SEARCH · Engineering Papers

Results for “training data quality”

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.

SourceResearch connectionFrom yearThrough year

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

At least 631 records · Page 35

All sources NASA NTRS DOE OSTI

NASA NTRS2022

Identifying Chemical Aerosol Signatures Using Optical Suborbital Observations: How Much Can Optical Properties Tell us about Aerosol Composition?

Improvements in air quality and Earth’s climate predictions require improvements of the aerosol speciation in chemical transport models, using observational constraints. Aerosol speciation (e.g., organic aerosols, black carbon, sulfate, nitrate, ammonium, dust or sea salt) is typically determined using in situ instrumentation. Continuous, routine surface network aerosol composition measurements are not uniformly widespread over the globe. Satellites, on the other hand, can provide a maximum coverage of the horizontal and vertical atmosphere but observe aerosol optical properties (and not aerosol speciation) based on remote sensing instrumentation. Combinations of satellite-derived aerosol optical properties can inform on air mass aerosol types (AMTs e.g., clean marine, dust, polluted continental). However, these AMTs are subjectively defined, might often be misclassified and are hard to relate to the critical parameters that need to be refined in models. In this paper, we derive AMTs that are more directly related to sources and hence to speciation. They are defined, characterized, and derived using simultaneous in situ gas-phase, chemical and optical instruments on the same aircraft during the Study of Emissions and Atmospheric Composition, Clouds, and Climate Coupling by Regional Surveys (SEAC4RS, US, summer of 2013). First, we prescribe well-informed AMTs that display distinct aerosol chemical and optical signatures to act as a training AMT dataset. These in situ observations reduce the errors and ambiguities in the selection of the AMT training dataset. We also investigate the relative skill of various combinations of aerosol optical properties to define AMTs and how much these optical properties can capture dominant aerosol speciation. We find distinct optical signatures for biomass burning (from agricultural or wildfires), biogenic and dust-influence AMTs. Useful aerosol optical properties to characterize these signatures are the extinction angstrom exponent (EAE), the single scattering albedo, the difference of single scattering albedo in two wavelengths, the absorption coefficient, the absorption angstrom exponent (AAE), and the real part of the refractive index (RRI). We find that all four AMTs studied when prescribed using mostly airborne in situ gas measurements, can be successfully extracted from at least three combinations of airborne in situ aerosol optical properties (e.g., EAE, AAE and RRI) over the US during SEAC4RS. However, we find that the optically based classifications for BB from agricultural fires and polluted dust include a large percentage of misclassifications that limit the usefulness of results relating to those classes. The technique and results presented in this study are suitable to develop a representative, robust and diverse source-based AMT database. This database could then be used for widespread retrievals of AMTs using existing and future remote sensing suborbital instruments/networks. Ultimately, it has the potential to provide a much broader observational aerosol data set to evaluate chemical transport and air quality models than is currently available by direct in situ measurements. This study illustrates how essential it is to explore existing airborne datasets to bridge chemical and optical signatures of different AMTs, before the implementation of future spaceborne missions (e.g., the next generation of Earth Observing System (EOS) satellites addressing Aerosol, Cloud, Convection and Precipitation (ACCP) designated observables).

Meloe S F Kacenelenbogen↗

NASA NTRS2023

Use of Machine Learning and Principal Component Analysis to Retrieve Nitrogen Dioxide (NO 2 ) With Hyperspectral Imagers and Reduce Noise in Spectral Fitting

