Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “model driven architecture”

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 163 records · Page 9
All sourcesNASA NTRSDOE OSTI
NASA NTRS2017

Data Production on Past and Future NASA Missions

Data return is a metric that is commonly publicized for all space science missions. In the early days of the Space Program, this figure was small, and could be described in bits or maybe even megabits. But now, missions are capable of returning data volumes two or three orders of magnitude larger. For example, Voyager 1 and 2 combined produced a little over 5 Terabits of data in 39 years of operation. In contrast, the Cassini mission, launched two decades after Voyager, produced about one and a half times those data volumes in half the time. NISAR, an Earth Science Mission currently in implementation, plans to produce over 28 Petabits of raw data in just 3 years. This means that NISAR will produce about as many data in 30 days as the combined data production of nearly all planetary missions to date. These increases in capability are a result of technology enhancements in two main areas: telecommunications architecture (both space and ground segments) and data storage technology. This paper describes the progression of these two technologies over the course of more than three decades of space missions and provides additional insight into the design of the end-to-end NISAR Data System Architecture. Trends in the data are briefly explored and compared to Moore’s Law which provides only a qualitative model for memory growth but not for data production. In summary, early missions are found to be driven by unrefined processes while later missions, having utilized earlier lessons learned, focus more on improvements to flight and ground capabilities. Data return seems to fall into three categories. First, deep space missions are driven by the large distances that limit data return to the Earth. Next, the orbiter infrastructure around Mars helps these missions generate more data than other deep space spacecraft. Finally, near-Earth missions have the greatest capabilities for the studied metrics due to their close proximity to Earth and the ground network availability.

Xaypraseuth, Peter↗
DOE OSTI2021

Domain knowledge-informed, process-mapping AI graph for designing Fe-based alloys

<span style="font-family: Calibri, sans-serif; font-size: 12pt;">Continuous improvement in efficiency of a power plant relies on designing materials for use at increasingly higher temperature and/or pressure, for 100,000s hours of operation. Due to complexity, non-linearity and high-dimensionality of the problem, traditional Machine Learning (ML) approaches require unreasonably large datasets for the data-driven model development. Science-based material and process engineering complements hard data with, sometimes soft and intuitive, empirical domain knowledge. Artificial Intelligence (AI) was used in this study to incorporate such knowledge into computational graph architecture (process-mimicking artificial neuron design, causal layer and graph structures, ensemble modeling of latent states) and learning procedures (variable transformation, fuzzy physics pre-training and freezing of deep layers, virtual microstructure representation, and adversarial multi-objective optimization). The first alloys design pathways suggested by the AI tool (pyroMind) passed a preliminary engineering review on soundness and transparency.</span>

Romanov, Vyacheslav↗
DOE OSTI2026

From natural language to control signals: a conceptual framework for semantic channel finding in complex experimental infrastructure

Modern experimental platforms such as particle accelerators, fusion devices, telescopes, and industrial process control systems expose tens to hundreds of thousands of control and diagnostic channels, accumulated over decades of hardware evolution. Operators and AI systems alike depend on informal expert knowledge, inconsistent naming conventions, and scattered documentation to locate the signals required for monitoring, troubleshooting, and automated control, creating a persistent bottleneck for reliability, scalability, and emerging language-model-driven interfaces. We formalize semantic channel finding, the task of mapping natural-language intent to concrete control-system signals, as a general problem in complex experimental infrastructure, and introduce a four-paradigm conceptual framework to guide architecture selection based on facility-specific data regimes. The paradigms span (i) direct in-context lookup over small, curated channel dictionaries, (ii) constrained hierarchical navigation through structured trees, (iii) interactive agent exploration using iterative reasoning and tool-based database queries, and (iv) ontology-grounded semantic search that decouples channel meaning from facility-specific naming conventions. We demonstrate the practical feasibility of each paradigm through proof-of-concept implementations at four operational facilities spanning two orders of magnitude in scale: from compact free-electron lasers to large synchrotron light sources, operating under diverse control-system architectures ranging from clean hierarchical naming schemes to legacy environments with decades of heterogeneous conventions. Where evaluated against expert-curated operational queries, these instantiations achieve 90%–97% accuracy, validating the framework’s applicability across real-world deployment scenarios. To accelerate adoption across the broader scientific and industrial control-system community, we release open-source, plug-and-play implementations of all three interactive paradigms-direct lookup, hierarchical navigation, and middle-layer exploration-within the Osprey framework, together with tools for channel database generation, interactive testing, and minimal-configuration deployment. This work establishes semantic channel finding as a foundational capability for human-centric and agentic AI interfaces at large-scale facilities, providing both a systematic framework for architecture design and practical resources to enable adoption without building custom infrastructure from scratch.

