Neural Networks Solutions of Thermistor Problem Tuned by Genetic Algorithms and Gradient Descent Method

C. Wongsathan and N. Suyaroj (Thailand)


Thermistor, neural networks, genetic algorithms, and gradient descent method.


This paper presents approximate steady-state solutions of a one-dimensional Positive Temperature Coefficient (PTC) thermistor problem having a ramp electrical conductivity. The predicted temperature distributions are obtained by using Neural Networks (NNs) model adjusted their initial parameters by Genetic Algorithms (GAs) and approved by Gradient Descent Methods (GDM). It is shown that numerical solutions exhibit the correct physical characteristics of the problem, and they are in good agreement with the exact solution and also those obtained by earlier authors.

Important Links:

Go Back