Abstract
In this paper, an improved heuristic algorithm based on the disturbance and somersault foraging grey wolf optimizer (IDSFGWO) is proposed to optimize the design of multilayer wideband microwave absorbers for normal incidence. The multilayer absorber is designed to reduce maximum reflection coefficient by choosing suitable layers of materials from a predefined database. Three improvement strategies are given to enhance the performance of GWO, including tent map, nonlinear perturbation, and somersault foraging. The optimization results show that the reflection coefficients optimized by IDSFGWO are better than those of other algorithms for multilayer absorber design.
Introduction
Microwave absorbers are critical components in stealth technology and electromagnetic compatibility, designed to minimize radar cross-sections by dissipating incident electromagnetic energy. While single-layer absorbers are relatively simple to implement, they often suffer from narrow bandwidths and excessive thickness. Multilayer absorbers address these limitations by stacking diverse materials to broaden the absorption spectrum. However, the design process involves selecting optimal material sequences and layer thicknesses from a vast database, resulting in a complex, non-linear optimization problem prone to "local optima" where standard algorithms fail to find the global best solution.
Methodology
The proposed design approach utilizes an Improved Grey Wolf Optimizer (IDSFGWO). Standard GWO mimics the social hierarchy and hunting strategies of grey wolves. To enhance its performance for electromagnetic design, three modifications were introduced:
- Tent Map Initialization: Replaces random initialization with a chaotic tent map to ensure a more uniform distribution of the initial population across the search space.
- Nonlinear Perturbation Factor: Adjusts search parameters nonlinearly to better balance global exploration and local exploitation.
- Somersault Foraging Strategy: Inspired by manta rays, this allows the algorithm to "flip" its search position around the current best solution, facilitating escape from local traps during the final stages of optimization.
The physical performance of the multilayer structures is evaluated using the Equivalent Transmission Line Method, which models each material layer as a circuit segment to compute the total reflection coefficient.
Key Insights and Ideas
A significant challenge in multilayer absorber design is the "Local Optima" hurdle, where optimization algorithms like Particle Swarm Optimization (PSO) plateau at suboptimal solutions. The IDSFGWO's somersault strategy is specifically designed to overcome this by exploring the vicinity of the best solution more dynamically.
Furthermore, the optimization process revealed a strong synergy among materials, particularly relaxation-type magnetic materials, which were frequently selected for high-performance designs due to their frequency-dependent properties. The study also highlights the engineering trade-off between absorption depth and bandwidth; increasing the number of layers from 5 to 7 significantly expanded the bandwidth but slightly reduced the peak absorption.
Key Results
The IDSFGWO was benchmarked against standard GWO, PSO, and the Whale Optimization Algorithm (WOA) across different design scenarios:
- 5-Layer Design (2–12 GHz): IDSFGWO achieved a maximum reflection coefficient of -34.2 dB, significantly outperforming standard GWO (-28.0 dB) and PSO (-14.7 dB).
- 7-Layer Design (0.1–20 GHz): Despite the much broader frequency range, the improved algorithm maintained a reflection of -30.5 dB, whereas the next best algorithm (GWO) only reached -21.5 dB.
Statistical analysis over 20 independent runs confirmed the stability of IDSFGWO, which exhibited the lowest standard deviation among the tested algorithms, indicating consistent convergence to high-quality solutions rather than relying on a lucky random start.
Conclusion
The IDSFGWO algorithm demonstrates superior performance in designing multilayer microwave absorbers, offering significantly lower reflection coefficients and wider bandwidths compared to traditional heuristic methods. While current results focus on normal incidence and a 10mm thickness constraint, future work could extend this optimizer to oblique incidence angles and multi-objective optimization considering weight and structural integrity. This improved optimizer holds promise for various complex electromagnetic engineering problems, such as antenna design or metasurface optimization.
Summary updated