Intelligent computing for electromagnetohydrodynamic bioconvection flow of micropolar nanofluid with thermal radiation and stratification: Levenberg–Marquardt backpropagation algorithm

AIP Advances, Год журнала: 2024, Номер 14(3)

Опубликована: Март 1, 2024

The Levenberg–Marquardt (LM) backpropagation optimization algorithm, an artificial neural network is used in this study to perform integrated numerical computing evaluate the electromagnetohydrodynamic bioconvection flow of micropolar nanofluid with thermal radiation and stratification. model then reduced a collection boundary value problems, which are solved help technique proposed scheme, i.e., LM iterative approach determine minimum nonlinear function defined as sum squares. As blend steepest descent Gauss–Newton method, it has become typical for least-squares problems. Furthermore, stability consistency algorithm ensured. For validation purposes, results also compared those previous research MATLAB bvp4c solver. Neural networking utilized velocity, temperature, concentration profile mapping from input output. These findings demonstrate accuracy forecasts optimizations produced by networks. performance solver, reduce mean square error, generalize dataset. network-based operates using data based on ratio testing (13%), (17%), training (70%). This stochastic work presents activation log-sigmoid tens neurons hidden output layers solving learning language model. overlapping small computed absolute errors, range 10−3 10−10 106 108 each class, indicate algorithm. case’s regression evaluated if were ideal In addition, fitness histogram validate dependability Numerical approaches networks excellent combination fluid dynamics, could lead new advancements many domains. contribute systems, resulting increased efficiency production across various technical

Язык: Английский

---

Spiking neural networks for nonlinear regression

Royal Society Open Science, Год журнала: 2024, Номер 11(5)

Опубликована: Май 1, 2024

Spiking neural networks (SNN), also often referred to as the third generation of networks, carry potential for a massive reduction in memory and energy consumption over traditional, second-generation networks. Inspired by undisputed efficiency human brain, they introduce temporal neuronal sparsity, which can be exploited next-generation neuromorphic hardware. Energy plays crucial role many engineering applications, instance, structural health monitoring. Machine learning contexts, especially data-driven mechanics, focuses on regression. While regression with SNN has already been discussed variety publications, this contribution, we provide novel formulation its accuracy efficiency. In particular, network topology decoding binary spike trains real numbers is introduced, using membrane spiking neurons. Several different architectures, ranging from simple feed-forward complex long short-term are derived. Since proposed architectures do not contain any dense layers, exploit full terms At same time, demonstrated numerical examples, namely material models. Linear nonlinear, well history-dependent models, examined. contribution mechanical interested reader may regress custom function adapting published source code.

Язык: Английский

---

An accuracy-enhanced transonic flow prediction method fusing deep learning and a reduced-order model

Physics of Fluids, Год журнала: 2024, Номер 36(5)

Опубликована: Май 1, 2024

It is difficult to accurately predict the flow field over an aircraft in presence of shock waves due its strong nonlinear characteristics. In this study, we developed accuracy-enhanced prediction method that fuses deep learning and a reduced-order model achieve accurate for various aerodynamic shapes. Herein, establish convolutional neural network/proper orthogonal decomposition (CNN-POD) mapping geometries overall field. Then, local regions containing structures can be identified by POD reconstruction build enhanced model. A CNN established map The proposed was applied two cases involving transonic airfoils. results indicate reduce error properties with values ranging from 13% 66.27%. Additionally, demonstrates better efficiency robustness comparison existing methods, it also address problem complex multiple structures.

Язык: Английский

---

An enhanced hybrid deep neural network reduced-order model for transonic buffet flow prediction

Aerospace Science and Technology, Год журнала: 2022, Номер 126, С. 107636 - 107636

Опубликована: Май 13, 2022

Язык: Английский

---

On the prediction of the turbulent flow behind cylinder arrays via Echo State Networks

Machine Learning Science and Technology, Год журнала: 2024, Номер 5(3), С. 035005 - 035005

