Some methods of training radial basis neural networks in solving the Navier‐Stokes equations

Abstract

Summary The purpose of this research is to analyze the application of neural networks and specific features of training radial basis functions for solving 2‐dimensional Navier‐Stokes equations. The authors developed an algorithm for solving hydrodynamic equations with representation of their solution by the method of weighted residuals upon the general neural network approximation throughout the entire computational domain. The article deals with testing of the developed algorithm through solving the 2‐dimensional Navier‐Stokes equations. Artificial neural networks are widely used for solving problems of mathematical physics; however, their use for modeling of hydrodynamic problems is very limited. At the same time, the problem of hydrodynamic modeling can be solved through neural network modeling, and our study demonstrates an example of its solution. The choice of neural networks based on radial basis functions is due to the ease of implementation and organization of the training process, the accuracy of the approximations, and smoothness of solutions. Radial basis neural networks in the solution of differential equations in partial derivatives allow obtaining a sufficiently accurate solution with a relatively small size of the neural network model. The authors propose to consider the neural network as an approximation of the unknown solution of the equation. The Gaussian distribution is used as the activation function.

Neural ordinary differential equations for ecological and evolutionary time‐series analysis

“…In this paper, we consider a specific class of ANNs, called single layer perceptrons (SLPs), which is a widespread and mathematically convenient choice due to their simplicity (e.g. Aarts & Van Der Veer, 2001; Funahashi & Nakamura, 1993; Mai et al., 2016; Sinchev et al., 2018; Wu et al., 2005). They are the most simple form of ANNs as they correspond to a weighted sum of basis functions , more formallywhere are the parameters of the network.…”

“…Hybrid approaches combining ANNs and ODEs have been applied with success in the past across a range of fields. For example, physicists have recently applied approaches combining partial differential equations (PDEs) and ANNs to model fluid dynamics (Sinchev et al., 2018), and to discover governing physical equations from data (Berg & Nyström, 2019). Neuroscientists have modelled the activation dynamics of interacting populations of neurons in an Ape with a NODE‐like system, faithfully reproducing the behaviour of motoneurons (Sussillo et al., 2015).…”

Research on Discrete Dynamic System Modeling of Vocal Performance Teaching Platform Based on Big Data Environment

“…e basic implementation process of this step is as follows: for individuals, s = (C1, C2, CN, CN + 1). However, it is not very easy for such individuals to code [17]. Because if the coding is not correct, the illegal student sequence or invalid solution will appear when the crossover or neuron transfer operation is implemented.…”

Characteristics of the Waviness End-Face Mechanical Seal in Reactor Coolant Pump Considering the Viscosity-Temperature Effect

“…The shaft seal is an isolation device to prevent the leakage of reactor coolant pumps (RCP) (Salant et al, 2018), which is directly related to the safe operation of the nuclear power plant (Nilsson et al, 2009;Baraldi et al, 2011;Feng et al, 2016;Kok and Benli, 2017;Luqman et al, 2019). As one of the key components of shaft seals, the mechanical seal (Harp and Salant, 1998;Clark et al, 2002;Simon, 2018) plays a vital role in the safety and reliability of the entire structure, where different shapes of its end face lead to the different hydraulic effects (hydrodynamic and hydrostatic effects). In comparison with the hydrostatic seals, hydrodynamic seals can reach better performance owing to its hydrodynamic effect (Pascovici and Etsion, 1992;Batten et al, 2008).…”

“…Most of the previous works on mechanical seals are mainly based on the bearing lubrication theory, in which the Reynolds equation (Chandramoorthy, Hadjiconstantinou;Gustafsson et al, 2017) is often used as the governing equation to solve the pressure distribution of the liquid film in the seal ring gap. The Reynolds equation is a simplified approximation of the Navier-Stokes equation (Temam, 2001;Pietarila Graham et al, 2008;Sinchev et al, 2018;Danchin and Mucha, 2019) to solve the narrow gap laminar flow. In this approximation, the pressure change along the film thickness direction is neglected by assuming that the thickness of the seal ring gap liquid film is small enough.…”