Journal
APPLIED MATHEMATICAL MODELLING
Volume 40, Issue 11-12, Pages 5964-5977Publisher
ELSEVIER SCIENCE INC
DOI: 10.1016/j.apm.2016.01.034
Keywords
One-dimensional Bratu equations; Neural networks; Genetic algorithms; Nonlinear ODEs; Hybrid techniques; Active-set method
Ask authors/readers for more resources
In this study, a bio-inspired computing approach is developed to solve Bratu-type equations arising in modeling of electrically conducting solids and various other physical phenomena. We employ feed-forward artificial neural networks (ANN) optimized with genetic algorithm (GA) and the active-set method (ASM). The mathematical formulation consists of an ANN with an unsupervised error, which is minimized by tuning weights of the network. The evolutionary technique based on GAs is used as a tool for global search of the weights in conjunction with the ASM for rapid local convergence. The designed methodology is applied to solve a number of initial and boundary value problems based on Bratu equations. Monte Carlo simulations and their statistical analyses are used to validate accuracy, convergence and effectiveness of the scheme. Comparison of results is made with exact solutions, the fully explicit Runge-Kutta numerical method, and other reported solutions of analytical and numerical solvers to establish correctness of the designed scheme. (C) 2016 Elsevier Inc. All rights reserved.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available