With the goal of net-zero expected to be accomplished in recent decades, the development of a thermoelectric generator, one of the energy harvesting technologies, is important. Along with efforts to discover more cost-effective thermoelectric materials, geometric and structural optimization of thermoelectric generators is essential to maximize power and efficiency. This work demonstrates a segmented thermoelectric generator, one of the advanced structures of a thermoelectric generator, modeling using artificial neural networks. After training the artificial neural networks, we have achieved 98.9% accuracy compared to COMSOL simulation results under constant temperature difference while speeding up the computational speed over a few thousand times. This new approach illustrates the advantages of the modeling of segmented thermoelectric generators.
Keywords artificial neural network, segmented thermoelectric generator, modeling