An official website of the United States government
Abstract Monitoring machine health and product quality enables predictive maintenance that optimizes repairs to minimize factory downtime. Data-driven intelligent manufacturing often relies on probabilistic techniques with intractable distributions. For example, generative models of data distributions can balance fault classes with synthetic data, and sampling the posterior distribution of hidden model parameters enables prognosis of degradation trends. Normalizing flows can address these problems while avoiding the training instability or long inference times of other generative Deep Learning (DL) models like Generative Adversarial Networks (GAN), Variational Autoencoders (VAE), and diffusion networks. To evaluate normalizing flows for manufacturing, experiments are conducted to synthesize surface defect images from an imbalanced data set and estimate parameters of a tool wear degradation model from limited observations. Results show that normalizing flows are an effective, multi-purpose DL architecture for solving these problems in manufacturing. Future work should explore normalizing flows for more complex degradation models and develop a framework for likelihood-based anomaly detection. Code is available at https://github.com/uky-aism/flows-for-manufacturing.
more »
« less
Reconstructing the Tropical Pacific Upper Ocean Using Online Data Assimilation With a Deep Learning Model
Abstract A deep learning (DL) model, based on a transformer architecture, is trained on a climate‐model data set and compared with a standard linear inverse model (LIM) in the tropical Pacific. We show that the DL model produces more accurate forecasts compared to the LIM when tested on a reanalysis data set. We then assess the ability of an ensemble Kalman filter to reconstruct the monthly averaged upper ocean from a noisy set of 24 sea‐surface temperature observations designed to mimic existing coral proxy measurements, and compare results for the DL model and LIM. Due to signal damping in the DL model, we implement a novel inflation technique by adding noise from hindcast experiments. Results show that assimilating observations with the DL model yields better reconstructions than the LIM for observation averaging times ranging from 1 month to 1 year. The improved reconstruction is due to the enhanced predictive capabilities of the DL model, which map the memory of past observations to future assimilation times.
more »
« less
- Award ID(s):
- 2202526
- PAR ID:
- 10625170
- Publisher / Repository:
- American Geophysical Union
- Date Published:
- Journal Name:
- Journal of Advances in Modeling Earth Systems
- Volume:
- 16
- Issue:
- 11
- ISSN:
- 1942-2466
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Near-wall flow simulation remains a central challenge in aerodynamics modelling: Reynolds-averaged Navier–Stokes predictions of separated flows are often inaccurate, and large-eddy simulation (LES) can require prohibitively small near-wall mesh sizes. A deep learning (DL) closure model for LES is developed by introducing untrained neural networks into the governing equations and training in situ for incompressible flows around rectangular prisms at moderate Reynolds numbers. The DL-LES models are trained using adjoint partial differential equation (PDE) optimization methods to match, as closely as possible, direct numerical simulation (DNS) data. They are then evaluated out-of-sample – for aspect ratios, Reynolds numbers and bluff-body geometries not included in the training data – and compared with standard LES models. The DL-LES models outperform these models and are able to achieve accurate LES predictions on a relatively coarse mesh (downsampled from the DNS mesh by factors of four or eight in each Cartesian direction). We study the accuracy of the DL-LES model for predicting the drag coefficient, near-wall and far-field mean flow, and resolved Reynolds stress. A crucial challenge is that the LES quantities of interest are the steady-state flow statistics; for example, a time-averaged velocity component $$\langle {u}_i\rangle (x) = \lim _{t \rightarrow \infty } ({1}/{t}) \int _0^t u_i(s,x)\, {\rm d}s$$ . Calculating the steady-state flow statistics therefore requires simulating the DL-LES equations over a large number of flow times through the domain. It is a non-trivial question whether an unsteady PDE model with a functional form defined by a deep neural network can remain stable and accurate on $$t \in [0, \infty )$$ , especially when trained over comparatively short time intervals. Our results demonstrate that the DL-LES models are accurate and stable over long time horizons, which enables the estimation of the steady-state mean velocity, fluctuations and drag coefficient of turbulent flows around bluff bodies relevant to aerodynamics applications.more » « less
