Answers are grounded in this paper and cite their sources.
This paper presents a novel hybrid analytical–computational framework based on the bilinear neural network method (BNNM) for solving the fourth-order Ablowitz–Kaup–Newell–Segur (4AKNS) water wave equation with perturbation. By employing Hirota's bilinear transformation, the original nonlinear partial differential equation is systematically converted into a bilinear form, providing the construction of a test function representation. Instead of directly approximating the physical solution, the proposed approach learns the test function using a feedforward neural network combined with automatic differentiation. To enhance training stability and convergence, a softmax-based adaptive weighting strategy is incorporated into the loss function, allowing the model to focus on low-residual regions. Several analytical solution structures, including kink waves, soliton–periodic interactions, and periodic waves, are constructed and validated through multiple case studies. Numerical experiments demonstrate that the proposed method achieves stable convergence, high accuracy, and strong robustness across diverse solution types. The results confirm that the bilinear neural network framework is an effective and flexible tool for solving nonlinear integrable systems, providing both accurate approximations and the ability to maintain complex wave dynamics.
No takes yet. Share an insight, caveat, or question.
Tuan et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: