Anomaly Events Research Articles

A Robotic Teleoperation System with Integrated Augmented Reality and Digital Twin Technologies for Disassembling End-of-Life Batteries

Disassembly is a key step in remanufacturing, especially for end-of-life (EoL) products such as electric vehicle (EV) batteries, which are challenging to dismantle due to uncertainties in their condition and potential risks of fire, fumes, explosions, and electrical shock. To address these challenges, this paper presents a robotic teleoperation system that leverages augmented reality (AR) and digital twin (DT) technologies to enable a human operator to work away from the danger zone. By integrating AR and DTs, the system not only provides a real-time visual representation of the robot’s status but also enables remote control via gesture recognition. A bidirectional communication framework established within the system synchronises the virtual robot with its physical counterpart in an AR environment, which enhances the operator’s understanding of both the robot and task statuses. In the event of anomalies, the operator can interact with the virtual robot through intuitive gestures based on information displayed on the AR interface, thereby improving decision-making efficiency and operational safety. The application of this system is demonstrated through a case study involving the disassembly of a busbar from an EoL EV battery. Furthermore, the performance of the system in terms of task completion time and operator workload was evaluated and compared with that of AR-based control methods without informational cues and ‘smartpad’ controls. The findings indicate that the proposed system reduces operation time and enhances user experience, delivering its broad application potential in complex industrial settings.

High-resolution oceanic anoxic event 2 (OAE2) records from the north of eastern Tethys and evidence for short-term sea regression and wildfire at its early phase

Oceanic Anoxic Event 2 (OAE2), occurring at the Cenomanian-Turonian boundary (CTB; 93.9 Ma), is a representative global event of carbon cycling fluctuations and transient climate anomalies. OAE2 was initially discovered and intensively studied in the trans-Atlantic and western Tethys Oceans; however, research on the eastern Tethys Ocean, especially the vast northern regions, is inadequate. Further, there is still considerable controversy regarding the triggering mechanisms and early environmental changes of OAE2. In this study, we present high-resolution carbonate carbon isotopes (δ13Ccarb), total organic carbon, and mercury (Hg) concentration records from the northern part of the eastern Tethys Ocean during the OAE2 interval, sampled at the ZK20-1 drillcore section of the Kukebai Formation in the western Tarim Basin, China. Following the global classification scheme for carbon isotope excursion (CIE) at the CTB, the carbon isotope change curve of this study is classified into six stages (C1-C6), and three substages (C3a, C3b, C3c) in the C3 stage. Sedimentary features reveal a short-term sea-level regression in the Tarim Basin during the onset stage (C2) of OAE2, which likely involved vast regions of the eastern Tethys Ocean and even the Pacific region. The mercury (Hg) concentration data reveal increasing Hg levels in the lower part of the Kukebai Formation, mainly corresponding to the onset stage (C3) of the OAE2, but it is significantly lower than that in other typical volcanogenic Hg anomalies. We preliminarily speculated that the intensified wildfire activity at the onset stage of OAE2 potentially accelerate the Hg extracted from the terrestrial soil or organic matter into the offshore area. This is also supported by the evidence of wildfire and Hg anomalies from the Western Interior Seaway (WIS) of North America.

The Streamflow Response to Multi‐Day Warm Anomaly Events: Sensitivity to Future Warming and Spatiotemporal Variability by Event Magnitude

AbstractPersistent warm temperature anomalies can drive streamflow in regions where snow and glacier melt are important constituents of streamflow. However, the spatiotemporal variability of the streamflow response depends on both the magnitude of the forcing temperature anomalies and the nature of the underlying hydrological system. Here we ask: when, where, and for what magnitude of temperature anomalies will the streamflow response change most rapidly under warming? We use observed streamflow and temperature for 868 basins across Canada to quantify the streamflow response during warm temperature anomalies and how such responses vary in space, time, and by anomaly magnitude. We first identify two temporal modes of the streamflow response, one in autumn and one in spring, the relative strength of which varies by climate. We then use sinusoidal approximations of seasonal temperature cycles to characterize the sensitivity of such modes to changes in annual temperature. At individual basins, we find that relative to moderate warm events, the streamflow response to more extreme warm events is more sensitive to changes in mean annual temperatures, and this sensitivity is greatest in the coastal, southern, and central regions of Canada. Our results have implications for how the hydrological impacts of extreme events, such as heatwaves, will change in space and time under future climate change.

