A review of the artificial neural network surrogate modeling in aerodynamic design

Abstract

Artificial neural network surrogate modeling with its economic computational consumption and accurate generalization capabilities offers a feasible approach to aerodynamic design in the field of rapid investigation of design space and optimal solution searching. This paper reviews the basic principle of artificial neural network surrogate modeling in terms of data treatment and configuration setup. A discussion of artificial neural network surrogate modeling is held on different objectives in aerodynamic design applications, various patterns of realization via cutting-edge data technique in numerous optimizations, selection of network topology and types, and other measures for improving modeling. Then, new frontiers of modern artificial neural network surrogate modeling are reviewed with regard to exploiting the hidden information for bringing new perspectives to optimization by exploring new data form and patterns, e.g. quick provision of candidates of better aerodynamic performance via accumulated database instead of random seeding, and envisions of more physical understanding being injected to the data manipulation.

Non-intrusive reduced-order modeling for fluid problems: A brief review

“…In practice, ANNs have been widely used in a wide range of application fields, including image recognition, speech recognition, and natural language processing as well as aerodynamic optimization. 53 The architecture of an ANN is variable and task-dependent. In this paper, multi-layer feedforward ANNs, known as the most common ANN model, will be described.…”

The Objective Space and the Formulation of Design Requirement in Natural Laminar Flow Optimization

“…The solution proposed in this paper is done via Principal Component Analysis (PCA). PCA is used to strengthen preprocessing of data to improve manipulation efficiency [21,22], dimension reduction [23], and in the sample selection of initialization [24], etc. The methods for data manipulation have been utilized in aerodynamic designs, but mostly in design space than in objective space.…”

“…The methods for data manipulation have been utilized in aerodynamic designs, but mostly in design space than in objective space. In fact, apart from extending the subjects of aerodynamic design to complex aircraft components [25][26][27] and strengthening the link between geometry and aerodynamic performance [21,[28][29][30], aerodynamic designers deployed data preparation in design space for preprocessing, e.g., dimensional reduction by parameter filtering according to physical intuitiveness(e.g., cutting out geometry parameters with less impact on ultimate aerodynamic design goal [31]) and feature extracting from flowfield for economical use of data [32,33], etc. It is noted that the intention of PCA in this paper is not simply for variable reduction but for extracting the main modes from objective space considered to be the essential information that may help establish a more reasonable formulation of design requirement in aerodynamic designs.…”

DNN-assisted statistical analysis of a model of local cortical circuits

“…In neuroscience, DNNs, specifically the hierarchical convolutional neural networks, have been used to model single-unit and population responses in higher visual cortical areas 10 . Our DNN-assisted approach falls into the general framework of surrogate-based modeling, a well established practice in engineering with wide applications to many problems that involve complex simulations or experiments (see 11 – 13 for reviews). In biology, the use of surrogate models has been more limited but there are precedents, e.g., support vector machines have been recently explored in hemorrhage and renal denervation 14 and yeast mating polarization 15 .…”