Data quality up to the third observing run of advanced LIGO: Gravity Spy glitch classifications

Classical and Quantum Gravity, Journal Year: 2023, Volume and Issue: 40(6), P. 065004 - 065004

Published: Jan. 25, 2023

Understanding the noise in gravitational-wave detectors is central to detecting and interpreting signals. Glitches are transient, non-Gaussian features that can have a range of environmental instrumental origins. The Gravity Spy project uses machine-learning algorithm classify glitches based upon their time-frequency morphology. resulting set classified be used as input detector-characterisation investigations how mitigate glitches, or data-analysis studies ameliorate impact glitches. Here we present results analysis data up end third observing run Advanced LIGO. We 233981 from LIGO Hanford 379805 Livingston into morphological classes. find distribution differs between two sites. This highlights potential need for quality individually tailored each observatory.

Language: Английский

---

Gravity Spy: lessons learned and a path forward

The European Physical Journal Plus, Journal Year: 2024, Volume and Issue: 139(1)

Published: Jan. 30, 2024

Language: Английский

---

Detecting causality in the frequency domain with Cross-Mapping Coherence

Physica D Nonlinear Phenomena, Journal Year: 2025, Volume and Issue: unknown, P. 134708 - 134708

Published: May 1, 2025

Language: Английский

---

Unsupervised constrained discord detection in IoT-based online crane monitoring

Advanced Engineering Informatics, Journal Year: 2024, Volume and Issue: 60, P. 102444 - 102444

Published: March 2, 2024

Maritime transport is an indispensable element of the global logistics network. Most maritime loading-unloading operations are supported by quay cranes, making their availability and condition critical to port operations. This work identifies discordant trends arising during vibration-based monitoring these cranes in one busiest container ports United Kingdom. proposes unsupervised constrained discord detection approach for irregular but near real-time time series data obtained from multi-modal IoT-based sensors installed on transmitted over a 5G Due live nature seaport's operations, development controlled anomaly signatures baseline reference was not possible. To address challenges incomplete asset health information, batched time-series sensor data, massive volumes, lack assets' vibration signature baselines, this paper unsupervised, robust, fast mechanism that can rapidly highlight chunks received at central server. A Support Vector Machine based One-Class Classifier (OCC-SVM) used identify data. During approach, timestamped add-on were clustered into two weight classes (loaded unloaded) crane's default Programmable Logic Controller (PLC) sensors. The efficacy method checked against crane maintenance logs separate PLC Further, generating synthetic noise-embedded test effectiveness has been devised. Finally, practicality proposed OCC-SVM inspected environment setting.

Language: Английский

---

Transportation Mode Detection Using Learning Methods and Self-Contained Sensors: Review

Sensors, Journal Year: 2024, Volume and Issue: 24(22), P. 7369 - 7369

Published: Nov. 19, 2024

Due to increasing traffic congestion, travel modeling has gained importance in the development of transportion mode detection (TMD) strategies over past decade. Nowadays, recent smartphones, equipped with integrated inertial measurement units (IMUs) and embedded algorithms, can play a crucial role such development. In particular, obtaining much more information on transportation modes used by users through smartphones is very challenging due variety data (accelerometers, magnetometers, gyroscopes, proximity sensors, etc.), standardization issue datasets pertinence learning methods for that purpose. Reviewing latest progress TMD systems important inform readers about detection, best practices classification issues remaining challenges still impact performances. Existing review papers until now offer overviews applications algorithms without tackling specific faced real-world collection classification. Compared these works, proposed provides some novelties as an in-depth analysis current state-of-the-art techniques systems, relying references focusing particularly major existing problems, evaluation methodologies detecting using smartphone IMUs (including dataset structures, sensor types, feature extraction, etc.). This paper help researchers focus their efforts main problems identified.