On the Relation between Thermohaline Anomalies and Water Mass Transformation in the Eastern Subpolar North Atlantic

Abstract Decadal thermohaline anomalies carried northward by the North Atlantic Current are an important source of predictability in the North Atlantic region. Here, we investigate whether these thermohaline anomalies influence surface-forced water mass transformation (SFWMT) in the eastern subpolar gyre using the reanalyses EN4.2.2 for the ocean and the ERA5 for the atmosphere. In addition, we follow the propagation of thermohaline anomalies along two paths: in the subpolar North Atlantic and the Norwegian Sea. We use observation-based datasets (HadISST, EN4.2.2, and Ishii) between 1947 and 2021 and apply complex empirical orthogonal functions. Our results show that when a warm anomaly enters the eastern subpolar gyre, more SFWMT occurs in light-density classes (27.0–27.2 kg m−3). In contrast, when a cold anomaly enters the eastern subpolar gyre, more SFWMT occurs in denser classes (27.4–27.5 kg m−3). Following the thermohaline anomalies in both paths, we find alternating warm–salty and cold–fresh subsurface anomalies, repeating throughout the 74-yr-long record with four warm–salty and cold–fresh periods after the 1950s. The cold–fresh anomaly periods happen simultaneously with the Great Salinity Anomaly events. Moreover, the propagation of thermohaline anomalies is faster in the subpolar North Atlantic (SPNA) than in the Norwegian Sea, especially for temperature anomalies. These findings might have implications for our understanding of the decadal variability of the lower limb of the Atlantic meridional overturning circulation and predictability in the North Atlantic region. Significance Statement Anomalously warm–salty or cold–fresh water, carried by the North Atlantic Current toward the Arctic, is a source of climate predictability. In this study, we investigate 1) how these ocean anomalies influence the transformation of water masses in the eastern subpolar gyre and 2) their subsequent propagation poleward and northwestward. The key findings reveal that anomalously warm waters entering the eastern subpolar gyre lead to increased transformation in lighter water masses, while cold anomalies affect denser water masses. These anomalies propagate more than 2 times faster toward the Greenland coast (northwestward) than toward the Arctic (poleward). Our findings contribute to enhancing the understanding of decadal predictability in the northern North Atlantic, including its influence on the Atlantic meridional overturning circulation.

Distilling Privileged Knowledge for Anomalous Event Detection From Weakly Labeled Videos.

Weakly supervised video anomaly detection (WS-VAD) aims to identify the snippets involving anomalous events in long untrimmed videos, with solely text video-level binary labels. A typical paradigm among the existing text WS-VAD methods is to employ multiple modalities as inputs, e.g., RGB, optical flow, and audio, as they can provide sufficient discriminative clues that are robust to the diverse, complicated real-world scenes. However, such a pipeline has high reliance on the availability of multiple modalities and is computationally expensive and storage demanding in processing long sequences, which limits its use in some applications. To address this dilemma, we propose a privileged knowledge distillation (KD) framework dedicated to the WS-VAD task, which can maintain the benefits of exploiting additional modalities, while avoiding the need for using multimodal data in the inference phase. We argue that the performance of the privileged KD framework mainly depends on two factors: 1) the effectiveness of the multimodal teacher network and 2) the completeness of the useful information transfer. To obtain a reliable teacher network, we propose a text cross-modal interactive learning strategy and an anomaly normal discrimination loss, which target learning task-specific cross-modal features and encourage the separability of anomalous and normal representations, respectively. Furthermore, we design both representation- and text logits-level distillation loss functions, which force the unimodal student network to distill abundant privileged knowledge from the text well-trained multimodal teacher network, in a snippet-to-video fashion. Extensive experimental results on three public benchmarks demonstrate that the proposed privileged KD framework can train a lightweight yet effective detector, for localizing anomaly events under the supervision of video-level annotations.

