Search
6 articles for “Neural network surrogate”
-
AI-Assisted Optimization of Supersonic Airfoil Shapes Using CFD Coupling
Abstract: This paper presents a novel framework for optimizing supersonic airfoil geometries through integrated artificial intelligence and computational fluid dynamics coupling. Traditional gradient-based optimization methods for high-speed aerodynamic shapes suffer from computational expense and convergence difficulties in non-convex design spaces. The proposed methodology employs a deep neural network surrogate model trained on high-fidelity Reynolds-Averaged Navier-Stokes solutions to approximate aerodynamic performance metrics across the design space. A hybrid particle swarm-genetic algorithm searches …
Published in Journal of Aerospace Engineering & Technology · Vol. 16, Issue 1, 2026 Read article
-
Machine Learning Optimization for VARTM Carbon Polymer Laminates
Abstract: Vacuum-assisted resin transfer moulding (VARTM) is a key low-cost, out-of-autoclave process for manufacturing large-scale carbon-fibre reinforced polymer (CFRP) laminates crucial to aerospace wings, wind-turbine blades, marine hulls, and automotive structures. Unpredictable resin flow often leads to voids, dry spots, and race-tracking defects, resulting in 27.9% scrap rates and lengthy, costly trial-and-error design cycles. Although surrogate models provide rapid impregnation predictions for simple flat-plate geometries, vision-based monitoring is limited to idealized …
Published in Journal of Polymer & Composites · Vol. 14, Issue 3, 2026 · pp. 229–245 Read article
-
Artificial Intelligence–Assisted Reduced-Order Modeling and Stability Control in Granular Couette Flow
Abstract: This study develops a reduced-order and stability-aware modeling framework for dense granular Couette flow by integrating continuum mechanics, bifurcation analysis, and data-driven stability estimation. Starting from coupled governing equations for momentum, granular temperature, and microstructural evolution, the system is nondimensionalized and reduced using a Galerkin projection consistent with shear-driven boundary conditions. This yields a low-dimensional nonlinear dynamical system that preserves the essential coupling between velocity, fluctuation energy, and structural relaxation. …
Published in Emerging Trends in Chemical Engineering · Vol. 13, Issue 3, 2026 Read article
-
Machine Learning Assisted Optimization of Nanoscale MOSFET Parameters Using TCAD Simulation
Abstract: This paper presents a machine learning (ML) assisted framework for the multi-objective optimization of nanoscale bulk n-channel metal-oxide-semiconductor field-effect transistors (nMOSFETs) with a 10 nm physical gate length, high-k HfO₂ gate dielectric, and TiN metal gate. Technology computer-aided design (TCAD) simulations employing drift-diffusion transport, Shockley-Read-Hall recombination, Lombardi mobility degradation, and density- gradient quantum correction models are used to generate a parametric dataset of 2,400 device configurations spanning gate length (L), …
Published in Journal of Microelectronics and Solid State Devices · Vol. 13, Issue 1, 2026 · pp. 10–19 Read article
-
Neuromorphic Spiking Neural Networks and Event-Driven Hardware: Architectures, Learning Paradigms, and Experimental Realizations
Abstract: With the current computational boom the research community is seeking for more sustainable energy efficient i.e. biologically inspired models of conventional Artificial Neural Networks (ANNs). Spiking Neural Networks (SNNs) known as the third generation of neural network models, provide a revolutionary approach by mimicking the asynchronized event-driven and temporally accurate signaling of the mammalian brain. Whereas conventional deep learning models operate with real-valued activations and dense matrix multiplications, SNNs use …
Published in Current Trends in Signal Processing · Vol. 16, Issue 2, 2026 Read article
-
AI-Accelerated Development of Gradient Polymer Nanocomposite Thin Films
Abstract: Gradient polymer nanocomposite thin films are an active field of materials research due to the fact that it enables scientists to de-facto regulate the optical, electrical, and mechanical properties of a film by merely altering its composition on a layer-by-layer basis. This type of control opens the gate to the improved flexible electronics, long lasting protective coats, and the new smart gadgets. The problem is, though, that it is a …
Published in Journal of Polymer & Composites · Vol. 14, Issue 2, 2026 · pp. 456–469 Read article