The idea of using artificial neural network in measurement system with hot probe for testing parameters of heat-insulating materials

Stanislaw Chudzika,
a.Institute of Electronics and Control Systems, Czestochowa University of Technology, al. Armii Krajowej 17, 42-200 Czestochowa, Poland
Received 30 April 2007;
revised 17 November 2008;
accepted 29 December 2008.
Available online 14 January 2009.

Abstract
The article presents a mathematical model of a measurement system with hot probe for testing thermal parameters of heat-insulating materials. Currently in situ measurement of thermal conductivity is widely done by the line heat source (LHS) method. The basic problem with this method is the number and type of the assumptions needed. In this study, another method was proposed to measure the thermal parameters by using an artificial neural network. The model of a nonstationary heat flow process in the sample of material with hot probe and auxiliary thermometer is based on a two-dimensional heat-conduction model. For solving a system of partial differential equations that describe the model, the finite element method (FEM) was applied. The measurement system uses an artificial neural network (ANN) to estimate the coefficients of inverse heat conduction problem for solid. The network determines the value of effective thermal conductivity and effective thermal diffusivity on the basis of temperature responses of hot probe and auxiliary thermometer. In developing of the ANN model, several configurations were evaluated. The optimal ANN model was capable of predicting the thermal conductivity values with a relative error <1%. The influence of measurands errors on identified values of the thermal parameters was analysed. Learning process and simulation analyses were conducted in the Matlab environment. It is possible to implement the architecture of a trained neural network with a simple microcontroller embedded system.

Keywords: Neural networks; Thermal conductivity; Thermal diffusivity; Inverse heat conduction problem
Article Outline
1. Introduction
2. The idea of measurement system with hot probe
3. Mathematic model and results of the simulation of a nonstationary heat flow process in the sample
4. The application of artificial neural networks for thermal parameters identification
5. Results of simulations
6. Conclusion
Acknowledgements
References

Taken from Sciencedirect.