Statistical Characteristics of Remote Sensing Extreme Temperature Anomaly Events in the Taiwan Strait

Statistical Characteristics of Remote Sensing Extreme Temperature Anomaly Events in the Taiwan Strait

A new data-driven framework for progressive anomaly event alerts in spacecraft based on reconstruction discrepancy

In the use of deep learning for spacecraft anomaly detection, a key issue arises from the insufficient extraction of temporal dependencies in telemetry data. This can lead to an inability to accurately discern whether distribution changes in the data are caused by substantive anomalies or merely a consequence of model underfitting. To address this issue, we design a Temporal Dependency Extraction Enhanced Autoencoder model for multi-scale learning of telemetry data. Firstly, this model incorporates Multi-Scale Temporal Dependency Extraction blocks, which integrate self-attention, autoregressive, and feed-forward networks, aimed at systematically dissecting the long-term dependencies, historical information, and complex patterns in telemetry data. Building on these blocks, our model can efficiently and accurately reconstruct telemetry data while maintaining computational efficiency. Furthermore, we utilize an anomaly quantification metric based on the smoothed Manhattan distance, combined with the Drift Streaming Peaks-over-Threshold strategy for setting anomaly thresholds, thus establishing a comprehensive and precise anomaly alerts framework. Finally, we validate our approach using a dataset from the Attitude Control System of a Geostationary Earth Orbit satellite. The experimental results show that our method not only detects anomalies earlier than traditional methods but also provides an in-depth quantitative analysis of anomaly characteristics.

Semantic-driven dual consistency learning for weakly supervised video anomaly detection

Video anomaly detection presents a significant challenge in computer vision, with the aim of distinguishing various anomaly events from numerous normal ones. Weakly supervised video anomaly detection has recently emerged as a promising solution, enabling the detection of anomaly snippets with only video-level annotations. However, knowledge about anomaly annotation remains underutilized, resulting in a gap between visual space and semantic understanding of anomalies, thus failing to capture the clear boundary between anomalies and normalities. Therefore, we propose a weakly supervised paradigm of cross-modal detection and consistency learning, leveraging dual consistency to provide discriminative representations for anomalies at both the semantic-to-target and target-to-snippet levels. Specifically, we introduce a cross-modal detection network, which detects the targets in each frame according to given semantic rules, to derive semantic-consistent visual embeddings. To depict the clear boundary between anomalies and normalities, a cross-domain alignment module is proposed to enhance the discriminative representation of abnormal targets by learning the contextual consistency between the target and snippet embeddings. Our architecture integrates the detection of semantic-consistent targets based on variable semantic rules, ensuring transferable deployment across scenarios and enabling comprehensive identification, localization, and recognition of abnormal events through a “when-where-which” pipeline. The evaluation of our approach is conducted on four widely used public benchmarks: ShanghaiTech, UCSD Ped2, CUHK Avenue, and UBnormal through extensive qualitative and quantitative analyzes. The results demonstrate the remarkable performance of our approach in dealing with the VAD task.

Analysis of Arctic Sea Ice Concentration Anomalies Using Spatiotemporal Clustering

The dynamic changes of sea ice exhibit spatial clustering, and this clustering has characteristics extending from its origin, through its development, and to its dissipation. Current research on sea ice change primarily focuses on spatiotemporal variation trends and remote correlation analysis, and lacks an analysis of spatiotemporal evolution characteristics. This study utilized monthly sea ice concentration (SIC) data from the National Snow and Ice Data Center (NSIDC) for the period from 1979 to 2022, utilizing classical spatiotemporal clustering algorithms to analyze the clustering patterns and evolutionary characteristics of SIC anomalies in key Arctic regions. The results revealed that the central-western region of the Barents Sea was a critical area where SIC anomaly evolutionary behaviors were concentrated and persisted for longer durations. The relationship between the intensity and duration of SIC anomaly events was nonlinear. A positive correlation was observed for shorter durations, while a negative correlation was noted for longer durations. Anomalies predominantly occurred in December, with complex evolution happening in April and May of the following year, and concluded in July. Evolutionary state transitions mainly occurred in the Barents Sea. These transitions included shifts from the origin state in the northwestern margin to the dissipation state in the central-north Barents Sea, from the origin state in the central-north to the dissipation state in the central-south, and from the origin state in the northeastern to the dissipation state in the central-south Barents Sea and southeastern Kara Sea. Various evolutionary states were observed in the same area on the southwest edge of the Barents Sea. These findings provide insights into the evolutionary mechanism of sea ice anomalies.

