Predicting the Output Performance of Triboelectric Nanogenerators Using Highly Representative Data‐Based Neural Networks

Abstract

Triboelectric nanogenerators (TENGs) are promising potential sustainable power sources for wireless sensing networks within the Internet of Things (IoT) realm. Developing an efficient TENG evaluation model, characterized by high speed, accuracy, and representativeness, facilitates its integration into practical applications, which is urgent and lack of investigation currently. Herein, an artificial intelligence (AI) based evaluation model is developed to predict the performance of freestanding rotational TENGs (FR‐TENGs) for demonstration. An accurate and representative train dataset is essential for development of AI‐based evaluation model, which has been generated using finite element analysis and equivalent circuit simulation alongside the non‐dominated sorting genetic algorithm II. Through comprehensive experiments and simulations, the accuracy of the model has been verified in predicting the power output performance of FR‐TENGs, which has 99.6% (three design parameters) and 99.2% (seven design parameters) maximum train set accuracy. More importantly, the predicted results from the AI‐based evaluation model have notably expanded the coverage of data and significantly expedited the generation time from days to seconds. Herein, the use of AI in assessing the performance of TENGs is enhanced. The TENG design process can be significantly simplified, while maintaining a high evaluation model accuracy, thus promising advancements of IoT applications in future.