channel finding↗
DOE OSTI2015

Load‐balanced and locality‐aware scheduling for data‐intensive workloads at extreme scales

Summary Data‐driven programming models such as many‐task computing (MTC) have been prevalent for running data‐intensive scientific applications. MTC applies over‐decomposition to enable distributed scheduling. To achieve extreme scalability, MTC proposes a fully distributed task scheduling architecture that employs as many schedulers as the compute nodes to make scheduling decisions. Achieving distributed load balancing and best exploiting data locality are two important goals for the best performance of distributed scheduling of data‐intensive applications. Our previous research proposed a data‐aware work‐stealing technique to optimize both load balancing and data locality by using both dedicated and shared task ready queues in each scheduler. Tasks were organized in queues based on the input data size and location. Distributed key‐value store was applied to manage task metadata. We implemented the technique in MATRIX, a distributed MTC task execution framework. In this work, we devise an analytical suboptimal upper bound of the proposed technique, compare MATRIX with other scheduling systems, and explore the scalability of the technique at extreme scales. Results show that the technique is not only scalable but can achieve performance within 15% of the suboptimal solution. Copyright © 2015 John Wiley & Sons, Ltd.

Wang, Ke↗
DOE OSTI2026

Ion Transport in Charged Membranes: Linking Electric-Field-Driven Mechanisms to Pore Size via Perturbation Analysis

Ion-exchange membranes are a critical component in electrochemical systems. Nevertheless, the understanding and modeling of ion transport within these porous structures have been limited by particular complexity reductions, either ignoring the dimensionality of their porous network architecture or imposing geometric assumptions (i.e., overlapping double layers). Before addressing this morphology-transport gap, a framework that relates the driving forces of transport to the geometry of a single pore is required. In this work, our modeling domain consists of a two-dimensional single pore with charged walls, connecting two identical electrolyte reservoirs. Using the Poisson-Nernst-Planck equations and regular perturbation theory, we decouple the electric fields and analyze the driving forces of ion transport, specifically electromigration and induced electroosmosis within the pore. These processes are described as analytical functions of the interaction aspect ratio, ?, defined as the ratio of the pore radius to the Debye length. Using this parameter, our study (i) describes the interplay between electromigrative and electroosmotic mechanisms that set ionic conductivity, (ii) identifies a dimensionless group of intrinsic electrolyte properties that indicates the predominant driving force, and (iii) provides a qualitative, confinement-dependent perspective on selectivity in ion-conducting membranes.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

Synergy of semiempirical models and machine learning in computational chemistry

Catalyzed by enormous success in the industrial sector, many research programs have been exploring data-driven, machine learning approaches. Performance can be poor when the model is extrapolated to new regions of chemical space, e.g., new bonding types, new many-body interactions. Another important limitation is the spatial locality assumption in model architecture, and this limitation cannot be overcome with larger or more diverse datasets. The outlined challenges are primarily associated with the lack of electronic structure information in surrogate models such as interatomic potentials. Given the fast development of machine learning and computational chemistry methods, we expect some limitations of surrogate models to be addressed in the near future; nevertheless spatial locality assumption will likely remain a limiting factor for their transferability. Here, we suggest focusing on an equally important effort—design of physics-informed models that leverage the domain knowledge and employ machine learning only as a corrective tool. In the context of material science, we will focus on semi-empirical quantum mechanics, using machine learning to predict corrections to the reduced-order Hamiltonian model parameters. The resulting models are broadly applicable, retain the speed of semiempirical chemistry, and frequently achieve accuracy on par with much more expensive ab initio calculations. These early results indicate that future work, in which machine learning and quantum chemistry methods are developed jointly, may provide the best of all worlds for chemistry applications that demand both high accuracy and high numerical efficiency.

