Search NASA⌕ Search

SEARCH · Search NASA

Results for “Sensor anomaly detection”

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 55 records · Page 3

Event-driven simulation in SELMON: An overview of EDSE

EDSE (event-driven simulation engine), a model-based event-driven simulator implemented for SELMON, a tool for sensor selection and anomaly detection in real-time monitoring is described. The simulator is used in conjunction with a causal model to predict future behavior of the model from observed data. The behavior of the causal model is interpreted as equivalent to the behavior of the physical system being modeled. An overview of the functionality of the simulator and the model-based event-driven simulation paradigm on which it is based is provided. Included are high-level descriptions of the following key properties: event consumption and event creation, iterative simulation, synchronization and filtering of monitoring data from the physical system. Finally, how EDSE stands with respect to the relevant open issues of discrete-event and model-based simulation is discussed.

Rouquette, Nicolas F.↗

Towards Resilient Critical Infrastructures: Application of Type-2 Fuzzy Logic in Embedded Network Security Cyber Sensor

Resiliency and cyber security of modern critical infrastructures is becoming increasingly important with the growing number of threats in the cyber-environment. This paper proposes an extension to a previously developed fuzzy logic based anomaly detection network security cyber sensor via incorporating Type-2 Fuzzy Logic (T2 FL). In general, fuzzy logic provides a framework for system modeling in linguistic form capable of coping with imprecise and vague meanings of words. T2 FL is an extension of Type-1 FL which proved to be successful in modeling and minimizing the effects of various kinds of dynamic uncertainties. In this paper, T2 FL provides a basis for robust anomaly detection and cyber security state awareness. In addition, the proposed algorithm was specifically developed to comply with the constrained computational requirements of low-cost embedded network security cyber sensors. The performance of the system was evaluated on a set of network data recorded from an experimental cyber-security test-bed.

97 MATHEMATICS AND COMPUTING↗

Integrating event detection system operation characteristics into sensor placement optimization.

We consider the problem of placing sensors in a municipal water network when we can choose both the location of sensors and the sensitivity and specificity of the contamination warning system. Sensor stations in a municipal water distribution network continuously send sensor output information to a centralized computing facility, and event detection systems at the control center determine when to signal an anomaly worthy of response. Although most sensor placement research has assumed perfect anomaly detection, signal analysis software has parameters that control the tradeoff between false alarms and false negatives. We describe a nonlinear sensor placement formulation, which we heuristically optimize with a linear approximation that can be solved as a mixed-integer linear program. We report the results of initial experiments on a real network and discuss tradeoffs between early detection of contamination incidents, and control of false alarms.

54 ENVIRONMENTAL SCIENCES↗

Flight-Tested Prototype of BEAM Software

Researchers at JPL have completed a software prototype of BEAM (Beacon-based Exception Analysis for Multi-missions) and successfully tested its operation in flight onboard a NASA research aircraft. BEAM (see NASA Tech Briefs, Vol. 26, No. 9; and Vol. 27, No. 3) is an ISHM (Integrated Systems Health Management) technology that automatically analyzes sensor data and classifies system behavior as either nominal or anomalous, and further characterizes anomalies according to strength, duration, and affected signals. BEAM (see figure) can be used to monitor a wide variety of physical systems and sensor types in real time. In this series of tests, BEAM monitored the engines of a Dryden Flight Research Center F-18 aircraft, and performed onboard, unattended analysis of 26 engine sensors from engine startup to shutdown. The BEAM algorithm can detect anomalies based solely on the sensor data, which includes but is not limited to sensor failure, performance degradation, incorrect operation such as unplanned engine shutdown or flameout in this example, and major system faults. BEAM was tested on an F-18 simulator, static engine tests, and 25 individual flights totaling approximately 60 hours of flight time. During these tests, BEAM successfully identified planned anomalies (in-flight shutdowns of one engine) as well as minor unplanned anomalies (e.g., transient oil- and fuel-pressure drops), with no false alarms or suspected false-negative results for the period tested. BEAM also detected previously unknown behavior in the F- 18 compressor section during several flights. This result, confirmed by direct analysis of the raw data, serves as a significant test of BEAM's capability.

Mackey, Ryan↗

System and Method for Outlier Detection via Estimating Clusters

An efficient method and system for real-time or offline analysis of multivariate sensor data for use in anomaly detection, fault detection, and system health monitoring is provided. Models automatically derived from training data, typically nominal system data acquired from sensors in normally operating conditions or from detailed simulations, are used to identify unusual, out of family data samples (outliers) that indicate possible system failure or degradation. Outliers are determined through analyzing a degree of deviation of current system behavior from the models formed from the nominal system data. The deviation of current system behavior is presented as an easy to interpret numerical score along with a measure of the relative contribution of each system parameter to any off-nominal deviation. The techniques described herein may also be used to "clean" the training data.