Controls on Upper Ocean Salinity Variability in the Eastern Subpolar North Atlantic During 1992–2017

AbstractThe upper ocean salinity in the eastern subpolar North Atlantic undergoes decadal fluctuations. A large fresh anomaly event occurred during 2012–2016. Using the ECCOv4r4 state estimate, we diagnose and compare mechanisms of this low salinity event with those of the 1990s fresh anomaly event. To avoid issues related to the choice of reference salinity values in the freshwater budget, we perform a salt mass content budget analysis of the eastern subpolar North Atlantic. It shows that the recent low salt content anomaly occurs due to the circulation of anomalous salinity by mean currents entering the eastern subpolar basin from its western boundary via the North Atlantic Current. This is in contrast to the early 1990s, when the dominant mechanism governing the low salt content anomaly was the transport of the mean salinity field by anomalous currents.

Unraveling the propagation dynamics of passenger congestion in public transportation systems using big smart card data

We propose a comprehensive framework to explore propagation of passenger crowdedness in public transportation systems using 74.16 million trip chains. A deep learning method HSTGCNT is adopted which firstly models normal spatio-temporal dependencies in time series data, and assigns an anomaly score to new data based on its prediction error. The derived discrete anomaly events are further aggregated into groups based on their spatio-temporal proximity by using a clustering algorithm that binds shared nearest neighbors together based on pairwise correlations. The clustering results uncover several key stages located in a propagation chain, such as a core event, adjacent events with spatio-temporal proximity, or a broadcasting group. Weibo data serves as a third-party dataset to verify the above anomaly clusters. Results show that the framework is effective to detect outlier clusters, with recall value being 0.86. Findings may assist public safety departments in taking pro-active actions to interrupt chaos from broadcasting.

A survey of large language models for cyber threat detection

With the increasing complexity of cyber threats and the expanding scope of cyberspace, there exist progressively more challenges in cyber threat detection. It is proven that most previous threat detection models may become inadequate due to the escalation of hacker attacks. However, recent research has shown that some of these problems can be effectively addressed by Large Language Models (LLMs) directly or indirectly. Nowadays, a growing number of security researchers are adopting LLMs for analyzing various cyber threats. According to the investigation, we found that while there are numerous emerging reviews on the utilization of LLMs in some fields of cyber security, there is currently a lack of a comprehensive review on the application of LLMs in the threat detection stage. Through retrieving and collating existing works in recent years, we examined various threat detection and monitoring tasks for which LLMs may be well-suited, including cyber threat intelligence, phishing email detection, threat prediction, logs analysis, and so on. Additionally, the review explored the specific stages of different detection tasks in which LLMs are involved, evaluating the points at which LLMs are optimized. For instance, LLMs have been found to enhance the interpretability of log analysis in real-time anomaly event discovery. Additionally, we discussed some tasks where LLMs may not be suitable and explored future directions and challenges in this field. By providing a detailed status update and comprehensive insights, this review aims to assist security researchers in leveraging LLMs to enhance existing detection frameworks or develop domain-specific LLMs.

Advanced deep learning for masked individual surveillance

During Covid-19 pandemic, face masks have become a ubiquitous protective measure. This poses new challenges for surveillance systems that heavily rely on facial recognition. To address this critical issue, we present a novel enhanced surveillance system that leverages deep learning techniques to tackle two crucial tasks simultaneously: anomaly detection of masked individuals' activities and masked face completion for accurate recognition. For anomaly detection, we employ a custom-designed deep neural network capable of processing real-time video streams. Finding a dataset of anomaly events of masked individuals is a big challenge for us. We handle this challenge using efficient techniques such as Dlib library and other image processing techniques. The network is trained on a diverse dataset encompassing normal and abnormal activities of masked individuals, enabling it to identify suspicious behaviors effectively. The surveillance cameras will exchange information, using a suitable network protocol, about detected anomalies and share relevant image data to aid in decision-making and choose the best images for further processing. In the context of masked face completion, we present a novel architecture called CCGAN network that is a combination of convolutional neural network (CNN) and conditioned generative adversarial network (CGAN) to generate the hidden parts of the face in a form that is accurate and close to the original face shape, as shown in our results. We conduct extensive experiments on publicly available datasets, demonstrating superior performance in both anomaly detection and masked face completion tasks. We have achieved 90% accuracy for anomaly detection of masked people.