Nitrogen dioxide (NO 2 ) is an important trace-gas pollutant and climate agent whose presence also leads to spectral interference in ocean color retrievals. NO 2 column densities have been retrieved with satellite UV–Vis spectrometers such as the Ozone Monitoring Instrument (OMI) and the Tropospheric Monitoring Instrument (TROPOMI) that typically have spectral resolutions of the order of 0.5 nm or better and spatial footprints as small as 3.6 km × 5.6 km. These NO 2 observations are used to estimate emissions, monitor pollution trends, and study effects on human health. Here, we investigate whether it is possible to retrieve NO 2 amounts with lower-spectral-resolution hyperspectral imagers such as the Ocean Color Instrument (OCI) that will fly on the Plankton, Aerosol, Cloud, ocean Ecosystem (PACE) satellite set for launch in early 2024. OCI will have a spectral resolution of 5 nm and a spatial resolution of ∼ 1 km with global coverage in 1–2 d. At this spectral resolution, small-scale spectral structure from NO 2 absorption is still present. We use real spectra from the OMI to simulate OCI spectra that are in turn used to estimate NO 2 slant column densities (SCDs) with an artificial neural network (NN) trained on target OMI retrievals. While we obtain good results with no noise added to the OCI simulated spectra, we find that the expected instrumental noise substantially degrades the OCI NO 2 retrievals. Nevertheless, the NO 2 information from OCI may be of value for ocean color retrievals. OCI retrievals can also be temporally averaged over timescales of the order of months to reduce noise and provide higher-spatial-resolution maps that may be useful for downscaling lower-spatial-resolution data provided by instruments such as OMI and TROPOMI; this downscaling could potentially enable higher-resolution emissions estimates and be useful for other applications. In addition, we show that NNs that use coefficients of leading modes of a principal component analysis of radiance spectra as inputs appear to enable noise reduction in NO 2 retrievals. Once trained, NNs can also substantially speed up NO 2 spectral fitting algorithms as applied to OMI, TROPOMI, and similar instruments that are flying or will soon fly in geostationary orbit.

NO2↗

DOE OSTI2022

Deep learning-based noise filtering toward millisecond order imaging by using scanning transmission electron microscopy

Application of scanning transmission electron microscopy (STEM) to in situ observation will be essential in the current and emerging data-driven materials science by taking STEM’s high affinity with various analytical options into account. As is well known, STEM’s image acquisition time needs to be further shortened to capture a targeted phenomenon in real-time as STEM’s current temporal resolution is far below the conventional TEM’s. However, rapid image acquisition in the millisecond per frame or faster generally causes image distortion, poor electron signals, and unidirectional blurring, which are obstacles for realizing video-rate STEM observation. Here we show an image correction framework integrating deep learning (DL)-based denoising and image distortion correction schemes optimized for STEM rapid image acquisition. By comparing a series of distortion corrected rapid scan images with corresponding regular scan speed images, the trained DL network is shown to remove not only the statistical noise but also the unidirectional blurring. This result demonstrates that rapid as well as high-quality image acquisition by STEM without hardware modification can be established by the DL. The DL-based noise filter could be applied to in-situ observation, such as dislocation activities under external stimuli, with high spatio-temporal resolution.

47 OTHER INSTRUMENTATION↗

NASA NTRS2011

Tools and Technologies Needed for Conducting Planetary Field Geology While On EVA: Insights from the 2010 Desert RATS Geologist Crewmembers

Observation is the primary role of all field geologists, and geologic observations put into an evolving conceptual context will be the most important data stream that will be relayed to Earth during a planetary exploration mission. Sample collection is also an important planetary field activity, and its success is closely tied to the quality of contextual observations. To test protocols for doing effective planetary geologic fieldwork, the Desert RATS (Research and Technology Studies) project deployed two prototype rovers for two weeks of simulated exploratory traverses in the San Francisco volcanic field of northern Arizona. The authors of this paper represent the geologist crewmembers who participated in the 2010 field test. We document the procedures adopted for Desert RATS 2010 and report on our experiences regarding these protocols. Careful consideration must be made of various issues that impact the interplay between field geologic observations and sample collection, including time management; strategies related to duplication of samples and observations; logistical constraints on the volume and mass of samples and the volume/transfer of data collected; and paradigms for evaluation of mission success. We find that the 2010 field protocols brought to light important aspects of each of these issues, and we recommend best practices and modifications to training and operational protocols to address them. Underlying our recommendations is the recognition that the capacity of the crew to "flexibly execute" their activities is paramount. Careful design of mission parameters, especially field geologic protocols, is critical for enabling the crews to successfully meet their science objectives.

