Find research datasets worth reusing

ARC Code TI: Multiple Kernel Anomaly Detection (MKAD) Algorithm

The Multiple Kernel Anomaly Detection (MKAD) algorithm is designed for anomaly detection over a set of files.

Recurring Anomaly Detection System (ReADS)

Overview: ReADS can analyze text reports, such as aviation reports and problem or maintenance records. ReADS uses text clustering algorithms to group loosely related reports and documents, this reduces human error and fatigue. Plus, ReADS identifies interconnected reports; automating the discovery of possible recurring anomalies. ReADS provides a visualization of the clusters and recurring anomalies. ReADS has been integrated into a secure web-based search tool to allow uses to perform their own text mining. Recurring Anomaly Identification ReADS identifies reports which mention other reports as a recurring anomaly using regular expressions to search documents and identify references of other reports by name. ReADS also detects recurring anomalies by determining the similarity between documents using a cosine distance similarity measure. Then according to the similarity measure, ReADS will run a hierarchical clustering algorithm to detect the recurring anomalies. The hierarchical tree is partitioned into clusters by setting a threshold. A low threshold implies that the reports must be very similar to be sorted into the same cluster. Here's more [info](http://ti.arc.nasa.gov/projects/pria/). The figure below is a screenshot of the clustering results.

Detecting Anomalies in Multivariate Data Sets with Switching Sequences and Continuous Streams

The world-wide aviation system is one of the most complex dynamical systems ever developed and is generating data at an extremely rapid rate. Most modern commercial aircraft record several hundred flight parameters including information from the guidance, navigation, and control systems, the avionics and propulsion systems, and the pilot inputs into the aircraft. These parameters may be continuous measurements or binary or categorical measurements recorded in one second intervals for the duration of the flight. Currently, most approaches to aviation safety are reactive, meaning that they are designed to react to an aviation safety incident or accident. Here, we discuss a novel approach based on the theory of multiple kernel learning to detect potential safety anomalies in very large data bases of discrete and continuous data from world-wide operations of commercial fleets. We pose a general anomaly detection problem which includes both discrete and continuous data streams, where we assume that the discrete streams have a causal influence on the continuous streams. We also assume that atypical sequence of events in the discrete streams can lead to off-nominal system performance. We discuss the application domain, novel algorithms, and also briefly discuss results on synthetic and real-world data sets. Our algorithm uncovers operationally significant events in high dimensional data streams in the aviation industry which are not detectable using state of the art methods.

Space Shuttle Main Propulsion System Anomaly Detection: A Case Study

The space shuttle main engine (SSME) is part of the Main Propnlsion System (MPS) which is an extremely complex system containing several sub-systems and components, each of which must work precisely in order to achieve a successful mission. A critical component under study is the flow control valve (FCV) which controls the pressure of the gaseous hydrogen between the SSME and the external fuel tank. The FCV has received added attention since a Space Shuttle Mission in November 2008, where it was discovered during the mission that an anomaly had occurred in one of the three FCV's. Subsequent inspection revealed that one FCV cracked during ascent. This type of fault is of high criticality because it can lead to potentially catastrophic gaseous hydrogen leakage. A supervised learning method known as Virtual Sensors (VS), and an unsupervised learning method known as the Inductive Monitoring System (IMS) were used to detect anomalies related to the FCV in the MPS. Both algorithms identify the time of the anomaly in a multi-dimensional time series of temperatures, pressures, and control signals related to the FCV. This discovery corroborates the results of the inspection and also reveals the time at which the anomaly likely occurred. The methods were applied to data obtained from the March 2009 launch of Space Shuttle Discovery to determine whether an anomaly occurred in the same sub-system. According to our models, the FCV SUb-system showed nominal behavior during ascent.

On-line intermittent connector anomaly detection

This paper investigates a non-traditional use of differential current sensor and current sensor to detect intermittent disconnection problems in connectors. An intermittent disconnect, often resulting in an arc, creates an imbalance which is manifested in the current. The traveling wave generated due to the perturbation can be detected using current sensors. This paper shows the feasibility to detect disconnection based on this principle.