Language: Английский

---

An anomaly detection method for identifying locations with abnormal behavior of temperature in school buildings

Scientific Reports, Journal Year: 2023, Volume and Issue: 13(1)

Published: Dec. 21, 2023

Time series data collected using wireless sensors, such as temperature and humidity, can provide insight into a building's heating, ventilation, air conditioning (HVAC) system. Anomalies of these sensor measurements be used to identify locations building that are poorly designed or maintained. Resolving the anomalies present in improve thermal comfort occupants, well quality energy efficiency levels space. In this study, we developed scoring method sensors shows collective due environmental issues. This leads identifying problematic within commercial institutional buildings. The Dynamic Warping (DTW) based anomaly detection was applied anomalies. Then, score for each obtained by taking weighted sum number anomalies, vertical distance an point, dynamic time-warping distance. weights were optimized well-defined simulation study applying grid search algorithm. Finally, synthetic set results case could evaluate performance our method. conclusion, newly successfully detects even with over one week, compared machine learning models which need more train themselves.

Language: Английский

---

A machine learning-based algorithm for automated detection of frequency-based events in recorded time series of sensor data

Engineering Applications of Artificial Intelligence, Journal Year: 2024, Volume and Issue: 133, P. 108536 - 108536

Published: May 17, 2024

Automated event detection methods, vital for monitoring technical systems via sensor data, are highly sought after in the automotive industry tracing events time series data. For assessing active vehicle safety systems, a diverse range of driving scenarios is conducted. These involve recording vehicle's behavior using external sensors, enabling evaluation operational performance. In such setting, automated methods not only accelerate but also standardize and objectify by avoiding subjective, human-based appraisals data inspection. This study introduces novel method identifying frequency-based To this aim, mapped to representations time-frequency domain, known as scalograms. After filtering scalograms enhance relevant parts signal, an object model trained detect desired objects By leveraging accuracy networks localizing objects, precise intervals identified. analysis unseen can be detected their with thereafter back mark corresponding interval. The algorithm, evaluated on datasets, achieves precision rate 0.97 detection, providing sharp interval boundaries whose accurate indication human visual inspection challenging. Incorporating into development process enhances reliability which holds major importance rapid testing analysis.

Language: Английский

---

Effects of RF Signal Eventization Encoding on Device Classification Performance

Electronics, Journal Year: 2024, Volume and Issue: 13(11), P. 2020 - 2020

Published: May 22, 2024

The results of first-step research activity are presented for realizing an envisioned “event radio” capability that mimics neuromorphic event-based camera processing. energy efficiency processing is orders magnitude higher than traditional von Neumann-based and realized through synergistic design brain-inspired software hardware computing elements. Relative to cameras, the development devices supporting Radio Frequency (RF) applications severely lagging considerable interest remains in obtaining RF signal In Operational Technology (OT) protection arena, this includes efficient provide reliable device classification. A Random Forest (RndF) classifier considered here as a precursor Spiking Neural Network (SNN) benefits. Both 1D 2D eventized fingerprints generated bursts from NDev = 8 WirelessHART devices. Average correct classification (%C) show fingerprinting best overall using detected events burst Gabor transform responses. This %C ≥ 90% under multiple access interference conditions average NEPB 400 per burst. sufficiently promising motivate next-step aimed at (1) reducing fingerprint dimensionality minimizing required computational resources, (2) transitioning neuromorphic-friendly SNN classifier—two significant steps toward developing necessary elements achieve full benefits event radio.

Language: Английский

---

Q residual non-parametric Distribution on Fault Detection Approach Using Unsupervised LSTM-KDE

International Journal of Prognostics and Health Management, Journal Year: 2024, Volume and Issue: 15(2)

Published: Aug. 19, 2024

It is well known among practitioner, majority collected data from industrial process plant are unlabeled. The historical if utilize, able to provide vital information of condition. Learning unlabeled dataset, this study proposed Unsupervised LSTM-KDE approach as a measure predict fault in plant. residual based detection framework utilized with long short-term memory (LSTM) the main pattern learner for nonlinear and multimode condition that usually appear Furthermore, kernel density (KDE) used determine threshold value non-parametric data. later evaluated numerical Tennessee Eastman dataset. performance also was compared Principal Component Analysis (PCA), Local outlier factor (LOF) Auto-associative Kernel Regression (AAKR) further examine performance. experimental results indicate has better learning accuracy other approaches

Language: Английский

---

RpDelta: Supporting UCR-Suite on Multi-versioning Time Series Data

Lecture notes in computer science, Journal Year: 2023, Volume and Issue: unknown, P. 205 - 220

Published: Jan. 1, 2023

Language: Английский

---