Young, Kelsey↗

DOE OSTI2025

Open Power System Datasets and Open Simulation Engines: A Survey Toward Machine Learning Applications

A major factor behind the success of machine learning (ML) models in multiple domains is the availability and accessibility of large, labeled, and well-organized datasets for training and benchmarking. In comparison, power grid datasets face three major challenges: (i) real-world data is often restricted by regulatory constraints, privacy reasons, or security concerns, making it difficult to obtain and work with; (ii) synthetic datasets, which are created to address these limitations, often have incomplete information and are released using specialized tools, making them inaccessible to the broader community; and, (iii) input-output datasets are difficult to generate through simulation for non-experts because open-source simulators are not known outside the power system community. This survey addresses these challenges by serving as an entry point to publicly available datasets and simulators for researchers venturing in this area. We review the current landscape of open-source power network data, machine models, consumer demand profiles, renewable generation data, and inverter models. We also examine open-source power system simulators, which are crucial for generating high-quality, high-fidelity power grid datasets. We aim to provide a foundation for overcoming data scarcity and advance towards a structured web of datasets and simulators to support the development of ML for power systems.

42 ENGINEERING↗

NASA NTRS2014

Simulation Technology at NASA

NASA Ames Research Center is home to several high-fidelity research flight and air-traffic control simulation facilities which, together with an experienced workforce, produce high-quality research data and findings that have proven to be applicable in the real world. These assets include the Vertical Motion Simulator (VMS), Crew Vehicle Systems Research Facility (CVSRF), Future Flight Central (FFC) air traffic control tower simulator, and several air-traffic control (ATC) simulators. The VMS combines a high-fidelity simulation capability with an adaptable simulation environment, enabling customization for numerous human-in-the-loop research applications. The distinctive feature of the VMS is its unparalleled large amplitude, high-fidelity motion capability. In over 30 years of continuous operation, the VMS has contributed significantly to the body of knowledge in a range of disciplines directly benefiting several aerospace programs and flight safety, including the design and development of flight control systems for the Joint Strike Fighter, Space Shuttle Orbiter, and rotorcraft. It continues to be used for researching new vehicle configurations, vehicle control and safety, transfer-of-training, etc., by NASA, other government agencies, and Industry.The CVSRF includes two motion-based flight simulators: a Boeing 747-400 full flight simulator and the reconfigurable Advanced Concepts Flight Simulator (ACFS). These simulators are primarily used to research air-traffic management concepts and procedures, advanced navigation and avionics concepts, and cockpit human factors. FFC is a full-sized control tower simulator with a 360-degree external field-of-view display system and reconfigurable system architecture. FFC and the ATC simulators are used for testing air-traffic management automation and decision support tools and demonstrate their feasibility in a realistic environment prior to technology transfer for implementation in the National Airspace System (NAS).To support integrated simulations and flight-tests for NASA's Unmanned Aircraft Systems (UAS) in the National Airspace System (NAS) Project, NASA developed a distributed test environment incorporating Live, Virtual, Constructive, (LVC) concepts. Development of the software enabling the LVC is conducted primarily at the Distributed Simulation Research Lab (DSRL) at NASA Ames. The LVC components provide the core infrastructure supporting simulation of UAS operations by integrating live and virtual aircraft in a realistic air traffic environment. This provides the ability to conduct tests more efficiently by promoting the use of existing distributed assets. The LVC infrastructure was used in several human-in-the-loop simulations to evaluate acceptance of Detect and Avoid (DAA) advisories used by UAS pilots to maintain well clear of other virtual traffic and to negotiate maneuvers with air traffic control. It is currently being used to support testing of self-separation algorithms between unmanned and manned aircraft in live flight. Further simulations with more comprehensive air traffic scenarios mixing live and virtual aircraft is planned.In the current fiscal environment, maintaining and upgrading these high-fidelity simulation assets and retaining the skilled workforce necessary to meet future research needs is the primary non-technical challenge.