36 MATERIALS SCIENCE↗
DOE OSTI2023

Proxy Applications for Converged Workloads: DMC LDRD Initiative

Modern scientific applications are complicated and require coordination of several components. Proxy application driven software-hardware co-design plays a vital role in driving innovation among the developments of applications, software infrastructure and hardware architecture. Proxy applications are self-contained and simplified codes that are intended to model the performance-critical computations within applications. Applications executing on modern High Performance Computing (HPC) systems are susceptible to network congestion, insufficient memory bandwidth within and across compute nodes, and inadvertent loss of performance due to bugs and unoptimized programming models. Modern numerical simulations and machine learning models play a critical role in studying physical phenomenon under myriad uncertainties. Such applications often exhibit irregular computation and memory accesses at specific regions of the application code, which can contribute to various performance bottlenecks at scale. To mitigate such issues and prepare the next generation hardware for a variety of computation and data movement contingencies, a well-known practice is to consider "proxy" applications as representative motifs for various classes of scientific applications. While there is disagreement in the HPC community on the mechanisms of construction of the proxy applications, there is a strong consensus on their positive impact in co-design. Proxy Applications for Converged Workloads (PACER) is about facilitating software-hardware co-design through proxy applications with the goal of improving the performance of converged science workflows on heterogeneous systems.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2020

Architecture and Information Requirements to Assess and Predict Flight Safety Risks During Highly Autonomous Urban Flight Operations

As aviation adopts new and increasingly complex operational paradigms, vehicle types, and technologies to broaden airspace capability and efficiency, maintaining a safe system will require recognition and timely mitigation of new safety issues as they emerge and before significant consequences occur. A shift toward a more predictive risk mitigation capability becomes critical to meet this challenge. In-time safety assurance comprises monitoring, assessment, and mitigation functions that proactively reduce risk in complex operational environments where the interplay of hazards may not be known (and therefore not accounted for) during design. These functions can also help to understand and predict emergent effects caused by the increased use of automation or autonomous functions that may exhibit unexpected non-deterministic behaviors. The envisioned monitoring and assessment functions can look for precursors, anomalies, and trends (PATs) by applying model-based and data-driven methods. Outputs would then drive downstream mitigation(s) if needed to reduce risk. These mitigations may be accomplished using traditional design revision processes or via operational (and sometimes automated) mechanisms. The latter refers to the ‘in-time’ aspect of the system concept. This report comprises architecture and information requirements and considerations toward enabling such a capability within the domain of low altitude highly autonomous urban flight operations. This domain may span, for example, public-use surveillance missions flown by small unmanned aircraft (e.g., infrastructure inspection, facility management, emergency response, law enforcement, and/or security) to transportation missions flown by larger aircraft that may carry passengers or deliver products. Caveat: Any stated requirements in this report should be considered initial requirements that are intended to drive research and development (R&D). These initial requirements are likely to evolve based on R&D findings, refinement of operational concepts, industry advances, and new industry or regulatory policies or standards related to safety assurance.

Young, Steven↗
DOE OSTI2021

Data assimilation empowered neural network parametrizations for subgrid processes in geophysical flows

In the past couple of years, there has been a proliferation in the use of machine learning approaches to represent subgrid-scale processes in geophysical flows with an aim to improve the forecasting capability and to accelerate numerical simulations of these flows. Despite its success for different types of flow, the online deployment of a data-driven closure model can cause instabilities and biases in modeling the overall effect of subgrid-scale processes, which in turn leads to inaccurate prediction. To tackle this issue, we exploit the data assimilation technique to correct the physics-based model coupled with the neural network as a surrogate for unresolved flow dynamics in multiscale systems. In particular, we use a set of neural network architectures to learn the correlation between resolved flow variables and the parametrizations of unresolved flow dynamics and formulate a data assimilation approach to correct the hybrid model during their online deployment. We illustrate our framework in a set of applications of the multiscale Lorenz 96 system for which the parametrization model for unresolved scales is exactly known, and the two-dimensional Kraichnan turbulence system for which the parametrization model for unresolved scales is not known a priori. Our analysis, therefore, comprises a predictive dynamical core empowered by (i) a data-driven closure model for subgrid-scale processes, (ii) a data assimilation approach for forecast error correction, and (iii) both data-driven closure and data assimilation procedures. We show significant improvement in the long-term prediction of the underlying chaotic dynamics with our framework compared to using only neural network parametrizations for future prediction. Moreover, we demonstrate that these data-driven parametrization models can handle the non-Gaussian statistics of subgrid-scale processes, and effectively improve the accuracy of outer data assimilation workflow loops in a modular nonintrusive way.

