Andy C. C. Tan’s research while affiliated with Tunku Abdul Rahman University and other places

... The results indicate that the average relative error of the BRBP model in predicting the flow rate is between 1% and 4%, and that of the head prediction is between 1% and 2%. Xu et al. 41 employed two descriptors to capture mesh information and structural features in neural networks, refining these features through mesh convolution layers. This enabled them to accurately predict the hydrodynamic performance of a horizontal axis tidal turbine. ...

Reference:

Optimization of organic Rankine cycle turbine expander based on radial basis function neural network and nondominated sorting genetic algorithm II

... 17 Xu proposed an innovative coupling approach to solve the complexity of FSI prediction, which can significantly improve the efficiency and structural integrity of FSI prediction. 18 Moreira used the computational fluid dynamics (CFD) FSI numerical method to analyze the performance of double oscillating webbed wing structures in ship propulsion and wave energy absorption. 19 Zhang examined the effects of hydrofoil spacing and stiffness on propulsion. ...

Reference:

Optimizing fluid–structure interaction dynamic characteristic of flexible webbed wings in wave glider under super-elastic deformation

... The ANN-Jensen model [108] improves the prediction of wake-induced power losses. In the paper by Luo et al. [109], DWFE used combined ML models for dynamic wake estimation. The HHO-XGBoost model proposed by Dong et al. [110] predicts offshore turbine vibrations. ...

Reference:

A Comprehensive Review of Machine Learning Models for Optimizing Wind Power Processes

... 44,45 This study refers to the blade structure and material layout of the NREL S814 HATT from the literature and employs the finite element software CalculiX to assess the structural performance of the blades. 46,47 As illustrated in Fig. 4, the composite blade consists of the skin, shear webs, and spar caps. Specifically, the blade skin is made of unidirectional carbon fiber reinforced polymer (UD CFRP) and primarily bears torsional and partial bending loads. ...

Reference:

A novel generative approach to the parametric design and multi-objective optimization of horizontal axis tidal turbines

... This approach significantly reduces computational costs while maintaining high accuracy in structural performance predictions. Similarly, Xu et al. [56] developed the DLFSI model, a deep learning framework that integrates Convolutional Neural Networks (CNN) with Blade Element Momentum theory and the Finite Element Method (FEM). This model facilitates multi-objective optimization of tidal turbine blades, predicting hydrodynamic performance and structural stresses while achieving a computational speedup of 19 times compared to traditional methods. ...

Reference:

Numerical Simulation of Earthquake Impacts on Marine Structures: A Comprehensive Review

... Examples include convolutional neural networks mostly with UNet structures (i.e. see [47,48]), artificial neural networks with autoencoder architectures [49], and physics-informed frameworks [42,50]. ...

Reference:

Wind Field Reconstruction and Uncertainty Quantification at Wildland Fires Based on Sparse UAV-based Wind Measurements

... Notably, techniques like proper orthogonal decomposition (POD) 38 and dynamic mode decomposition (DMD) 39 have been integrated into DDML to compress data dimensions effectively while preserving essential physical information. For instance, Luo et al. 40,41 proposed a machine learning approach using DDML with sparse data for wind turbine wake flow estimation, offering a computationally efficient and cost-effective solution for wind energy optimization. These methods are especially beneficial for addressing the inherent nonlinearities in wake flows, providing a deeper insight compared to conventional linear models. ...

Reference:

Innovative sparse data reconstruction approaches for yawed wind turbine wake flow via data-driven and physics-informed machine learning

... Li et al. 40 employed long short-term memory (LSTM) networks to predict the unsteady aerodynamics of airfoils. Luo et al. 41 proposed a multiscaled autoencoder (MS-AE) framework for reconstructing the missing flow field in an experimental hydrofoil. However, these applications predominantly concentrate on fundamental two-dimensional airfoil predictions, exhibiting fewer applications in the domain of 3D fluid machinery. ...

Reference:

TurbineNet: Advancing tidal turbine blade hydrodynamic performance prediction with neural networks

... Liu et al. [17] proposed a static convolutional neural network (SCNN) and a multiple temporal paths convolutional neural network to simultaneously input spatiotemporal information to improve the performance of the model. Chen et al. [18] extended and modified the down-sampled skip-connection and multiscale network (DSC/MS) framework [19] to improve a 16-fold increase in resolution for global wind turbine wake simulations. To more accurately reconstruct the turbulent flow field structures, Yang et al. [20] attempted to process three-dimensional flow field data using a back-projection network. ...

Reference:

A multi-scale hybrid attention Swin-transformer-based model for the super-resolution reconstruction of turbulence

... [9][10][11] Analytical models, such as the Jensen model, 12 Frandsen model, 13 Bastankhah model, 14 and the three-dimensional wake models, 15 provide rapid predictions but inadequately capture real-world measurements due to their oversimplifications. [16][17][18][19][20] In light of these challenges, data-driven approaches, particularly employing machine learning techniques, have emerged as promising tools for wake flow estimation and have increasingly been applied in various aspects of wind power systems, including wind turbine wake prediction, [21][22][23][24] power forecasting, [25][26][27] load prediction, [28][29][30] and wind turbine control. [31][32][33] These approaches derive insights from data to offer potential for rapid and cost-effective solutions. ...

Reference:

Innovative sparse data reconstruction approaches for yawed wind turbine wake flow via data-driven and physics-informed machine learning