CVSRF↗

DOE OSTI2025

Using a Large Language Model as a Building Block to Generate Usable Validation and Verification Suite for OpenMP

In the HPC area, both hardware and software move quickly. Often new hardware is developed and deployed, the corresponding software stack, including compilers and other tools, are under active development while leading edge software developers are working to port and tune their applications, all at the same time. While the software ecosystem is in flux, one of the key challenges for users is obtaining insight into the state of implementation of key features in the programming languages and models their applications are using – whether they have been implemented, and whether the implementation conforms to the specification, especially for newly implemented features (less tested by widespread use). OpenMP is one of the most prominent shared memory programming models used for on-node programming in HPC. With the shift towards accelerators (such as GPUs and FPGAs) and heterogeneous programming OpenMP features are getting more complex. It is natural to ask whether generative AI approaches, and large language models (LLMs) in particular, can help in producing validation and verification test suites to allow users better and faster insights into the availability and correctness of OpenMP features of interest. In this work, we explore the use of ChatGPT-4 to generate a suite of tests for OpenMP features. We have chosen a set of directives and clauses, a total of 78 combinations, which first appeared in OpenMP 3.0 (released in May 2008) but are also relevant for accelerators. We prompted ChatGPT to generate tests in the C and Fortran languages, for both host (CPU) and device (accelerator). On the Summit super-computer using the GNU implementation, we found that, of the 78 generated tests 67 C tests and 43 Fortran tests compiled successfully and fewer than those executed to completion. On further analysis we show that not all generated tests are valid. We document the process, results, and provide detailed analysis regarding the quality of tests generated. With the aim of providing input to a production quality validation and verification suite, we manually implement the corrections required to make the tests valid according to the current OpenMP specification. We quantify this effort as small, medium, or large, and record the lines of code changed to correct the invalid tests. With the corrected tests we validate recent implementations from HPE, AMD, and GNU on the Frontier supercomputer. Our experiment and subsequent analysis show that although LLMs are capable of producing HPC specific codes, they are limited by their understanding of the deeper semantics and restrictions of programming models such as OpenMP. Unsurprisingly more commonly used features have better support, while some OpenMP 3.0 directives such as sections and tasking are not universally supported on accelerators. We demonstrate that successful compilation and execution to completion are inadequate metrics for evaluating generated code and that, at this time, commodity LLMs require expert intervention for code verification. This points to gaps in the training data that is currently available for HPC. We demonstrate that with "small" effort 37% of generated invalid C tests and 63% of generated invalid Fortran tests could be corrected. This improves productivity of test generation as we circumvent writing from scratch and the common programming errors associated with it.

Pophale, Swaroop [ORNL] (ORCID:0000000185446367)↗

DOE OSTI2024

Large-Scale Visualization of 3D Unstructured Groundwater Model Using Cave Automated Virtual Environment

The immersive three-dimensional (3D) virtual reality (VR) visualization of groundwater models allows us to deepen our understanding of aquifer systems and provide better solutions to present groundwater-related problems, such as groundwater recharge, water quality, and sustainability. Visualization assists in accurately developing groundwater models and revealing important subsurface features, including faulting, folding, and unconformity. However, assessing model accuracy poses challenges due to the complexity of geology and groundwater systems. This research demonstrates a workflow to visualize and analyze raw 3D unstructured groundwater model data using an immersive Cave Automated Virtual Environment (CAVE). To visualize the unstructured groundwater model data, the raw dataset is converted into interactive CAVE-compatible formats utilizing a set of tools: ParaView, Blender, and Unity. This enables researchers to immerse themselves in the data, identifying influential patterns and relationships. e resulting insights can inform the development of sophisticated machine-learning models for groundwater level prediction. The CAVE’s immersive capabilities allow intuitive exploration from various perspectives, providing a more holistic understanding of the factors affecting groundwater levels. These insights are crucial to improve predictive models. The CAVE results also facilitate collaborative analysis and have potential applications in training and education. is research demonstrates the value of immersive VR tools such as the CAVE for unraveling intricacies within high-dimensional scientific data to drive real-world forecasting and modeling applications.