Traffic Anomaly Detection based on Spatio-Temporal Hypergraph Convolution Neural Networks

Traffic Anomaly Detection (TAD) is an important and difficult task in Intelligent Transportation Systems (ITS) . Traffic anomaly events are sparse in both spatial and temporal spaces, posing a challenge to the performance of model. Moreover, a single traffic anomaly event can impact multiple road sections in the neighborhood, further undermining the accuracy of TAD. In this paper, we propose a new TAD method based on spatio-temporal hypergraph convolutional neural network. Specifically, we adopt a spatial–temporal augmentation approach for traffic data. This will enhance the performance of detecting sparse anomalies. Meanwhile, we introduce a hypergraph learning method to model the road network. This could capture the spreading features of anomalies for better detection results. Additionally, we design a dynamic hypergraph construction method to extract the evolving relationships of road segments. The proposed model evaluation on the Beijing (SE-BJ) dataset for TAD reveals superior performance compared to state-of-the-art ones.

Generalized Video Anomaly Event Detection: Systematic Taxonomy and Comparison of Deep Models

Video Anomaly Detection (VAD) serves as a pivotal technology in the intelligent surveillance systems, enabling the temporal or spatial identification of anomalous events within videos. While existing reviews predominantly concentrate on conventional unsupervised methods, they often overlook the emergence of weakly-supervised and fully-unsupervised approaches. To address this gap, this survey extends the conventional scope of VAD beyond unsupervised methods, encompassing a broader spectrum termed Generalized Video Anomaly Event Detection (GVAED). By skillfully incorporating recent advancements rooted in diverse assumptions and learning frameworks, this survey introduces an intuitive taxonomy that seamlessly navigates through unsupervised, weakly-supervised, supervised and fully-unsupervised VAD methodologies, elucidating the distinctions and interconnections within these research trajectories. In addition, this survey facilitates prospective researchers by assembling a compilation of research resources, including public datasets, available codebases, programming tools, and pertinent literature. Furthermore, this survey quantitatively assesses model performance, delves into research challenges and directions, and outlines potential avenues for future exploration.

Evaluation and Bias Correction of the ERA5 Reanalysis over the United States for Wind and Solar Energy Applications

The applicability of the ERA5 reanalysis for estimating wind and solar energy generation over the contiguous United States is evaluated using wind speed and irradiance variables from multiple observational data sets. After converting ERA5 and observed meteorological variables into wind power and solar power, comparisons demonstrate that significant errors in the ERA5 reanalysis exist that limit its direct applicability for a wind and solar energy analysis. Overall, ERA5-derived solar power is biased high, while ERA5-derived wind power is biased low. During winter, the ERA5-derived solar power is biased high by 23% on average, while on an annual basis, the ERA5-derived wind power is biased low by 20%. ERA5-derived solar power errors are found to have consistent characteristics across the contiguous United States. Errors for the shortest duration and most extreme solar negative anomaly events are relatively small in the ERA5 when completely overcast conditions occur in both the ERA5 and observations. However, longer-duration anomaly events on weekly to monthly timescales, which include partially cloudy days or a mix of cloudy and sunny days, have significant ERA5 errors. At 10 days duration, the ERA5-derived average solar power produced during the largest negative anomaly events is 62% greater than observed. The ERA5 wind speed and derived wind power negative biases are largely consistent across the central and northwestern U.S., and offshore, while the northeastern U.S. has an overall small net bias. For the ERA5-derived most extreme negative anomaly wind power events, at some sites at 10 days duration, the ERA5-derived wind power produced can be less than half of that observed. Corrections to ERA5 are derived using a quantile–quantile method for solar power and linear regression of wind speed for wind power. These methods are shown to avoid potential over-inflation of the reanalysis variability resulting from differences between point measurements and the temporally and spatially smoother reanalysis values. The corrections greatly reduce the ERA5 errors, including those for extreme events associated with wind and solar energy droughts, which will be most challenging for electric grid operation.