Опубликована: Июнь 4, 2024

Abstract This study aims at the prediction of turbulent flow behind cylinder arrays by application Echo State Networks (ESN). Three different arrangements seven cylinders are chosen for current study. These represent regimes: single bluff body flow, transient and co-shedding flow. allows investigation flows that fundamentally originate from wake yet exhibit highly diverse dynamics. The data is reduced Proper Orthogonal Decomposition (POD) which optimal in terms kinetic energy. Time Coefficients POD Modes (TCPM) predicted ESN. network architecture optimized with respect to its three main hyperparameters, Input Scaling (INS), Spectral Radius (SR), Leaking Rate (LR), order produce best predictions Weighted Prediction Score (WPS), a metric leveling statistic deterministic prediction. In general, ESN capable imitating complex dynamics even longer periods several vortex shedding cycles. Furthermore, mutual interdependencies TCPM well preserved. However, hyperparameters depend strongly on characteristics. Generally, as become faster more intermittent, larger LR INS values result better predictions, whereas less clear trends SR observable.

Язык: Английский

---

Dynamic stall modeling of wind turbine blade sections based on a data-knowledge fusion method

Energy, Год журнала: 2024, Номер 305, С. 132234 - 132234

Опубликована: Июль 3, 2024

Язык: Английский

---

A new composite neural network with spatiotemporal features extraction capability for unsteady flow fields predictions

Physics of Fluids, Год журнала: 2025, Номер 37(2)

Опубликована: Фев. 1, 2025

Artificial intelligence based on neural network technology has provided innovative methods for predicting unsteady flow fields. However, both purely data-driven and single physics-driven can only perform short-term predictions fields are unable to achieve medium- long-term predictions. A composite CNN-GRU-PINN (CGPINN) is proposed by combining convolutional (CNN), gated recurrent unit (GRU), physics-informed (PINN). CNN GRU used learn the spatial temporal characteristics of flows, respectively. PINN adopted constrain field prediction data according physical laws. The around a circular cylinder employed verify performances CGPINN. test results show that compared PINN, reconstruction accuracy CGPINN improved about 86.10% average, 96.18%. Compared pure approaches, an average 65.71%. Additionally, exhibits better robustness, demonstrating insensitivity variations in sample size noise levels, thereby ensuring stable reliable across diverse conditions. This study more accurate robust method

Язык: Английский

---

Unsteady flow prediction based on a hybrid network-operator learning model

Physics of Fluids, Год журнала: 2025, Номер 37(2)

Опубликована: Фев. 1, 2025

Fast and accurate forecasting of unsteady flow is a challenge for both fluid dynamics machine learning study. In this paper, hybrid network-operator model proposed to predict the spatiotemporal flow. The first integrates Fourier neural operator (FNO) convolutional long short-term memory (ConvLSTM) network in parallel. systematically evaluated three typical problems, including two-dimensional turbulence, around stationary cylinder, an oscillated airfoil. Numerical experiments indicate that exhibits higher accuracy than vanilla FNO ConvLSTM model. Additionally, shows excellent performance on long-period prediction can be well generalized cases with different dimensionless parameters (such as Reynolds number Mach number).

Язык: Английский

---

Data-knowledge-driven dynamic stall modeling guided by stall patterns and semi-empirical model

Physics of Fluids, Год журнала: 2025, Номер 37(4)

Опубликована: Апрель 1, 2025

Dynamic stall often causes unsteady loads and negatively affects the aerodynamic performance of aircraft. Thus, accurate modeling dynamic stalls is crucial for aircraft design. With development machine learning, existing data-driven models always rely on extensive, costly training data but lack physical knowledge, which limits their generalizability interpretability. Therefore, this study proposes a data-knowledge-driven procedure. First, by exploring damping evolution moment coefficient, three distinct patterns are identified. A transitional state, significantly differs from both deep light stall, proposed to assist neural network modeling. Subsequently, with P-based degree classification force component developed, integrates Leishman–Beddoes model. This model provides unified approach predict aerodynamics across different degrees stall. Compared purely network, incorporating expert knowledge improved generalization accuracy 50%. Moreover, insights reduce reliance high-precision network.

Язык: Английский

---

A Unified Vortex-Induced Vibration Plot Incorporating Scaling Parameter Derived by Symbolic Regression

Опубликована: Янв. 1, 2025

Язык: Английский

---