Sensing and characterization of EMI during intermittent connector anomalies

This paper presents a new on-line methodology for detecting intermittent disconnection failures. The detection principle operates on the fundamental Lorentz Law that states that sudden changes in flux create a large voltage, resulting in an arc. This arc propagates as a traveling wave through the circuit until all the energy associated with it is dissipated. It is possible to detect that traveling wave as an indication of an intermittent disconnection failure. A test bench was implemented to test and validate the theory. Preliminary results presented in this paper show the feasibility of detecting disconnection failures and the possibility of locating the failing connector based on distance.

Comparative Analysis of Data-Driven Anomaly Detection Methods

This paper provides a review of three different advanced machine learning algorithms for anomaly detection in continuous data streams from a ground-test firing of a subscale Solid Rocket Motor (SRM). This study compares Orca, one-class support vector machines, and the Inductive Monitoring System (IMS) for anomaly detection on the data streams. We measure the performance of the algorithm with respect to the detection horizon for situations where fault information is available. These algorithms have been also studied by the present authors (and other co-authors) as applied to liquid propulsion systems. The trade space will be explored between these algorithms for both types of propulsion systems.

VIIRS/NPP Thermal Anomalies/Fire 6-Min L2 Swath 750m V001

The Visible Infrared Imaging Radiometer Suite (VIIRS) Thermal Anomalies (VNP14) Version 1 product is produced in 6-minute temporal satellite increments (swaths) at 750 meter resolution from the VIIRS sensor located on the NASA/NOAA Suomi National Polar-orbiting Partnership (Suomi NPP) satellite. This product is designed after the Terra and Aqua Moderate Resolution Imaging Spectroradiometer (MODIS) Thermal Anomalies and Fire data products to promote the continuity of the Earth Observation System (EOS) mission. This data product can enable users to understand the location and intensity of fire events as well as identifying thermal anomalies. The VNP14 product includes 31 science dataset layers to analyze key factors in fire detection, including atmospheric conditions (e.g. atmospheric reflectance, solar zenith angle, brightness temperature) and fuel type for the event. The fire mask layer in the VNP14 product is the primary layer and can be used to identify fires and other thermal anomalies such as volcanoes. In addition to the fire mask, brightness temperature is provided for VIIRS channels M5, M7, M11, M13, M15, and M16. Each swath of data is approximately 3,060 kilometers along track (long) and 3,060 kilometers across track (wide). The VNP14 product is also used to generate higher-level fire data products. Use of the (VNP03MODLL) (https://doi.org/10.5067/viirs/vnp03modll.001) data product is required to apply accurate geolocation information to the VNP14 Science Datasets (SDS).

VIIRS/NPP Thermal Anomalies and Fire Daily L3 Global 1km SIN Grid V001

The daily NASA/NOAA Suomi National Polar-orbiting Partnership (Suomi NPP) Visible Infrared Imaging Radiometer Suite (VIIRS) Thermal Anomalies/Fire (VNP14A1) Version 1 data product provides daily information about active fires and other thermal anomalies. The VNP14A1 data product is a global, 1 kilometer (km) gridded composite of fire pixels detected from VIIRS 750 meter (m) bands over a daily (24-hour) period. The VNP14 data products are designed after the Moderate Resolution Imaging Spectroradiometer (MODIS) Thermal Anomalies/Fire product suite. The VNP14A1 product provides a total of four Science Dataset (SDS) layers for the confidence of fire, maximum fire radiative power (FRP), quality assessment (QA), and position of fire within scan. Each data product file is provided in HDF-EOS5 format. A low resolution browse is also provided showing the fire mask layer with a color map applied in JPEG format.