Iverson, David J.↗

Detecting Process Equipment Failures Using Acoustic Data and Machine Learning

Nuclear power plant (NPP) process equipment such as fans, motors, valves, and pumps generate frequent or continuous noise, and deviations from the normal operational sounds made by this equipment can indicate potential issues. These deviations can be identified via automated acoustic anomaly detection, which involves using acoustic sensors (i.e., microphones) alongside detection algorithms to continuously monitor for changes in acoustic signatures. This task is made challenging by the substantial background noise that exists, such as operators opening and closing doors, manipulating valves, and conversing—in addition to typical plant noises. In collaboration with a nuclear power utility partner, this effort assessed the efficacy of acoustic anomaly detection when using a specific acoustic sensor that compresses data into a fixed set of features that are transferable over a standard Internet of Things communication protocol, thereby improving usability but potentially degrading detection performance. Two methods of performing automated acoustic anomaly detection were evaluated: one-class support vector machine (OC-SVM) and isolation forest (iForest). To enable the use of high-quality acoustic data encompassing both normal and anomalous conditions, the study utilized the publicly available Malfunctioning Industrial Machine Investigation and Inspection dataset, which includes real measured acoustic sensor data for a range of equipment types, model numbers, and signal-to-noise ratios (SNRs), along with a benchmark set of detection results. Using this dataset, the methods were tested and then compared against the benchmark results. The results indicated that although the specific acoustic sensor did not enable as rich a feature set extraction, the proposed methods with the limited feature set performed just as well. This provides solid justification for both the methods and the use of the proposed acoustic sensor.

46 - INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AN↗

MAD: Self-Supervised Masked Anomaly Detection Task for Multivariate Time Series

In this paper, we introduce Masked Anomaly Detection (MAD), a general self-supervised learning task for multivariate time series anomaly detection. With the increasing availability of sensor data from industrial systems, being able to detecting anomalies from streams of multivariate time series data is of significant importance. Given the scarcity of anomalies in real-world applications, the majority of literature has been focusing on modeling normality. The learned normal representations can empower anomaly detection as the model has learned to capture certain key underlying data regularities. A typical formulation is to learn a predictive model, i.e., use a window of time series data to predict future data values. In this paper, we propose an alternative self-supervised learning task. By randomly masking a portion of the inputs and training a model to estimate them using the remaining ones, MAD is an improvement over the traditional left-to-right next step prediction (NSP) task. Our experimental results demonstrate that MAD can achieve better anomaly detection rates over traditional NSP approaches when using exactly the same neural network (NN) base models, and can be modified to run as fast as NSP models during test time on the same hardware, thus making it an ideal upgrade for many existing NSP-based NN anomaly detection models.

97 MATHEMATICS AND COMPUTING↗

Monte Carlo Dropout Uncertainty Quantification of Long Short-Term Memory Autoencoder Anomaly Detection in a Liquid Sodium Cold Trap

Advanced high-temperature fluid reactors, such as sodium-cooled fast reactors (SFRs) and molten salt–cooled reactors (MSCRs), require coolant purification systems to prevent fluid contamination and local freezing that can lead to plugging. Liquid sodium purification can be achieved with a cold trap, where the sodium temperature is reduced to a near-freezing point to precipitate out impurities. Automation of monitoring of the cold trap performance with machine learning algorithms can aid in early detection of incipient anomalies. An efficient approach to loss-of-coolant–type anomaly detection in a cold trap monitored with more than two dozen thermal-hydraulic sensors consists of a long short-term memory (LSTM) autoencoder. This work develops the uncertainty quantification of the LSTM autoencoder performance for cold trap anomaly detection using the Monte Carlo (MC) dropout method. The MC dropout methodology creates a distribution of sister distributions that all slightly differ from each other because of random neurons being turned off for testing. The variances of the sister network distributions are used to make an uncertainty interval. Our analysis shows that the uncertainty in the autoencoder performance is largest near the peak of the anomaly signal. Using the MC dropout method, we investigate the uncertainty in the anomaly detection with missing sensor inputs. This capability allows the reactor operator to evaluate resilience of the anomaly detection system and to make informed decisions about continuity of operation in the event of sensor failure.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

On-Line Monitoring to Detect Sensor and Process Degradation under Normal Operational Transients - 327

