Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Neural Network”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,315 records · Page 73

EFIT-Prime: Probabilistic and physics-constrained reduced-order neural network model for equilibrium reconstruction in DIII-D

We introduce EFIT-Prime, a novel machine learning surrogate model for EFIT (Equilibrium FIT) that integrates probabilistic and physics-informed methodologies to overcome typical limitations associated with deterministic and ad hoc neural network architectures. EFIT-Prime utilizes a neural architecture search-based deep ensemble for robust uncertainty quantification, providing scalable and efficient neural architectures that comprehensively quantify both data and model uncertainties. Physically informed by the Grad–Shafranov equation, EFIT-Prime applies a constraint on the current density J tor and a smoothness constraint on the first derivative of the poloidal flux, ensuring physically plausible solutions. Furthermore, the spatial location of the diagnostics is explicitly incorporated in the inputs to account for their spatial correlation. Extensive evaluations demonstrate EFIT-Prime's accuracy and robustness across diverse scenarios, most notably showing good generalization on negative-triangularity discharges that were excluded from training. Timing studies indicate an ensemble inference time of 15 ms for predicting a new equilibrium, offering the possibility of plasma control in real-time, if the model is optimized for speed.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Structural Health Monitoring and Impact Detection Using Neural Networks for Damage Characterization

Detection of damage due to foreign object impact is an important factor in the development of new aerospace vehicles. Acoustic waves generated on impact can be detected using a set of piezoelectric transducers, and the location of impact can be determined by triangulation based on the differences in the arrival time of the waves at each of the sensors. These sensors generate electrical signals in response to mechanical motion resulting from the impact as well as from natural vibrations. Due to electrical noise and mechanical vibration, accurately determining these time differentials can be challenging, and even small measurement inaccuracies can lead to significant errors in the computed damage location. Wavelet transforms are used to analyze the signals at multiple levels of detail, allowing the signals resulting from the impact to be isolated from ambient electromechanical noise. Data extracted from these transformed signals are input to an artificial neural network to aid in identifying the moment of impact from the transformed signals. By distinguishing which of the signal components are resultant from the impact and which are characteristic of noise and normal aerodynamic loads, the time differentials as well as the location of damage can be accurately assessed. The combination of wavelet transformations and neural network processing results in an efficient and accurate approach for passive in-flight detection of foreign object damage.

Ross, Richard W.↗

System Identification of X-33 Neural Network

Modern flight control research has improved spacecraft survivability as its goal. To this end we need to have a failure detection system on board. In case the spacecraft is performing imperfectly, reconfiguration of control is needed. For that purpose we need to have parameter identification of spacecraft dynamics. Parameter identification of a system is called system identification. We treat the system as a black box which receives some inputs that lead to some outputs. The question is: what kind of parameters for a particular black box can correlate the observed inputs and outputs? Can these parameters help us to predict the outputs for a new given set of inputs? This is the basic problem of system identification. The X33 was supposed to have the onboard capability of evaluating the current performance and if needed to take the corrective measures to adapt to desired performance. The X33 is comprised of both rocket and aircraft vehicle design characteristics and requires, in general, analytical methods for evaluating its flight performance. Its flight consists of four phases: ascent, transition, entry and TAEM (Terminal Area Energy Management). It spends about 200 seconds in ascent phase, reaching an altitude of about 180,000 feet and a speed of about 10 to 15 Mach. During the transition phase which lasts only about 30 seconds, its altitude may increase to about 190,000 feet but its speed is reduced to about 9 Mach. At the beginning of this phase, the Main Engine is Cut Off (MECO) and the control is reconfigured with the help of aerosurfaces (four elevons, two flaps and two rudders) and reaction control system (RCS). The entry phase brings down the altitude of X33 to about 90,000 feet and its speed to about Mach 3. It spends about 250 seconds in this phase. Main engine is still cut off and the vehicle is controlled by complex maneuvers of aerosurfaces. The last phase TAEM lasts for about 450 seconds and the altitude and speed, both are reduced to zero. The present attempt, as a start, focuses only on the entry phase. Since the main engine remains cut off in this phase, there is no thrust acting on the system. This considerably simplifies the equations of motion. We introduce another simplification by assuming the system to be linear after some non-linearities are removed analytically from our consideration. Under these assumptions, the problem could be solved by Classical Statistics by employing the least sum of squares approach. Instead we chose to use the Neural Network method. This method has many advantages. It is modern, more efficient, can be adapted to work even when the assumptions are diluted. In fact, Neural Networks try to model the human brain and are capable of pattern recognition.