-
Abstract Exploring new techniques to improve the prediction of tropical cyclone (TC) formation is essential for operational practice. Using convolutional neural networks, this study shows that deep learning can provide a promising capability for predicting TC formation from a given set of large-scale environments at certain forecast lead times. Specifically, two common deep-learning architectures including the residual net (ResNet) and UNet are used to examine TC formation in the Pacific Ocean. With a set of large-scale environments extracted from the NCEP–NCAR reanalysis during 2008–21 as input and the TC labels obtained from the best track data, we show that both ResNet and UNet reach their maximum forecast skill at the 12–18-h forecast lead time. Moreover, both architectures perform best when using a large domain covering most of the Pacific Ocean for input data, as compared to a smaller subdomain in the western Pacific. Given its ability to provide additional information about TC formation location, UNet performs generally worse than ResNet across the accuracy metrics. The deep learning approach in this study presents an alternative way to predict TC formation beyond the traditional vortex-tracking methods in the current numerical weather prediction. Significance StatementThis study presents a new approach for predicting tropical cyclone (TC) formation based on deep learning (DL). Using two common DL architectures in visualization research and a set of large-scale environments in the Pacific Ocean extracted from the reanalysis data, we show that DL has an optimal capability of predicting TC formation at the 12–18-h lead time. Examining the DL performance for different domain sizes shows that the use of a large domain size for input data can help capture some far-field information needed for predicting TCG. The DL approach in this study demonstrates an alternative way to predict or detect TC formation beyond the traditional vortex-tracking methods used in the current numerical weather prediction.more » « less
-
ABSTRACT We forecast the prospects for cross-correlating future line intensity mapping (LIM) surveys with the current and future Ly-α forest measurements. Using large cosmological hydrodynamic simulations, we model the emission from the CO rotational transition in the CO Mapping Array Project LIM experiment at the 5-yr benchmark and the Ly-α forest absorption signal for extended Baryon Acoustic Oscillations (BOSS), Dark energy survey instrument (DESI), and Prime Focus multiplex Spectroscopy survey (PFS). We show that CO × Ly-α forest significantly enhances the detection signal-to-noise ratio (S/N) of CO, with up to $$300{{\ \rm per\, cent}}$$ improvement when correlated with the PFS Ly-α forest survey and a 50–75 per cent enhancement with the available eBOSS or the upcoming DESI observations. This is competitive with even CO × spectroscopic galaxy surveys. Furthermore, our study suggests that the clustering of CO emission is tightly constrained by CO × Ly-α forest due to the increased sensitivity and the simplicity of Ly-α absorption modelling. Foreground contamination or systematics are expected not to be shared between LIM and Ly-α forest observations, providing an unbiased inference. Ly-α forest will aid in detecting the first LIM signals. We also estimate that [C ii] × Ly-α forest measurements from Experiment for Cryogenic Large-Aperture Intensity Mapping and DESI/eBOSS should have a larger S/N than planned [C ii] × quasar observations by about an order of magnitude.more » « less
-
Abstract We evaluate Linear Inverse Models (LIMs) trained on last millennium model data to predict Arctic sea‐ice concentration, thickness, and other atmospheric and oceanic variables on monthly timescales. We find that more than 500 years of training data and 100 years of validation data are needed to reliably estimate LIM forecast skill. The best LIM has skill up to 8 months lead time and outperforms an autoregressive model of order one (AR1) forecast at all locations, with particularly large outperformance near the ice edge. However, for out‐of‐sample validation tests using data from various different model simulations and reanalysis products, they underperform an AR1 model due to differences in the location of the sea‐ice edge from the training data. We present a metric for predicting LIM forecast skill, based on the spatial correlation of the variance in the training and validation data sets.more » « less
