearity issues in precision measurement. The software techniques handle the situation leading to complexity in architecture. This research article presents a design of robust linearizer using data driven approaches such as Backpropagation neural network (BPN) and polynomial fitting techniques. The neural network architecture regarding the number of layers, hidden neurons in each layer is a design challenge. BPN finds the best solution to overcome nonlinearity in the sensors has to overcome problems in training, testing and validation. Linear fit is simple but not sufficiently accurate data driven approach and higher order polynomials requiredfor accurate estimation which needs more experimental points for coefficient calculations. The output data from the „K‟ type thermocouple is given to both the approaches. The comparative result analysis is reported regarding mean square error (MSE) for both the approaches shows that polyfit outperform the BPN and linearfit approaches over span of measurement.

Downloads

Published

2023-09-28

How to Cite

Prashant Dattatraya Sarawade. (2023). Software based data driven approaches for K type thermocouple linearization. Chinese Journal of Computational Mechanics, (5), 188–193. Retrieved from http://jslxxb.cn/index.php/jslxxb/article/view/4348

Issue

Section

Articles