Sharing instant delivery UAVs for crowdsensing: A data-driven performance study

In recent years, there has been a significant increase in demand for instant deliveries, such as rapid delivery of takeaway food and medicine. Many logistics companies are planning to realize real-time delivery services through unmanned aerial vehicles (UAVs). However, costs of running such an autonomous delivery system are too expensive. Fortunately, urban management departments (UMD) that are responsible for monitoring urban environments have the intentions to cooperate with external companies and design crowdsensing methods, so that the duty can be ensured in an autonomous and low-cost method. Motivated by this, instant delivery UAVs are introduced into crowdsensing, so that UAVs can earn additional money for logistics companies while ensuring priority completion of instant deliveries, as well as provide environmental monitoring data to UMD. To be specific, some challenges are firstly identified to combine these two tasks based on real-world data in China. Then a set of data-driven UAVs sharing algorithms is proposed to achieve sensing for urban POIs (Points of Interest) and urban anomaly events. The results of simulations based on real-world datasets demonstrate the efficiency of our methods. The average sensing interval of the POIs within a 368.64π km2 area can be only ≈2.6 min. more than 81% of stochastic events are successfully predicted, over 84% (401 out of 476) of events are successfully sensed, and 60% of events have a UAV to arrive within 6.65min of their occurrence.

Optimal Design of Resilient Carbon Capture, Utilization and Storage (CCUS) Supply Chain Networks under Facility Disruption

In recent years, various kinds of carbon dioxide capture, utilization and storage supply chain network design (CCUS SCND) problems have been extensively studied by scholars from the supply chain management community and other fields. The existing works mainly focus on the various deterministic or uncertainty problems; few works consider the CCUS SCND resilience problem in the context of utilization/storage facility disruptions due to unexpected natural disasters or other geological anomaly events. This paper aims to study the CCUS SCND resilience problem under utilization/storage facility capacity disruption risk. We propose a stochastic mixed-integer linear programming model for the considered problem. In the considered problem, the main decisions related to the following areas are taken into account: supply chain design and planning; facility disruption risk handling, including the optimal determination of facility locations and the matching of carbon dioxide emission sources and utilization/storage facilities; carbon dioxide normal transportation planning; and transshipment planning for various disruption scenarios. Finally, an experimental study comprising a case study from China is conducted to validate the effectiveness and performance of our proposed model. The obtained results show that the supply chain networks for the case study obtained by our proposed model are efficient, cost-effective and resilient in mitigating various kinds of utilization/storage facility disruption scenarios, showing the model can be applied to large-scale CCUS projects to help managers effectively deal with disruption risks. Future research should consider multiple disruption events and propose multiple effective resilience strategies.

Audio–visual representation learning for anomaly events detection in crowds

In recent years, anomaly events detection in crowd scenes attracts many researchers’ attentions, because of its importance to public safety. Existing methods usually exploit visual information to analyze whether any abnormal events have occurred due to only visual sensors are generally equipped in public places. However, when an abnormal event in crowds occurs, sound information may be discriminative to assist the crowd analysis system to determine whether there is an abnormality. Compared with vision information that is easily occluded, audio signals have a certain degree of penetration. Thus, this paper attempt to exploit multi-modal learning for modeling the audio and visual signals simultaneously. To be specific, we design a two-branch network to model different types of information. The first is a typical 3D CNN model to extract temporal appearance feature from video clips. The second is an audio CNN for encoding Log Mel-Spectrogram of audio signals. Finally, by fusing the above features, the more accurate prediction will be produced. We conduct the experiments on SHADE dataset, a synthetic audio–visual dataset in surveillance scenes, and find introducing audio signals effectively improves the performance of anomaly events detection and outperforms other state-of-the-art methods. Furthermore, we will release the code and the pre-trained models as soon as possible.

LSTM Based Neural Network Model for Anomaly Event Detection in Care-Independent Smart Homes

LSTM Based Neural Network Model for Anomaly Event Detection in Care-Independent Smart Homes