42 ENGINEERING↗
DOE OSTI2022

Data-Driven PMU Noise Emulation Framework using Gradient-Penalty-Based Wasserstein GAN

Availability of phasor measurement unit (PMUs) data has led to research on data-driven algorithms for event monitoring, control and ensuring stability of the grid. Unavailability of infrequent critical event field PMU data with component failures is driving the need to generate realistic synthetic PMU data for research. The synthetic data from power system simulation software often neglect noise profiles of received phasors, thus creating some discrepancies between real PMU data and synthetic ones. To address this issue, this work presents an initial study on the noise characteristics of PMUs, as well as presenting models for recreating their unique noise signatures. The proposed method, utilizing the Wasserstein generative adversarial network with gradient penalty (WGAN-GP) architecture, provides an excellent benchmark for matching the noise distribution. One can use a well-learned GAN model to draw noise signatures from a distribution that seemingly mirrors the real PMU noise distribution, while also being able to be detached from the PMU data once the training is done. Based on the observed results and employed data-driven methodology, it is expected that the proposed methods can be adapted to replicate the behavior of other sensors, providing research and other applications with a tool for data synthesis and sensor characterization.

PMU↗
DOE OSTI2025

Enzyme property prediction using artificial intelligence

Artificial intelligence (AI)-driven enzyme property prediction enables rapid discovery and engineering of enzymes for a wide range of biotechnological and therapeutic applications. Here, we first introduce the key components in AI model development, including enzyme datasets, protein representation methods, and model architectures. We then highlight a variety of AI tools developed for the prediction of enzyme properties and functional annotations, including enzyme structure, kinetic parameters, substrate specificity, thermostability, solubility, Enzyme Commission number, and Gene Ontology term. Moreover, we describe representative downstream applications enabled by these AI tools. Finally, we discuss some challenges and opportunities as well as future prospects.

Yuan, Le [University of Illinois at Urbana-Champai↗
DOE OSTI2018

Final Technical Report for: A Unified Data-Driven Approach for Programming In Situ Analysis and Visualization

The placement and movement of data is becoming the key limiting factor on both performance and energy efficiency of high performance computations. As systems generate more data, it is becoming increasingly difficult to actually move that data elsewhere for post-processing, as the rate of improvements in supporting I/O infrastructure is not keeping pace. Together, these trends are creating a shift in how we think about exascale computations, from a viewpoint that focuses on FLOPS to one that focuses on data and data-centric operations as fundamental to the reasoning about, and optimization of, scientific workflows on extreme-scale architectures. The overarching goal of our effort was the study of a unified data-driven approach for programming applications and in situ analysis and visualization. Our work was to understand the interplay between data-centric programming model requirements at extreme-scale and the overall impact of those requirements on the design, capabilities, flexibility, and implementation details for both applications and the supporting in situ infrastructure. In this context, we made many improvements to the Legion programming system (one of the leading data-centric models today) and demonstrated in situ analyses on real application codes using these improvements.

97 MATHEMATICS AND COMPUTING↗
NASA NTRSDate not supplied

Mars Entry, Descent, Landing, and Ascent Systems Sensitivities to Landing Site and Atmospheric Dust

