Machine learning-assisted rapid electromagnetic design of flexible graphene-based absorptive composites

Electromagnetic interference and pollution have become increasingly prevalent due to the widespread use of electronic devices. Consequently, the development of high-performance electromagnetic wave-absorbing materials has emerged as a critical research focus. However, the design of traditional wave-absorbing materials is often limited by conventional trial-and-error methods, especially in the absence of well-defined physical mechanisms, complicating the direct design and optimization of material properties. In this study, a dielectric constant prediction and reflection optimization system (DCPRO) was developed. This system first establishes a machine learning model to predict the electromagnetic properties of flexible graphene-based wave-absorbing composites based on key processing parameters. Subsequently, a dual-layer impedance gradient design is achieved based on the expanded database from the predictions. The dual-layer impedance gradient design exhibited a broad absorption bandwidth (EAB) in the frequency range of 3.29–18 GHz, with a minimum reflection loss (RL_min) of −56.08 dB. The use of machine learning-based predictions allows for the extensive expansion of the database, significantly reducing the research cycle. The standardized DCPRO process provides a more efficient alternative to traditional experimental methods, thus accelerating the development and commercialization of advanced microwave absorbing materials.