Aggarwal, Shiv↗

Neural network based prediction of no-wall β N limits due to ideal external kink instabilities

Multi-layer perceptron based neural networks (NNs) are trained to predict the Troyon no-wall beta limits due to the onset of low-n (n = 1 – 3, n is the toroidal mode number) ideal external kink instabilities in tokamak plasmas. It is revealed that a well-trained NN can predict the n = 1 no-wall beta limit within 10% relative error. The NN performance is somewhat worse for the n = 2 and 3 no-wall beta limits, but still a relative error within 20% is achievable. The trained NNs well reproduce the known dependences of the no-wall beta limits on the plasma pressure (pressure peaking factor) and current (plasma internal inductance) profiles. Other scalings are further easily established with NNs, for parametric dependences such as on the aspect ratio, the elongation and triangularity of the plasma boundary shape. These NN-based Troyon beta limit predictors can be incorporated into integrated modeling platforms, or directly implemented as a real time stability estimator for the purpose of disruption avoidance or mitigation during experiments.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Multi-task graph neural networks for simultaneous prediction of global and atomic properties in ferromagnetic systems *

Abstract We introduce a multi-tasking graph convolutional neural network, HydraGNN, to simultaneously predict both global and atomic physical properties and demonstrate with ferromagnetic materials. We train HydraGNN on an open-source ab initio density functional theory (DFT) dataset for iron-platinum with a fixed body centered tetragonal lattice structure and fixed volume to simultaneously predict the mixing enthalpy (a global feature of the system), the atomic charge transfer, and the atomic magnetic moment across configurations that span the entire compositional range. By taking advantage of underlying physical correlations between material properties, multi-task learning (MTL) with HydraGNN provides effective training even with modest amounts of data. Moreover, this is achieved with just one architecture instead of three, as required by single-task learning (STL). The first convolutional layers of the HydraGNN architecture are shared by all learning tasks and extract features common to all material properties. The following layers discriminate the features of the different properties, the results of which are fed to the separate heads of the final layer to produce predictions. Numerical results show that HydraGNN effectively captures the relation between the configurational entropy and the material properties over the entire compositional range. Overall, the accuracy of simultaneous MTL predictions is comparable to the accuracy of the STL predictions. In addition, the computational cost of training HydraGNN for MTL is much lower than the original DFT calculations and also lower than training separate STL models for each property.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Convolutional Neural Networks for the CHIPS Neutrino Detector R&D Project.

The CHerenkov detectors In mine PitS (Chips) neutrino detector R&D project aims to develop novel strategies and technologies for very large yet ‘cheap as chips’ water Cherenkov neutrino detectors. Via deployment in a body of water, use of commercially available components, and instrumentation coverage optimisation for the study of exclusively accelerator beam neutrinos, Chips will enable megaton scale detectors to become a reality at the cost of $200k-$300k per kt of sensitive mass. During the summer of 2019 a prototype Chips detector, Chips-5, was deployed into the Wentworth 2W disused mine pit in northern Minnesota, 7 mrad off the NuMI beam axis. A novel data acquisition system was introduced using cheap single-board computers and open-source software. This work presents a novel approach to water Cherenkov neutrino detector event reconstruction and classification. Three forms of a Convolutional Neural Network, a type of deep learning algorithm, have been trained to reject cosmic muon events, classify beam events, and estimate neutrino energies, all using only the raw detector event as input. When evaluated on the expected distribution of Chips-5 events, this new approach is shown to be robust and explainable as well as providing a significant performance increase over the standard likelihood-based reconstruction and simple neural network classification. Promisingly, the performance presented here is comparable to the more complex (and expensive) neutrino oscillation experiments within the field.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Automated assignment of rotational spectra using artificial neural networks

Here, a typical broadband rotational spectrum may contain several thousand observable transitions, spanning many species. While these spectra often encode troves of chemical information, identifying and assigning the individual spectra can be challenging. Traditional approaches typically involve visually identifying a pattern. A more modern approach is to apply an automated fitting routine. In this approach, combinations of 3 transitions are searched by trial and error, to fit the A , B , and C rotational constants in a Watson-type Hamiltonian. In this work, we develop an alternative approach-to utilize machine learning to train a computer to recognize the patterns inherent in rotational spectra. Broadband high-resolution rotational spectra are perhaps uniquely suited for pattern recognition, assignment, and species identification using machine learning. Repeating patterns of transition frequencies and intensities are now routinely recorded in broadband chirped-pulse Fourier transform microwave experiments in which both the number of resolution elements and the dynamic range surpass 10 4 . At the same time, these high-resolution spectra are extremely sensitive to molecular geometry with each polar species having a unique rotational spectrum. Here we train the feed forward neural network on thousands of rotational spectra that we calculate, using the rules of quantum mechanics, from randomly generated sets of rotational constants and other Hamiltonian parameters. Reasonable physical constraints are applied to these parameter sets, yet they need not belong to existing species. A trained neural network presented with a spectrum identifies its type (e.g., linear molecule, symmetric top, or asymmetric top) and infers the corresponding Hamiltonian parameters (rotational constants, distortion, and hyperfine constants). The classification and prediction times, about 160 μs and 50 μs, respectively, seem independent of the spectral complexity or the number of molecular parameters. We describe how the network works, provide benchmarking results, and discuss future directions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