Comparison of Unsupervised Anomaly Detection Methods

Several different unsupervised anomaly detection algorithms have been applied to Space Shuttle Main Engine (SSME) data to serve the purpose of developing a comprehensive suite of Integrated Systems Health Management (ISHM) tools. As the theoretical bases for these methods vary considerably, it is reasonable to conjecture that the resulting anomalies detected by them may differ quite significantly as well. As such, it would be useful to apply a common metric with which to compare the results. However, for such a quantitative analysis to be statistically significant, a sufficient number of examples of both nominally categorized and anomalous data must be available. Due to the lack of sufficient examples of anomalous data, use of any statistics that rely upon a statistically significant sample of anomalous data is infeasible. Therefore, the main focus of this paper will be to compare actual examples of anomalies detected by the algorithms via the sensors in which they appear, as well the times at which they appear. We find that there is enough overlap in detection of the anomalies among all of the different algorithms tested in order for them to corroborate the severity of these anomalies. In certain cases, the severity of these anomalies is supported by their categorization as failures by experts, with realistic physical explanations. For those anomalies that can not be corroborated by at least one other method, this overlap says less about the severity of the anomaly, and more about their technical nuances, which will also be discussed.

Anomaly Detection in a Fleet of Systems

A fleet is a group of systems (e.g., cars, aircraft) that are designed and manufactured the same way and are intended to be used the same way. For example, a fleet of delivery trucks may consist of one hundred instances of a particular model of truck, each of which is intended for the same type of service—almost the same amount of time and distance driven every day, approximately the same total weight carried, etc. For this reason, one may imagine that data mining for fleet monitoring may merely involve collecting operating data from the multiple systems in the fleet and developing some sort of model, such as a model of normal operation that can be used for anomaly detection. However, one then may realize that each member of the fleet will be unique in some ways—there will be minor variations in manufacturing, quality of parts, and usage. For this reason, the typical machine learning and statis- tics algorithm’s assumption that all the data are independent and identically distributed is not correct. One may realize that data from each system in the fleet must be treated as unique so that one can notice significant changes in the operation of that system.

Discovery of Recurring Anomalies in Text Reports

This paper describes the results of a significant research and development effort conducted at NASA Ames Research Center to develop new text mining algorithms to discover anomalies in free-text reports regarding system health and safety of two aerospace systems. We discuss two problems of significant import in the aviation industry. The first problem is that of automatic anomaly discovery concerning an aerospace system through the analysis of tens of thousands of free-text problem reports that are written about the system. The second problem that we address is that of automatic discovery of recurring anomalies, i.e., anomalies that may be described in different ways by different authors, at varying times and under varying conditions, but that are truly about the same part of the system. The intent of recurring anomaly identification is to determine project or system weakness or high-risk issues. The discovery of recurring anomalies is a key goal in building safe, reliable, and cost-effective aerospace systems.

Multiple Kernel Learning for Heterogeneous Anomaly Detection: Algorithm and Aviation Safety Case Study

The world-wide aviation system is one of the most complex dynamical systems ever developed and is generating data at an extremely rapid rate. Most modern commercial aircraft record several hundred flight parameters including information from the guidance, navigation, and control systems, the avionics and propulsion systems, and the pilot inputs into the aircraft. These parameters may be continuous measurements or binary or categorical measurements recorded in one second intervals for the duration of the flight. Currently, most approaches to aviation safety are reactive, meaning that they are designed to react to an aviation safety incident or accident. In this paper, we discuss a novel approach based on the theory of multiple kernel learning to detect potential safety anomalies in very large data bases of discrete and continuous data from world-wide operations of commercial fleets. We pose a general anomaly detection problem which includes both discrete and continuous data streams, where we assume that the discrete streams have a causal influence on the continuous streams. We also assume that atypical sequences of events in the discrete streams can lead to off-nominal system performance. We discuss the application domain, novel algorithms, and also discuss results on real-world data sets. Our algorithm uncovers operationally significant events in high dimensional data streams in the aviation industry which are not detectable using state of the art methods.

