Efficient Extreme Learning Machine-Based Method for Time-Domain Electromagnetic Simulations

This letter proposes a novel time-domain electromagnetic simulation method based on the extreme learning machine (ELM). The governing equations of the proposed method are weighted Laguerre orthogonal polynomials-based Maxwell's equations, leading to unconditional stability. Domain decomposition and C$^{k}$ continuity conditions are introduced to enhance the efficiency of ELM. To truncate the computational domain, ELM-based wave-absorbing formulation derived from the one-way wave equation is proposed here. Numerical examples demonstrate that the proposed method achieves significant computational efficiency improvements while maintaining high accuracy compared with the Laguerre finitedifference time-domain (FDTD) method.