Measuring thermal conductivity of materials at room temperature in atmosphere by using a continuous-wave laser and neural network model

Abstract

To measure the thermal conductivity (TC) of materials at room temperature in the atmosphere via a continuous-wave laser, a new model differing from the existing heat transfer models for calculating the TC was constructed by combining a neural network (NN) with the finite element method (FEM). Massive FEM samples simulating the heat conduction process of specimens were generated to realise feature engineering and constructing an NN model for TC prediction. The accuracy of the NN model was validated through the experimental data of several samples measured using a self-developed apparatus equipped with a continuous-wave laser source. The maximum relative error between the predicted and real TC values was approximately 6%. The presented NN model is suitable for materials with thermal diffusivities less than 1 × 10−5 m2 s−1, corresponding to most ceramics and ceramic-based composites.