Anomaly Detection from ASRS Databases of Textual Reports

Our primary goal is to automatically analyze textual reports from the Aviation Safety Reporting System (ASRS) database to detect/discover the anomaly categories reported by the pilots, and to assign each report to the appropriate category/categories. We have used two state-of-the-art models for text analysis: (i) mixture of von Mises Fisher (movMF) distributions, and (ii) latent Dirichlet allocation (LDA) on a subset of all ASRS reports. The models achieve a reasonably high performance in discovering anomaly categories and clustering reports. Each category is represented by the most representative words with the highest probability in this category. In addition, since the inference algorithm for LDA was somewhat slow, we have developed a new fast LDA algorithm which is 5-10 times more efficient than the original one, therefore more applicable for the practical use. Further, we have developed a simple visualization tool based on non-linear manifold embedding (ISOMAP) to generate a 2-d visual representation of each report based on its content/topics, which gives a direct view of the structure of the whole dataset as well as the outliers.

Anomaly Detection in Sequences

We present a set of novel algorithms which we call sequenceMiner, that detect and characterize anomalies in large sets of high-dimensional symbol sequences that arise from recordings of switch sensors in the cockpits of commercial airliners. While the algorithms we present are general and domain-independent, we focus on a specific problem that is critical to determining system-wide health of a fleet of aircraft. The approach taken uses unsupervised clustering of sequences using the normalized length of he longest common subsequence (nLCS) as a similarity measure, followed by a detailed analysis of outliers to detect anomalies. In this method, an outlier sequence is defined as a sequence that is far away from a cluster. We present new algorithms for outlier analysis that provide comprehensible indicators as to why a particular sequence is deemed to be an outlier. The algorithm provides a coherent description to an analyst of the anomalies in the sequence when compared to more normal sequences. The final section of the paper demonstrates the effectiveness of sequenceMiner for anomaly detection on a real set of discrete sequence data from a fleet of commercial airliners. We show that sequenceMiner discovers actionable and operationally significant safety events. We also compare our innovations with standard HiddenMarkov Models, and show that our methods are superior

Rotor health monitoring combining spin tests and data-driven anomaly detection methods

Health monitoring is highly dependent on sensor systems that are capable of performing in various engine environmental conditions and able to transmit a signal upon a predetermined crack length, while acting in a neutral form upon the overall performance of the engine system. Efforts are under way at NASA Glenn Research Center through support of the Intelligent Vehicle Health Management Project (IVHM) to develop and implement such sensor technology for a wide variety of applications. These efforts are focused on developing high temperature, wireless, low cost, and durable products. In an effort to address technical issues concerning health monitoring, this article considers data collected from an experimental study using high frequency capacitive sensor technology to capture blade tip clearance and tip timing measurements in a rotating turbine engine-like-disk to detect the disk faults and assess its structural integrity. The experimental results composed at a range of rotational speeds from tests conducted at the NASA Glenn Research Center’s Rotordynamics Laboratory are evaluated and integrated into multiple data-driven anomaly detection techniques to identify faults and anomalies in the disk. In summary, this study presents a select evaluation of online health monitoring of a rotating disk using high caliber capacitive sensors and demonstrates the capability of the in-house spin system.

Application of Inductive Monitoring System to Plug Load Anomaly Detection

NASA Ames Research Center’s Sustainability Base is a new 50,000 sq. ft. LEED Platinum office building. Plug loads are expected to account for a significant portion of the overall energy consumption. This is because building design choices have resulted in greatly reduced energy demand from Heating, Ventilation, and Air Conditioning (HVAC) and lighting systems, which are major contributors to energy consumption in traditional buildings. In anticipation of the importance of plug loads in Sustainability Base, a pilot study was conducted to collect data from a variety of plug loads. A number of cases of anomalous or unhealthy behavior were observed including schedule-based rule failures, time-to-standby errors, changed loads, and inter-channel anomalies. These issues prevent effective plug load management; therefore, they are important to promptly identify and correct. The Inductive Monitoring System (IMS) data mining algorithm was chosen to identify errors. This paper details how an automated data analysis program was created, tested and implemented using IMS. This program will be applied to Sustainability Base to maintain effective plug load management system performance, identify malfunctioning equipment, and reduce building energy consumption.