Plans for human missions to Mars continue to go through several architectural changes, dating all the way back to the 1950s [1]. The continuous study, reformulation, and refinement of Mars architectures and system concepts is necessary in order to incorporate evolving mission objectives, technology advancements, and growth in the body of knowledge regarding human factors and the various environments of human space travel. This evolution has continued into the 21st century, with architectures concepts such as NASA’s Design Reference Architecture 5.0 in 2009 [2], The Evolvable Mars Campaign in 2016 [3], and as recently as 2020, an architecture focused on a crewed mission as early as the 2030s that aims to minimizing crewed duration and infrastructure investment for the first mission [4]. Within these architectures, numerous studies around the current concept designs for Mars entry, descent, landing, and ascent (EDLA) systems have been performed over the last half a decade [[5], [6], [7], [8], [9], [10]]. Despite the breadth and depth of these studies, landing site, a key design parameter relevant to the design of EDLA systems, has remained nebulous over the years, largely due to the ever evolving mission objectives and architecture concept over the decades. However, the specific landing site has direct impact on the altitude and atmospheric conditions, which subsequently impact the design of EDLA systems. To accommodate the lack of a specific landing site, a baseline reference altitude of 0 km relative to the Mars Orbiter Laser Altimeter (MOLA), which is similar to an Earth sea level reference, has typically been selected and fixed for these past studies. Similarly, a baseline reference atmosphere has typically been utilized in these studies, either the Mars Global Reference Atmospheric Model [11], or a general 1982 standard warm, high pressure atmosphere model derived from Viking lander data [12]. Fig. 1 shows the range of elevations across the surface of Mars. Current architectures are focused on latitudes greater than 30 degrees north in an effort to ensure access to frozen water ice. From the figure, it is quickly evident that elevations over the range of longitudes at or above this latitude are significantly varied between roughly -4 to +4 km MOLA. Based on these observations, initial qualitative assessments of the impacts of landing site elevation on EDLA systems were performed. Fig. 2 depicts the expected trends in EDLA system mass for variations in both land site latitude and elevation. The background coloring is a qualitative evaluation of the current knowledge on ice water availability at the latitudes. Further modeling and simulation was performed to obtain numerical predictions of sensitivities to these parameters. Results indicate up to +2% to -4% mass variation from the current baseline Mars Ascent Vehicle concept, with the potential for greater than 6% mass variation at latitudes greater than 70 degrees north. However, variations in the descent system due to landing site were much more significant with mass variations in the range of -15% to +30% around the current baseline Mars Descent System concept. Additionally, Fig. 3 provides a depiction of potential Mars atmospheric density variations with both dust and Martian season. Though the Martian atmosphere is relatively thin compared to Earth’s atmosphere, it still poses significant impact on the design of EDLA systems. Variations in atmospheric density indicated by this slice of data would have significant impact on the design of EDLA systems. However, due to the highly complex nature of atmospheric flight, a qualitative assessment could not be performed. Rather, sensitivity results relied on modeling and simulation to provide numerical results for sensitivities on the current EDLA design concepts under consideration. Understanding these sensitivities is vital to the overall systems design of a Mars architecture. The resulting mass impacts on EDLA systems, driven by landing site elevation and dust level variations, has rippling impacts throughout the architecture that, ultimately, impact the viability of the architecture. Results showed fairly minor mass impacts to the current Mars Ascent Vehicle baseline configuration, roughly -0.6% to + 0.4% mass variations, while the Mars Descent System say greater variations due to dust, roughly -4% to +6% around the baseline concept.

EDLA↗
DOE OSTI2021

Modeling of advanced accelerator concepts

Computer modeling is essential to research on Advanced Accelerator Concepts (AAC), as well as to their design and operation. This paper summarizes the current status and future needs of AAC systems and reports on several key aspects of (i) high-performance computing (including performance, portability, scalability, advanced algorithms, scalable I/Os and In-Situ analysis), (ii) the benefits of ecosystems with integrated workflows based on standardized input and output and with integrated frameworks developed as a community, and (iii) sustainability and reliability (including code robustness and usability).

47 OTHER INSTRUMENTATION↗
DOE OSTI2026

PowerMappeR: Power-Optimized Mapping of SNNs onto ReRAM Crossbars coupled via Packet-Switched NoCs

Many recent efforts in developing hardware-accelerated spiking neural networks (SNNs) are characterized by deep co-design between algorithms, architectures, and devices. Architectural advances overcome device constraints by coupling together many small resistive-RAM (ReRAM) crossbars via a network-on-chip (NoC) for neuromorphic component operation. Concurrently, improved SNN training methods increase accuracy and structural sparsity in networks despite growing problem sizes. Finally, compilers leverage these attributes to minimize area and inter-crossbar communication while mapping large SNNs to sophisticated architectures. However, for compiler-driven co-design to realize increasingly complex and profitable optimizations, a compile-time view of power consumption is critical. We present PowerMappeR to express and optimize over mapping-, architecture-, and device-specific power consumption information. By modeling the dynamic power of well-established components, we develop an integer linear programming (ILP)-based, encoding-agnostic, parametric power estimation model. Using this model, we demonstrate practical improvements in area and inter-crossbar communication by 0%–9.5% and 1.4%–5.1%, respectively. We also limit hotspot formation during optimization, achieving comparable or better results in targeted metrics with up to 96.4%–97.1% restriction of hotspot magnitude. Finally, we introduce profile-guided formulations to reduce worst-case and expected-case hotspot magnitude by 40.7%–69.5% and 40.6%–56.3%, respectively. Optimizing worst-case hotspot magnitude incidentally improves expected-case magnitude by 10.85%–33.45%. Reciprocally, optimizing expected-case magnitude incidentally improves worst-case magnitude by 4.33%–39.87%. Validation against hardware simulators confirms that PowerMappeR can decrease dynamic power consumption by 12.6%–27.3%.

