Adversarial autoencoders and adversarial LSTM for improved forecasts of urban air pollution simulations

Abstract

This paper presents an approach to improve the forecast of computational fluid dynamics (CFD) simulations of urban air pollution using deep learning, and most specifically adversarial training. This adversarial approach aims to reduce the divergence of the forecasts from the underlying physical model. Our twostep method integrates a Principal Components Analysis (PCA) based adversarial autoencoder (PC-AAE) with adversarial Long short-term memory (LSTM) networks. Once the reduced-order model (ROM) of the CFD solution is obtained via PCA, an adversarial autoencoder is used on the principal components time series. Subsequentially, a Long Short-Term Memory network (LSTM) is adversarially trained on the latent space produced by the PC-AAE to make forecasts. Once trained, the adversarially trained LSTM outperforms a LSTM trained in a classical way. The study area is in South London, including three-dimensional velocity vectors in a busy traffic junction.

Generalised Latent Assimilation in Heterogeneous Reduced Spaces with Machine Learning Surrogate Models

“…Both proper orthogonal decomposition (POD)-type (e.g., [2,16,3,12]) and neural networks (NNs)-based autoencoding methods [14,1] have been used to construct the reduced-order latent spaces. The work of [3] is extended in [17] which relies on an Adversarial RNN when the training dataset is insufficient. In terms of compression accuracy, much effort has been devoted to compare the performance of different auto-encoding approaches.…”

“…We provide both a theoretical and numerical analysis (based on a high-dimensional CFD application) of the proposed method. The surrogate models we build are based on AE and long short-term memory (LSTM) technologies which have been shown to provide stable and accurate solutions for ROMs [17]. • We provide a theoretical error upper-bound for the expectation of the cost function in LA when using the local surrogate polynomial function instead of the original DL function.…”

An AI-based Domain-Decomposition Non-Intrusive Reduced-Order Model for Extended Domains applied to Multiphase Flow in Pipes

“…whilst the first use of a convolutional autoencoder came 16 years later and was applied to Burgers Equation, advecting vortices and lid-driven cavity flow [31]. In the few years since 2018, many papers have appeared, in which convolutional autoencoders have been applied to sloshing waves, colliding bodies of fluid and smoke convection [32]; flow past a cylinder [33][34][35]; the Sod shock test and transient wake of a ship [36]; air pollution in an urban environment [37][38][39]; parametrised time-dependent problems [40]; natural convection problems in porous media [41]; the inviscid shallow water equations [42]; supercritical flow around an airfoil [43]; cardiac electrophysiology [44]; multiphase flow examples [45]; the Kuramoto-Sivashinsky equation [46]; the parametrised 2D heat equation [47]; and a collapsing water column [48]. Of these papers, those which compare autoencoder networks with POD generally conclude that autoencoders can outperform POD [31,33], especially when small numbers of reduced variables are used [41][42][43][44].…”

“…Therefore, any set of latent variables should be associated, through the decoder, with a realistic output. Not many examples exist of using an AAE for dimensionality reduction in fluid dynamics problems, however, it has been applied to model air pollution in an urban environment [38,39]. In this work we compare POD, CAE, AAE and the SVD-AE on flow past a cylinder and multiphase flow in a pipe, to assess their suitability as dimension reduction methods.…”

“…From the references in this paper alone, many examples exist of feed-forward and recurrent neural networks having been used for the purpose of learning the evolution of time series data, for example, by Multi-layer perceptrons [12,13,40,41,43,[54][55][56][57][58][59][60], Gaussian Process Regression [11,45,[61][62][63] and Long-Short Term Memory networks [31,32,34,35,38,51,64]. When using these types of neural network to predict in time, if the reduced variables stray outside of the range of values encountered during training, the neural network can produce unphysical, divergent results [39,51,52,64,65]. To combat this, a number of methods have been proposed.…”