AP-Net: An atomic-pairwise neural network for smooth and transferable interaction potentials

Intermolecular interactions are critical to many chemical phenomena, but their accurate computation using ab initio methods is often limited by computational cost. The recent emergence of machine learning (ML) potentials may be a promising alternative. Useful ML models should not only estimate accurate interaction energies but also predict smooth and asymptotically correct potential energy surfaces. However, existing ML models are not guaranteed to obey these constraints. Indeed, systemic deficiencies are apparent in the predictions of our previous hydrogen-bond model as well as the popular ANI-1X model, which we attribute to the use of an atomic energy partition. As a solution, we propose an alternative atomic-pairwise framework specifically for intermolecular ML potentials, and we introduce AP-Net—a neural network model for interaction energies. The AP-Net model is developed using this physically motivated atomic-pairwise paradigm and also exploits the interpretability of symmetry adapted perturbation theory (SAPT). We show that in contrast to other models, AP-Net produces smooth, physically meaningful intermolecular potentials exhibiting correct asymptotic behavior. Initially trained on only a limited number of mostly hydrogen-bonded dimers, AP-Net makes accurate predictions across the chemically diverse S66x8 dataset, demonstrating significant transferability. On a test set including experimental hydrogen-bonded dimers, AP-Net predicts total interaction energies with a mean absolute error of 0.37 kcal mol−1, reducing errors by a factor of 2–5 across SAPT components from previous neural network potentials. The pairwise interaction energies of the model are physically interpretable, and an investigation of predicted electrostatic energies suggests that the model “learns” the physics of hydrogen-bonded interactions.

Glick, Zachary L. (ORCID:0000000309002849)↗

Lower and upper threshold limit for artificial neural network based chilled and condenser water temperatures set-point control in a chilled water system

In this study, an ANN (artificial neural network) based real-time optimized control algorithm for a chilled water cooling system was developed and applied in an actual building to analyze its cooling energy saving effects through in-situ application and actual measurements. For this purpose, the cooling tower’s CndWT (condenser water temperature) and the chiller’s ChWT (chilled water temperature) were set as system control variables. To evaluate algorithm performance, the electric consumption and the COP (coefficient of performance) were compared and analyzed when ChWT and CndWTs were controlled conventionally and controlled based on the ANN. During the analysis, unexpected abnormal data was observed due to insufficient training data and limited consideration of OWBT (outdoor air wet-bulb temperature) when determining the CndWT set-point. Therefore, it is necessary to further build training data from a wider range of conditions and to set the lower limit of CndWT set-point to at least +3.6 °C above OWBT when the OWBT is higher than 23 °C, so that further energy savings can be achieved.

42 ENGINEERING↗

Learning model combining convolutional deep neural network with a self-attention mechanism for AC optimal power flow

Alternating current optimal power flow (OPF) analysis is critical for efficient and reliable operation of power systems. For large systems or repetitive computations, the traditional methods such as the direct and gradient methods, or non-traditional methods, such as the genetic algorithm and simulating annealing, are time-consuming and unsuitable for real-time computing. The work in this paper proposes a novel framework to obtain the optimal solution of power flow in real-time using a combination of convolutional neural networks and a self-attention mechanism. All parameters of the power networks are rearranged in an image-like shape of a multi-channel image where each channel is a two-dimensional matrix. The proposed approach is adaptive with every input size of power systems as well as frequent variations of network topologies without intervention to the framework core. The encompassment of all power system contexts in which all parameters of internal elements, generation costs, and topology information are included, contributes to the higher accuracy of inference compared to other current machine-learning-based OPF-solving methods. Besides, the proposed framework established on ubiquitous platforms is effortlessly integrated into current infrastructures of power systems, and the great efficiency along with the computation speed may serve as a critical point for practical implications, such as enabling faster decision-making during real-time operations, predicting system contingencies, and remedial actions based on an offline pre-trained model. Furthermore, this supervised learning process is applied to the dataset of four case studies of meshed power systems: the IEEE 5-bus system (IEEE-5), the IEEE 30-bus system (IEEE-30), the IEEE 39-bus system (IEEE-39), and the IEEE 57-bus system (IEEE-57) to prove the efficacy of the proposed method.