Pohl, Devin [ORNL] (ORCID:0009000040149027)↗
DOE OSTI2021

Characterization of Extremes and Compound Impacts: Applications of Machine Learning and Interpretable Neural Networks

Focal Area: This white paper responds to Focal area III by exploring data fusion, learning and explainable AI methods in characterizing hydrological extremes and interconnections. It also addresses Focal area II by using probabilistic AI and ensemble ML for predicting extremes and compound extremes. Science Challenge: A key question associated with the integrated water (or hydrological) cycle grand challenge in the Earth and Environmental Systems Sciences Division (EESSD) strategic plan, is how the frequency and intensity of hydrological events will change. Prediction of the tail behavior (extremes) of the hydrological cycle is especially challenging, because of their stochasticity and low probability. These extreme events and their compound impacts have significant societal and economic consequences. It is anticipated for the next-generation Earth System models (ESMs), that model predictability of the water cycle will improve with increased resolution (e.g., regionally refined E3SM), advanced software and computational architectures, and improved model physics based on the data from ARM measurements and high-fidelity models. However, the challenges for predictability of low-probability high-impact extreme events will unlikely be alleviated with conventional modeling and data-driven approaches, as ESMs are calibrated largely for capturing the high-frequency mean climate states. Recent AI and ML applications have shown great potential in quantifying well-defined climate extremes (e.g., supervised learning of tropical cyclones/atmospheric rivers by ClimateNet1) but few efforts are dedicated to compound events, extreme drivers and uncertainty estimation. We envision the opportunity to develop and apply ML and interpretable AI methods extended on the existing efforts, specifically, for: (1) identification of compound extremes, (2) diagnosing drivers of extremes, (3) bias correction in extreme predictions and (4) probabilistic modeling of extremes.

54 ENVIRONMENTAL SCIENCES↗
NASA NTRS2014

Space Launch System Ascent Flight Control Design

A robust and flexible autopilot architecture for NASA's Space Launch System (SLS) family of launch vehicles is presented. The SLS configurations represent a potentially significant increase in complexity and performance capability when compared with other manned launch vehicles. It was recognized early in the program that a new, generalized autopilot design should be formulated to fulfill the needs of this new space launch architecture. The present design concept is intended to leverage existing NASA and industry launch vehicle design experience and maintain the extensibility and modularity necessary to accommodate multiple vehicle configurations while relying on proven and flight-tested control design principles for large boost vehicles. The SLS flight control architecture combines a digital three-axis autopilot with traditional bending filters to support robust active or passive stabilization of the vehicle's bending and sloshing dynamics using optimally blended measurements from multiple rate gyros on the vehicle structure. The algorithm also relies on a pseudo-optimal control allocation scheme to maximize the performance capability of multiple vectored engines while accommodating throttling and engine failure contingencies in real time with negligible impact to stability characteristics. The architecture supports active in-flight disturbance compensation through the use of nonlinear observers driven by acceleration measurements. Envelope expansion and robustness enhancement is obtained through the use of a multiplicative forward gain modulation law based upon a simple model reference adaptive control scheme.

Orr, Jeb S.↗
DOE OSTI2019

An artificial neural network framework for reduced order modeling of transient flows

Here, this paper proposes a supervised machine learning framework for the non-intrusive model order reduction of unsteady fluid flows to provide accurate predictions of non-stationary state variables when the control parameter values vary. Our approach utilizes a training process from full-order scale direct numerical simulation data projected on proper orthogonal decomposition (POD) modes to achieve an artificial neural network (ANN) model with reduced memory requirements. This data-driven ANN framework allows for a nonlinear time evolution of the modal coefficients without performing a Galerkin projection. Our POD-ANN framework can thus be considered an equation-free approach for latent space dynamics evolution of nonlinear transient systems and can be applied to a wide range of physical and engineering applications. Within this framework we introduce two architectures, namely sequential network (SN) and residual network (RN), to train the trajectory of modal coefficients. We perform a systematic analysis of the performance of the proposed reduced order modeling approaches on prediction of a nonlinear wave-propagation problem governed by the viscous Burgers equation, a simplified prototype setting for transient flows. We find that the POD-ANN-RN yields stable and accurate results for test problems assessed both within inside and outside of the database range and performs significantly better than the standard intrusive Galerkin projection model. Finally, our results show that the proposed framework provides a non-intrusive alternative to the evolution of transient physics in a POD basis spanned space, and can be used as a robust predictive model order reduction tool for nonlinear dynamical systems.

42 ENGINEERING↗