Calibration assessment of nuclear plant sensors and transmitters is currently performed through periodic maintenance actions during plant refueling outages. On-line monitoring (OLM) can be used to reduce the time and cost of this maintenance process by using empirical models to evaluate sensor calibration. The implementation of OLM can also mitigate human error associated with unnecessary maintenance on sensors operating within calibration tolerances. OLM has been studied and validated for plants under steady-state baseload operation, but the extension to load-following operation is not well-documented in the literature. Auto-associative kernel regression (AAKR) provides an error-corrected prediction of process sensor measurements for OLM through a weighted average of historic data. The current study evaluates the performance of AAKR for prediction under normal operational transients (such as expected under load-following operation). Performance of the AAKR model with two anomaly detection routines, error uncertainty limit monitoring (EULM) and sequential probability ratio test (SPRT), was evaluated for normal operation, sensor faults, and process degradation. The data used to develop and evaluate the OLM system were collected from a forced flow loop with temperature and flow sensors. Three models of well-correlated groups of sensors are trained to predict normal system behavior under planned transients. The anomaly detection routines are tested for sensor faults and process faults. The results of this experimental study suggest AAKR and an appropriate anomaly detection routine can be applied for sensor calibration and process fault detection under transient operation. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Consolidated Hydropower Data Repository: Value and Opportunities

Hydropower is one of several types of generating assets that provides energy, capacity, and services to electric power systems. It does so under rubrics and objectives—market driven and regulated, internal and external to asset and fleet owners—that address reliability, cost, price, and, increasingly, flexibility of output. The aggregation of data from multiple hydropower units can provide insights into asset operations and maintenance practices and needs and assist in meeting hydropower objectives. This paper examines the concept and potential benefits of aggregating hydropower asset data—primarily supervisory control and data acquisition (SCADA) information—with examples of insights developed from data aggregated by the Hydropower Research Institute (HRI). Data aggregation as discussed herein, and as implemented by the HRI, extends beyond multiple units in a powerhouse and beyond multiple hydropower facilities in an electric utility fleet or river system. Examples of research and analytics from such aggregated datasets range from unit load dependency analyses to modeling sensor measurements to detect and diagnose anomalies in assets. These examples showed the benefits of utilizing the entire dataset for insights into how the sensor layout of a single unit or set of units compares to the hydropower industry overall. Such insights include whether additional sensors are needed to complete analyses or to make decisions. In addition, utilizing multiple sensors of the same kind within a unit can provide an indication of possible current or upcoming problems with equipment. Although other analyses are possible, their use requires the development of complex models and, potentially, access to types of data that are currently not included with the example dataset used in this study. However, the examples studied herein confirmed the value of data aggregation in the fleet and unit contexts, and the value extends beyond multiple units in a powerhouse and beyond multiple hydropower facilities in an electric utility fleet or river system. The assessments also provided insights into potential extensions to the data aggregation concept that could further add to their value to the hydropower community, and these are included in this document as a set of recommendations.

13 HYDRO ENERGY↗

Integrated System for Autonomous and Adaptive Caretaking (ISAAC): Phase 1 Low-Fidelity Demo

This presentation describes the ISAAC phase 1 low-fidelity demonstration results. The demonstration satisfied the milestone from the Gateway-ISAAC Memorandum of Understanding to "Demonstrate spatial and logical data registration between robotics and spacecraft". It integrated many new ISAAC components, including a spatially linked model, Astrobee multi-sensor mapping, and an integrated data interface. It advanced ISAAC key performance parameters related to mapping, in a lab setting. Areas for phase 1 forward work include: improve maturity toward the high-fidelity demo on the ISS; demonstrate mapping with more sensor modalities; expand initial anomaly detection implementation into a flexible framework with multiple detection algorithms for different tasks; begin open source software release process for releasable ISAAC components.

robotics↗

Sensor Fault Detection in Smart Extraterrestrial Habitats Using Unsupervised Learning

Various types of sensors are needed to monitor the health state of smart deep-space habitats. However, measured data can be affected by sensor faults, which influence the health management system and consequently the decision-making. In this paper, an unsupervised learning approach based on convolutional autoencoders (CAEs) is developed to detect anomalies in temperature and pressure sensors. The proposed method is systematically investigated using a habitat simulator (HabSim). Several illustrative examples are demonstrated in the nominal and hazardous states of the habitat, including micrometeorite impact and fire scenarios. The performance of the proposed method using CAEs is compared with that of existing methods using auto-associative neural networks (AANNs) and variational autoencoders. This comparison is based on typical evaluation metrics, including precision, recall, F1 score, training time, and testing time. The effect of temperature–pressure coupling on the detection performance of CAEs and AANNs is explored by training different data-driven models, including one with temperature sensors, one with pressure sensors, and one with both temperature and pressure sensors. The effect of the number of faulty sensors on the performance of CAEs is studied, as with an increase in the number of faulty sensors, redundant information among the sensors is reduced. The capability of CAEs to change the number of sensors without redesigning the network architecture and retraining the neural network is investigated and demonstrated. The capabilities and limitations of the proposed solution are discussed.