54 ENVIRONMENTAL SCIENCES↗

NASA NTRS1999

System Identification for Integrated Aircraft Development and Flight Testing [l'Identification Des Systemes Pour le Developpement Integre des Aeronefs et les Essais en Vol]

Over the last decades flight vehicles such as aircraft and helicopters entering service and requiring increased operational effectiveness have with few exceptions experienced prolonged flight test development to achieve full certification. In many cases the original requirements had later to be reduced to enable release to service. The impact on the customer, and manufacturer has been considerable leading to increased costs and or reduced operational capabilities. These costly experiences are largely a result of the flight vehicle not behaving as modelled and designed. The evaluation of flight test data can be used as a tool for validating windtunnel results and mathematical models describing the flight dynamical behaviour. In this sense the uncertainty of important aerodynamic stability and control parameters can be reduced and the confidence of aircraft mathematical models improved. An additional important factor comes from the implementation of active control systems offering the promise of significantly increased flight vehicle performance and operational capability. This approach extends the traditional trade-offs between aerodynamics, structures and propulsion systems to include full- time, full-authority fly-by-wire/light systems. It is imperative that the aerodynamic stability and control parameters of such integrated flight and propulsion control systems have to turn out inflight as predicted, since inherent stability margins will be lower and the flight control system must correct these deficiencies to provide flight critical redundancy and safety. With the methodology of system identification from flight tests it is possible to sense the control inputs and the flight vehicle reactions Such as accelerations, rates and attitudes. The mathematical model, e.g. the model structure and parameters, has to be determined from the relationship of the measured control inputs and the system's responses. The aim of this symposium was to review the present state of the art of flight vehicle system and parameter identification techniques, and to provide a critical appraisal of current methods developed and applied to flight test data in a number of NATO nations. Particular emphasis was placed on practical aspects and lessons learned in order to generate information useful to the flight test community in industry and government agencies. The technical papers share invaluable experience and emphasize the advances of flight vehicle system identification over the last years to the point where confidence and robustness level is now reasonably high. The symposium covered overviews of identification methodologies, flight test techniques, recent aircraft and helicopter application programs, and a session of short papers covering up-to-the-minute flight test results. A final discussion included prepared comments from experts and concluded with key issues learned in the application of system identification and future research needs. The essential benefits to NATO nations can be condensed as follows: More accurate mathematical models for high bandwidth flight control systems, Improved assessment and evaluation of flying qualities, High fidelity mathematical models for flight vehicle development and mission training simulators, and generally, Reduced flight test time and costs.

