TITLE
TEMPERATURE SENSOR BASED ON A THERMISTOR WITH A NEGATIVE TEMPERATURE COEFFICIENT AND LINEARIZATION VIA NEURAL NETWORK
AUTHOR(S)
Ventsislav Simonov*, Nikolay Shopov
ABSTRACT
An experimental setup was implemented in which, using measuring equipment and a controlled heat source, the electrical and temperature characteristics of a commercially available thermistor with a negative temperature coefficient were recorded. The data is used to implement an analog circuit to compensate for the nonlinear characteristic. By selecting the inflection point when tuning a Wheatstone bridge, a root mean square error of 1.7 mV was achieved over the operating temperature range. The characteristics of the implemented analog cir-
cuit are compared with a signal linearization method using a feedforward artificial neural network with a back propagation algorithm. This method showed significantly higher linearization performance. A root mean square error value of only 200μV was achieved at operating range.
DOI
DOWNLOAD
https://unitech.tugab.bg/images/2025/dokladi/2-Electronics%20and%20Sensors/p240_s2_u199_id281.pdf
How to cite this article:
Ventsislav Simonov*, Nikolay Shopov, TEMPERATURE SENSOR BASED ON A THERMISTOR WITH A NEGATIVE TEMPERATURE COEFFICIENT AND LINEARIZATION VIA NEURAL NETWORK, UNITECH – SELECTED PAPERS - 2025