Zixin Wang↗

Classification based anomaly detection

An example device includes processing circuitry and a memory. The memory includes instructions that cause the device to perform various functions. The functions include receiving datastreams from a plurality of sensors of a high performance computing system, classifying each datastream of the each sensor to one of a plurality of datastream models, selecting an anomaly detection algorithm from a plurality of anomaly detection algorithms for each datastream, determining parameters of the each anomaly detection algorithm, determining an anomaly threshold for each datastream, and generating an indication that the sensor associated with the datastream is acting anomalously.

Serebryakov, Sergey↗

Crosscutting Airborne Remote Sensing Technologies for Oil and Gas and Earth Science Applications

Airborne imaging spectroscopy has evolved dramatically since the 1980s as a robust remote sensing technique used to generate 2-dimensional maps of surface properties over large spatial areas. Traditional applications for passive airborne imaging spectroscopy include interrogation of surface composition, such as mapping of vegetation diversity and surface geological composition. Two recent applications are particularly relevant to the needs of both the oil and gas as well as government sectors: quantification of surficial hydrocarbon thickness in aquatic environments and mapping atmospheric greenhouse gas components. These techniques provide valuable capabilities for petroleum seepage in addition to detection and quantification of fugitive emissions. New empirical data that provides insight into the source strength of anthropogenic methane will be reviewed, with particular emphasis on the evolving constraints enabled by new methane remote sensing techniques. Contemporary studies attribute high-strength point sources as significantly contributing to the national methane inventory and underscore the need for high performance remote sensing technologies that provide quantitative leak detection. Imaging sensors that map spatial distributions of methane anomalies provide effective techniques to detect, localize, and quantify fugitive leaks. Airborne remote sensing instruments provide the unique combination of high spatial resolution (<1 m) and large coverage required to directly attribute methane emissions to individual emission sources. This capability cannot currently be achieved using spaceborne sensors. In this study, results from recent NASA remote sensing field experiments focused on point-source leak detection, will be highlighted. This includes existing quantitative capabilities for oil and methane using state-of-the-art airborne remote sensing instruments. While these capabilities are of interest to NASA for assessment of environmental impact and global climate change, industry similarly seeks to detect and localize leaks of both oil and methane across operating fields. In some cases, higher sensitivities desired for upstream and downstream applications can only be provided by new airborne remote sensing instruments tailored specifically for a given application. There exists a unique opportunity for alignment of efforts between commercial and government sectors to advance the next generation of instruments to provide more sensitive leak detection capabilities, including those for quantitative source strength determination.

hyperspectral↗

Remote Structural Health Monitoring and Advanced Prognostics of Wind Turbines

The prospect of substantial investment in wind energy generation represents a significant capital investment strategy. In order to maximize the life-cycle of wind turbines, associated rotors, gears, and structural towers, a capability to detect and predict (prognostics) the onset of mechanical faults at a sufficiently early stage for maintenance actions to be planned would significantly reduce both maintenance and operational costs. Advancement towards this effort has been made through the development of anomaly detection, fault detection and fault diagnosis routines to identify selected fault modes of a wind turbine based on available sensor data preceding an unscheduled emergency shutdown. The anomaly detection approach employs spectral techniques to find an approximation of the data using a combination of attributes that capture the bulk of variability in the data. Fault detection and diagnosis (FDD) is performed using a neural network-based classifier trained from baseline and fault data recorded during known failure conditions. The approach has been evaluated for known baseline conditions and three selected failure modes: pitch rate failure, low oil pressure failure and a gearbox gear-tooth failure. Experimental results demonstrate the approach can distinguish between these failure modes and normal baseline behavior within a specified statistical accuracy.

17 WIND ENERGY↗

Automated System Checkout to Support Predictive Maintenance for the Reusable Launch Vehicle

The Propulsion Checkout and Control System (PCCS) is a predictive maintenance software system. The real-time checkout procedures and diagnostics are designed to detect components that need maintenance based on their condition, rather than using more conventional approaches such as scheduled or reliability centered maintenance. Predictive maintenance can reduce turn-around time and cost and increase safety as compared to conventional maintenance approaches. Real-time sensor validation, limit checking, statistical anomaly detection, and failure prediction based on simulation models are employed. Multi-signal models, useful for testability analysis during system design, are used during the operational phase to detect and isolate degraded or failed components. The TEAMS-RT real-time diagnostic engine was developed to utilize the multi-signal models by Qualtech Systems, Inc. Capability of predicting the maintenance condition was successfully demonstrated with a variety of data, from simulation to actual operation on the Integrated Propulsion Technology Demonstrator (IPTD) at Marshall Space Flight Center (MSFC). Playback of IPTD valve actuations for feature recognition updates identified an otherwise undetectable Main Propulsion System 12 inch prevalve degradation. The algorithms were loaded into the Propulsion Checkout and Control System for further development and are the first known application of predictive Integrated Vehicle Health Management to an operational cryogenic testbed. The software performed successfully in real-time, meeting the required performance goal of 1 second cycle time.

Patterson-Hine, Ann↗