Segmented thermoelectric generator modelling and optimization using artificial neural networks by iterative training

Renewable energy technologies are central to emissions reduction and essential to achieve net-zero emission.Segmented thermoelectric generators(STEG)facilitate more efficient thermal energy recovery over a large temperature gradient.However,the additional design complexity has introduced challenges in the modelling and optimization of its performance.In this work,an artificial neural network(ANN)has been applied to build accurate and fast forward modelling of the STEG.More importantly,we adopt an iterative method in the ANN training process to improve accuracy without increasing the dataset size.This approach strengthens the proportion of the high-power performance in the STEG training dataset.Without increasing the size of the training dataset,the relative prediction error over high-power STEG designs decreases from 0.06 to 0.02,representing a threefold improvement.Coupling with a genetic algorithm,the trained artificial neural networks can perform design optimization within 10 s for each operating condition.It is over 5,000 times faster than the optimization performed by the conventional finite element method.Such an accurate and fast modeller also allows mapping of the STEG power against different parameters.The modelling approach demonstrated in this work indicates its future application in designing and optimizing complex energy harvesting technologies.

Accurate inverse design of Fabry–Perot-cavity-based color filters far beyond sRGB via a bidirectional artificial neural network

Structural color based on Fabry–Perot(F-P) cavity enables a wide color gamut with high resolution at submicroscopic scale by varying its geometrical parameters. The ability to design such parameters that can accurately display the desired color is therefore crucial to the manufacturing of F-P cavities for practical applications.This work reports the first inverse design of F-P cavity structure using deep learning through a bidirectional artificial neural network. It enables the production of a significantly wider coverage of color space that is over 215% of sRGB with extremely high accuracy, represented by an average ΔE_(2000) value below 1.2. The superior performance of this structural color-based neural network is directly ascribed to the definition of loss function in the uniform CIE 1976-Lab color space. Over 100,000 times improvement in the design efficiency has been demonstrated by comparing the neural network to the metaheuristic optimization technique using an evolutionary algorithm when designing the famous painting of "Haystacks, end of Summer" by Claude Monet. Our results demonstrate that, with the correct selection of loss function, deep learning can be very powerful to achieve extremely accurate design of nanostructured color filters with very high efficiency.