42 ENGINEERING↗

Dilated causal convolutional neural networks for forecasting zone airflow to estimate short-term energy consumption

Here this paper investigates the use of dilated causal convolutional neural networks for fine- grained temporal forecasting of building zone states. Specifically, we build and evaluate models using a small set of exogenous features (e.g., external temperature) to autoregressively predict zone airflow setpoints every minute for a 24-hour prediction window. We carefully explore the trade-off between generality and specificity in these models, training and evaluating them based on zone, zone type, month, season, and combinations thereof. When evaluated for a commercial office building in Eastern Washington with 16 zones served by variable air volume air handling units, we find that the highest performance comes from a zone-specific, season-agnostic approach; with it, we obtain an R 2 of 0.704 (averaged over zones) and an average normalized root mean square error (nRMSE) of 0.111. In contrast, the most general model (trained across all zones and seasons) yields an R 2 of only 0.416 and a nRMSE of 0.168, while a baseline zone-specific reduced order model obtains 0.443 R 2 and 0.159 nRMSE. We also report on factors affecting airflow forecasting performance, on the ability of models trained on a specific zone to generalize to other zones, and on the capability of those models trained on a specific month to generalize to other months.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Active learning of neural network potentials for rare events

Developing an automated active learning framework for Neural Network Potentials, focusing on accurately simulating bond-breaking in hexane chains through steered molecular dynamics sampling and assessing model transferability.

97 MATHEMATICS AND COMPUTING↗

Cartesian message passing neural networks for directional properties: Fast and transferable atomic multipoles

The message passing neural network (MPNN) framework is a promising tool for modeling atomic properties but is, until recently, incompatible with directional properties, such as Cartesian tensors. We propose a modified Cartesian MPNN (CMPNN) suitable for predicting atom-centered multipoles, an essential component of ab initio force fields. The efficacy of this model is demonstrated on a newly developed dataset consisting of 46 623 chemical structures and corresponding high-quality atomic multipoles, which was deposited into the publicly available Molecular Sciences Software Institute QCArchive server. We show that the CMPNN accurately predicts atom-centered charges, dipoles, and quadrupoles and that errors in the predicted atomic multipoles have a negligible effect on multipole–multipole electrostatic energies. The CMPNN is accurate enough to model conformational dependencies of a molecule’s electronic structure. This opens up the possibility of recomputing atomic multipoles on the fly throughout a simulation in which they might exhibit strong conformational dependence.

Glick, Zachary L. (ORCID:0000000309002849)↗

WPGNN and PLayGen (Wind Plant Graph Neural Network and Plant Layout Generator) [SWR-21-90]

WPGNN is the graph neural network machine learning based surrogate model and software that provides a streamlined approximation of wind plant wake models. It can rapidly estimate the energy production of the plant and turbines for any arbitrarily sized wind plant and layout under any inflow condition. Associated tools include graphing and visualization capabilities as well as a wind plant generator capable of creating randomized realizations of canonical wind plant layouts. The WPGNN architecture and application is extendable to multiple wake models, turbine technologies and features, and supports downstream optimizations of wind plant layouts and control strategies. In addition to the WPGNN, we include the code for the plant layout generator (PLayGen) playgen.py. This generator can produce random realizations of realistic wind plant layouts from one of the four canonical styles: cluster, single string, multiple string, or parallel string. The PLayGen_demo.ipynb notebook provides a demonstration of how to use the generator tool.

Harrison-Atlas, Dylan↗

User Manual - HydraGNN: Distributed PyTorch Implementation of Multi-Headed Graph Convolutional Neural Networks

This document serves as user manual for HydraGNN, a scalable graph neural network (GNN) architecture that allows for a simultaneous prediction of multiple target properties using multi-task learning (MTL). The HydraGNN architecture is constructed by successive superposition of three different sets of layers. The first set is made of message-passing layers to exchange information across nodes in the graph and use this to update the nodal features. The second set is made of global pooling layers that aggregate information from all the nodes in the graph and map it into a scalar, and is needed only for global target properties that are related to the entire graph. The third set of layers is dedicated to the implementation of MTL, which is enabled by forking of the architecture into separate heads, each one of them dedicated to the predictive task of one specific target property. Through an object-oriented programming paradigm, HydraGNN is templated over different message-passing policies, which allows for a user-friendly hyperparameter study to assess the sensitivity of the predictive performance of the HydraGNN architecture on a specific dataset with respect to the choice of the message-passing policy. The object-oriented paradigm used by HydraGNN also allows for a user-friendly inclusion of newly developed message passing policies within the existing framework. HydraGNN supports distributed computing capabilities for scalable data reading and scalable training on leadership-class supercomputers.

97 MATHEMATICS AND COMPUTING↗