Advisory Group for Aerospace Research and Developm↗

NASA NTRS2017

GES DISC Greenhouse Gas Data Sets and Associated Services

NASA Goddard Earth Sciences (GES) Data and Information Services Center (DISC) archives and distributes rich collections of data on atmospheric greenhouse gases from multiple missions. Hosted data include those from the Atmospheric Infrared Sounder (AIRS) mission (which has observed CO2, CH4, ozone, and water vapor since 2002); legacy water vapor and ozone retrievals from TIROS Operational Vertical Sounder (TOVS); and Upper Atmosphere Research Satellite (UARS) going back to the early 1980s. GES DISC also archives and supports data from seven projects of the Making Earth System Data Records for Use in Research Environments (MEaSUREs) program that have ozone and water vapor records. Greenhouse gases data from the A-Train satellite constellation is also available: (1) Aura-Ozone Monitoring Instrument (OMI) and Microwave Limb Sounder (MLS) ozone, nitrous oxide, and water vapor since 2004; (2) Greenhouse Gases Observing Satellite (GOSAT) CO2 observations since 2009 from the Atmospheric CO2 Observations from Space (ACOS) task; and (3) Orbiting Carbon Observatory-2 (OCO-2) CO2 data since 2014. The most recent related data set that the GES DISC archives is methane flux for North America, as part of NASAs Carbon Monitoring System (CMS) project. This dataset contains estimates of methane emission in North America based on an inversion of the GEOS-Chem chemical transport model constrained by GOSAT observations (Turner et al., 2015). Along with data stewardship, an important focus area of the GES DISC is to enhance the usability of its data and broaden its user base. Users have unrestricted access to a new user-friendly search interface, which includes many services such as variable subsetting, format conversion, quality screening, and quick browse. The majority of the GES DISC data sets are also accessible through Open-source Project for a Network Data Access Protocol (OPeNDAP) and Web Coverage Service (WCS). The latter two services provide more options for specialized subsetting, format conversion, and image viewing. Additional data exploration, data preview, and preliminary analysis capabilities are available via NASA Giovanni, which obviates the need forusers to download the data (Acker and Leptoukh, 2007). Giovanni provides a bridge between the data and science and has been very successful in extending GES DISC data to educational users and to users with limited resources.

data ordering↗

NASA NTRS2017

Probabilistic Forecasting of Surface Ozone with a Novel Statistical Approach

The recent change in the Environmental Protection Agency's surface ozone regulation, lowering the surface ozone daily maximum 8-h average (MDA8) exceedance threshold from 75 to 70 ppbv, poses significant challenges to U.S. air quality (AQ) forecasters responsible for ozone MDA8 forecasts. The forecasters, supplied by only a few AQ model products, end up relying heavily on self-developed tools. To help U.S. AQ forecasters, this study explores a surface ozone MDA8 forecasting tool that is based solely on statistical methods and standard meteorological variables from the numerical weather prediction (NWP) models. The model combines the self-organizing map (SOM), which is a clustering technique, with a step wise weighted quadratic regression using meteorological variables as predictors for ozone MDA8. The SOM method identifies different weather regimes, to distinguish between various modes of ozone variability, and groups them according to similarity. In this way, when a regression is developed for a specific regime, data from the other regimes are also used, with weights that are based on their similarity to this specific regime. This approach, regression in SOM (REGiS), yields a distinct model for each regime taking into account both the training cases for that regime and other similar training cases. To produce probabilistic MDA8 ozone forecasts, REGiS weighs and combines all of the developed regression models on the basis of the weather patterns predicted by an NWP model. REGiS is evaluated over the San Joaquin Valley in California and the northeastern plains of Colorado. The results suggest that the model performs best when trained and adjusted separately for an individual AQ station and its corresponding meteorological site.

Ozone variables↗

DOE OSTI2023

Neural network-based single material beam-hardening correction for X-ray CT in Additive Manufacturing

Beam-hardening (BH) artifacts are ubiquitous in X-ray CT scans of dense metal additively manufactured (AM) parts. While linearization approaches are useful for correcting beam-hardened data from single material objects, they either require a calibration scan or detailed system and material composition information. In this paper, we introduce a neural network-based, material-agnostic method to correct beam-hardening artifacts. We train a neural network to map the acquired beam-hardened projection values and the corresponding estimated thickness of the part based on an initial segmentation to beam-hardening related parameters, which can be used to compute the coefficients of a linearizing correction polynomial. A key strength of our approach is that, once the network is trained, it can be used for correcting beam hardening from a variety of materials without any calibration scans or detailed system and material composition information. Furthermore, our method is robust to errors in the estimated thickness due to the typical challenge of obtaining an accurate initial segmentation from reconstructions impacted by BH artifacts. We demonstrate the utility of our method to obtain high-quality CT reconstructions from a collection of AM parts -- suppressing cupping and streaking artifacts

