Self-Refining Kernel-Aware Physics-Informed Neural Network for Nonlinear Singular Integral Equations

Document Type : Research article

Authors

1 Department of Mathematics, Kongu Engineering College, Erode, India

2 Canadian Quantum Research Center, 106-460 Doyle Ave, Kelowna, British Columbia V1Y 0C2, Canada

3 Department of Statistics, University of Mazandaran, Babolsar, Iran

4 Payame Noor University (PNU), Tehran, Iran

5 Department of Physics, Sar.C., Islamic Azad University, Sari, Iran

6 Department of Mathemstics, University of Mazandaran, Babolsar, Iran

Abstract

Nonlinear weakly singular integral equations are challenging to solve numerically because the singular behaviour of the kernel can reduce the effectiveness of conventional uniform discretization and physics-informed learning strategies. This study develops a Self-Refining Kernel-Aware Physics-Informed Neural Network (SKA-PINN) for nonlinear weakly singular Fredholm and Volterra integral equations. The proposed framework combines a kernel-aware difficulty indicator based on residual magnitude, residual variation, and kernel complexity with adaptive residual weighting, self-refining collocation, and adaptive quadrature. Instead of distributing computational points uniformly, the method identifies numerically difficult regions and progressively increases the local resolution during training. The framework is evaluated using nonlinear Fredholm and Volterra benchmark problems with prescribed analytical solutions. For the Fredholm problem, the maximum reported absolute and relative errors are $6.3\times10^{-6}$ and $3.4\times10^{-6}$, respectively. For the Volterra problem, the maximum absolute error is $7.5\times10^{-6}$, with a maximum relative error of $1.2\times10^{-4}$. The refinement experiments show a consistent reduction in the maximum residual and solution error as the collocation and quadrature resolutions increase. In comparison with the classical and adaptive PINN configurations considered, SKA-PINN achieves the smallest reported $L^2$ error of $1.09\times10^{-5}$ while requiring fewer training epochs and lower CPU time. The ablation results further show that kernel-aware weighting and self-refinement both contribute to the overall improvement. These results indicate that incorporating kernel information directly into the learning and refinement process provides an effective computational strategy for nonlinear weakly singular integral equations.

Keywords

Main Subjects


 

Article PDF

[1] I. V. Nadal, J. Stiasny, and S. Chatzivasileiadis, Physics-Informed Neural Networks: A plug and play integration into power system dynamic simulations, Electric Power Systems Research, 248, 111885, (2025).
[2] Z. Ren, S. Zhou, D. Liu, and Q. Liu, EPINN: Enhanced physics-informed neural network for solving continuous integral equations, Journal of Artificial Intelligence Research, 84, (2025).
[3] G. H. Tipu, A. Y. Cherati, F. Yao, and H. Momeni, Data-driven soliton solutions and parameter discovery to the fourth-order nonlinear AKNS equation: A hybrid analytical and physics-informed neural network approach, Nonlinear Dynamics, 113(18), 25047–25069, (2025).
[4] J. Das, B. Bhaumik, S. De, and S. Changdar, Physics-informed neural network with symbolic regression for deriving analytical approximate solutions to nonlinear partial differential equations, Neural Computing and Applications, 37(24), 20205–20240, (2025).
[5] E. Lee, Least-squares enhanced physics-informed learning for singular and ill-posed partial differential equations, Computers & Mathematics with Applications, 206, 301–315, (2026).
[6] J. Wang and Y. Yang, Data-driven solutions and nonlinear dynamics for the integrable fractional nonlinear Schrödinger equation via physics-informed neural networks, Applied Mathematics and Computation, 528, 130121, (2026).
[7] J. Nishanth and S. Elango, A comparative study of optimization algorithms in physics-informed neural networks for singularly perturbed differential equations, Applied Soft Computing, 115496, (2026).
[8] K. Luo, J. Zhao, Y. Wang, J. Li, J. Wen, J. Liang, et al., Physics-informed neural networks for PDE problems: A comprehensive review, Artificial Intelligence Review, 58(10), 323, (2025).
[9] Y. Ding, S. Chen, H. Miyake, and X. Li, Physics-informed neural networks with Fourier features for seismic wavefield simulation in time-domain nonsmooth complex media, IEEE Transactions on Geoscience and Remote Sensing, 63, 1–13, (2025).
[10] H. Mofidi, M. Raissi, and W. Yang, Exploring multiple timescale dynamics using geometric singular perturbation-informed neural networks (GSPINNs), SIAM Journal on Scientific Computing, 48(2), C301–C329, (2026).
[11] D. A. Pratama, M. A. Bakar, S. Surono, A. Salhi, N. Baini Ismail, and N. Mohamed, Revolutionizing high-dimensional PDE solutions: A novel approach of integrated truncated singular value decomposition to enhance physics-informed neural networks, Neural Processing Letters, (2026).
[12] S. Agarwal and L. N. Mishra, A comparative study of the fractional differential equations using physics-informed neural networks, AIMS Mathematics, 11(8), 25705–25743, (2026).
[13] Y. V. Ngueabou and S. D. Oloniiju, Integrating spectral methods with neural network architectures: A review of hybrid approaches to solving differential equation, Archives of Computational Methods in Engineering, 1–57, (2025).
[14] Z. Y. Sun, Y. T. Guo, K. Ge, C. Hou, and Z. Z. Hu, ROM-PINN: A physics-informed neural network with reduced-order modelling for nonlinear structural response prediction, Structures, 86, 111350, (2026).
[15] A. K. Singh, M. Mehra, and R. Pulch, Non-local physics informed neural networks for forward and inverse problems containing non-local operators, Neural Computing and Applications, 37(6), 4111–4132, (2025).
[16] N. Ramya, M. Akhoundi, M. NouriJouybari and F. Mofidnakhaei, Artificial neural network and physics-informed neural network assisted analytical solutions for nonlinear fractional biological models using local fractional operators, Analytical and Numerical Solutions for Nonlinear Equations, 11(2), 220–237, (2026).
[17] G. Muhiuddin, N. Ramya, E. Alshaban, F. Maqsood, M. Al-Matrafi and A. H. Alshehri, Hybrid SWCNT/MWCNT nanofluid flow over porous surfaces with thermal radiation and microbial bioconvection under Soret–Dufour coupling, Journal of Nanotechnology, 2026(1), 5831339, (2026).
[18] G. Muhiuddin, N. Ramya, M. Deivanayaki, F. Maqsood and A. H. Alshehri, Analysis of thermophoretic and Brownian motion effects on Newtonian and non-Newtonian fluid flow through a porous stretching surface, Frontiers in Physics, 14, 1746700, (2026).
[19] A. Fallah, Y. Talebi Rostam and N. Ramya, Fuzzy-based numerical investigation of MHD nanofluid flow over a stretching surface with Darcy-Forchheimer porous medium effects, Caspian Journal of Mathematical Sciences, 14(2), 410–423, (2025).
[20] A. Aliyari Boroujeni, R. Pourgholi and S. H. Tabasi, A new improved teaching–learning-based optimization (ITLBO) algorithm for solving nonlinear inverse partial differential equation problems, Computational and Applied Mathematics, 42, 99, (2023).
[21] A. A. Boroujeni, R. Pourgholi and S. H. Tabasi, Numerical solutions of KDV and mKDV equations: Using sequence and multi-core parallelization implementation, Journal of Computational and Applied Mathematics, 454, 116184, (2025). 
Volume 12, Issue 1
June 2027
Pages 171-190
  • Receive Date: 31 August 2026
  • Revise Date: 16 October 2026
  • Accept Date: 17 October 2026
  • Publish Date: 02 November 2026