Rahman, Obaid↗

NASA NTRS2012

Rapid Analysis, Self-Calibrating Array for Air Monitoring

Human space missions have critical needs for monitoring and control for life support systems. These systems have monitoring needs that include feedback for closed loop processes and quality control for environmental factors. Sensors and monitoring technologies assure that the air environment and water supply for the astronaut crew habitat fall within acceptable limits, and that the life support system is functioning properly and efficiently. The longer the flight duration and the more distant the destination, the more critical it becomes to have carefully monitored and automated control systems for life support. Past experiments with the JPL ENose have demonstrated a lifetime of the sensor array, with the software, of around 18 months. The lifetime of the calibration, for some analytes, was as long as 24 months. We are working on a sensor array and new algorithms that will include sensor response time in the analysis. The preliminary array analysis for two analytes shows that the analysis time, of an event, can be dropped from 45 minutes to less than10 minutes and array training time can be cut substantially. We will describe the lifetime testing of an array and show lifetime data on individual sensors. This progress will lead to more rapid identification of analytes, and faster training time of the array.

Self Calibrating↗

DOE OSTI2025

Active Learning Meets Foundation Models: Fast Remote Sensing Data Annotation for Object Detection

Object detection in remote sensing demands extensive, high-quality annotations—a process that is both labor-intensive and time-consuming. In this work, we introduce a real-time active learning and semi-automated labeling framework that leverages foundation models to streamline dataset annotation for object detection in remote sensing imagery. For example, by integrating a Segment Anything Model (SAM), our approach generates mask-based bounding boxes that serve as the basis for dual sampling: (a) uncertainty estimation to pinpoint challenging samples, and (b) diversity assessment to ensure broad data coverage. Furthermore, our Dynamic Box Switching Module (DBS) addresses the well-known cold start problem for object detection models by replacing its suboptimal initial predictions with SAM-derived masks, thereby enhancing early-stage localization accuracy. Extensive evaluations on multiple remote sensing datasets plus a real-world user study, demonstrate that our framework not only reduces annotation effort, but also significantly boosts detection performance compared to traditional active learning sampling methods. The code for training and the user interface will be made available.

Burges, Marvin [ORNL] (ORCID:0000000312690769)↗

DOE OSTI2023

Explainable AI (XAI)-driven vibration sensing scheme for surface quality monitoring in a smart surface grinding process

Local Interpretable and Model-agnostic Explanation (LIME), an explainable artificial intelligence (XAI) approach is adapted to identify the globally important time-frequency bands for predicting average surface roughness (Ra) in a smart grinding process. The smart grinding setup consisted of a Supertech CNC precision surface grinding machine, instrumented with a Dytran piezoelectric accelerometer attached to the tailstock quill along the tangential direction (Y-axis). For every grinding pass, vibration signatures were captured, and the ground truth surface roughness values were recorded using a Mahr Marsurf M300C portable surface roughness profilometer. The roughness values ranged from 0.06 to 0.14 microns over the complete set of experiments. Time-frequency domain spectrogram frames were extracted for each of the vibration signals collected during the grinding process. Convolutional Neural Networks (CNNs) were modeled to predict the surface roughness based on these spectrogram frames and their image augmentations. The best CNN model was able to predict the roughness values with an overall R2-score of 0.95, training R2-score of 0.99, and testing R2-score of 0.81 with only 80 sets of vibration signals corresponding to 4 experiments with 20 trials each. Although the data size is not large enough to guarantee such performance metrics in real-world scenarios, one can extract statistically consistent explanations underlying the relationships these complex deep learning models capture. Further, the LIME methodology was implemented on the developed surface roughness CNN model to identify the important time-frequency bands (i.e., the superpixels of a spectrogram) influencing the predictions. Based on the identified important regions on the spectrogram frames, the corresponding frequency characteristics were determined that influence the surface roughness predictions. The important frequency range based on LIME results was approximately 11.7 to 19.1 kHz. The power of XAI was demonstrated by cutting down the sampling rate from 160 kHz to 30, 20, 10, and 5 kHz based on the important frequency range and considering Nyquist criteria. Separate CNN models were developed for these ranges by only extracting time-frequency contents below their corresponding Nyquist cut-offs. A proper data acquisition strategy is proposed by comparing the model performances to argue the selection of a sufficient sampling rate to capture the grinding process successfully and robustly.

42 ENGINEERING↗

DOE OSTI2023

Reprint of: Explainable AI (XAI)-driven vibration sensing scheme for surface quality monitoring in a smart surface grinding process

47 OTHER INSTRUMENTATION↗

DOE OSTI2020

Overcoming small minirhizotron datasets using transfer learning

Minirhizotron technology is widely used to study root growth and development. Yet, standard approaches for tracing roots in minirhiztron imagery is extremely tedious and time consuming. Machine learning approaches can help to automate this task. However, lack of enough annotated training data is a major limitation for the application of machine learning methods. Transfer learning is a useful technique to help with training when available datasets are limited. In this paper, we investigated the effect of pre-trained features from the massives-cale, irrelevant ImageNet dataset and a relatively moderate-scale, but relevant peanut root dataset on switchgrass root imagery segmentation applications. We compiled two minirhizotron image datasets to accomplish this study: one with 17,550 peanut root images and another with 28 switchgrass root images. Both datasets were paired with manually labeled ground truth masks. Deep neural networks based on the U-net architecture were used with different pre-trained features as initialization for automated, precise pixel-wise root segmentation in minirhizotron imagery. We observed that features pre-trained on a closely related but relatively moderate size dataset like our peanut dataset were more effective than features pre-trained on the large but unrelated ImageNet dataset. Here, we achieved high quality segmentation on peanut root dataset with 99.04% accuracy at the pixel-level and overcame errors in human-labeled ground truth masks. By applying transfer learning technique on limited switchgrass dataset with features pre-trained on peanut dataset, we obtained 99% segmentation accuracy in switchgrass imagery using only 21 images for training (fine tuning). Furthermore, the peanut pre-trained features can help the model converge faster and have much more stable performance.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2025

Anomaly Detection in Seismic Data with Deep Learning: Application for Instrument Failure Detection and Forecasting

Seismic data quality assessment (QA) is the first and one of the most important steps before conducting any further data analysis. Traditional methods involve checking various metrics, such as spike detection and power spectral density, by setting strict thresholds or comparing data against synthetic benchmarks. However, these approaches often rely on pre-existing knowledge and assumptions about data anomalies, leading to potential misclassification of unusual cases. Here, in this study, we propose a deep autoencoder model, an unsupervised learning approach that evaluates data quality without making assumptions about normal and anomalous data, which can be used to identify deviations in recorded data that may indicate nascent instrument failure. We test the model with the U.S. International Monitoring System (IMS) seismic stations and demonstrate the capability of detecting anomalies on a monthly scale. This could prompt station operators to examine potential problems early, allowing sufficient time for instrument maintenance to prevent data outages. In addition, we use a new manually selected testing dataset to compare our model performance against two supervised machine learning (ML) approaches and a standard QA package, as baseline models. When applied to the dataset containing known data anomalies, performance of the supervised and unsupervised ML approaches is similar, with an accuracy of 88.1% for our model compared to ∼90% for the supervised ML approach and 78.2% for the standard QA package. Our model outperforms the baseline models when applied to new stations, where new types of data anomalies can be station-specific and not included in the training dataset. Finally, we show model transferability by training the model with data from the Global Seismograph Network only and applying it to the IMS network data. The results suggest that our model is generalizable and can be applied to new stations with good accuracy.

Lin, Jiun-Ting [Lawrence Livermore National Labora↗