ArticlePublisher preview available

Real-time anomaly detection system within the scope of smart factories

Authors:
Cihan Bayraktar at Karabük University
Cihan Bayraktar
Ziya Karakaya at University of Turkish Aeronautical Association
Ziya Karakaya
Hadi Gökçen at Gazi University
Hadi Gökçen
Read publisher preview
Download citation
To read the full-text of this research, you can request a copy directly from the authors.

Abstract and Figures

Anomaly detection is the process of identifying patterns that move differently from normal in a certain order. This process is considered one of the necessary measures for the safety of intelligent production systems. This study proposes a real-time anomaly detection system capable of using and analyzing data in smart production systems consisting of interconnected devices. Synthetic data were preferred in the study because it has difficulties such as high cost and a long time to obtain real anomaly data naturally for learning and testing processes. In order to obtain the necessary synthetic data, a simulation was developed by taking the popcorn production systems as an example. Multi-class anomalies were defined in the obtained data set, and the analysis performances were tested by creating learning models with AutoML libraries. In the field of production systems, while studies on anomaly detection generally focus on whether there is an anomaly in the system, it is aimed to determine which type of anomaly occurs in which device, together with the detection of anomaly by using multi-class tags in the data of this study. As a result of the tests, the Auto-Sklearn library presented the learning models with the highest performance on all data sets. As a result of the study, a real-time anomaly detection system was developed on dynamic data by using the obtained learning models.
Figures - available from: The Journal of Supercomputing
This content is subject to copyright. Terms and conditions apply.
A preview of this full-text is provided by Springer Nature.
Learn more
Content available from The Journal of Supercomputing
This content is subject to copyright. Terms and conditions apply.
Vol.:(0123456789)
The Journal of Supercomputing (2023) 79:14707–14742
https://doi.org/10.1007/s11227-023-05236-w
1 3
Real‑time anomaly detection system withinthescope
ofsmart factories
CihanBayraktar1· ZiyaKarakaya2,3· HadiGökçen4
Accepted: 26 March 2023 / Published online: 9 April 2023
© The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature
2023
Abstract
Anomaly detection is the process of identifying patterns that move differently from nor-
mal in a certain order. This process is considered one of the necessary measures for the
safety of intelligent production systems. This study proposes a real-time anomaly detec-
tion system capable of using and analyzing data in smart production systems consist-
ing of interconnected devices. Synthetic data were preferred in the study because it has
difficulties such as high cost and a long time to obtain real anomaly data naturally for
learning and testing processes. In order to obtain the necessary synthetic data, a simula-
tion was developed by taking the popcorn production systems as an example. Multi-class
anomalies were defined in the obtained data set, and the analysis performances were
tested by creating learning models with AutoML libraries. In the field of production sys-
tems, while studies on anomaly detection generally focus on whether there is an anomaly
in the system, it is aimed to determine which type of anomaly occurs in which device,
together with the detection of anomaly by using multi-class tags in the data of this study.
As a result of the tests, the Auto-Sklearn library presented the learning models with the
highest performance on all data sets. As a result of the study, a real-time anomaly detec-
tion system was developed on dynamic data by using the obtained learning models.
Keywords Industry 4.0· Smart factories· Anomaly detection· AutoML· Machine
learning
* Cihan Bayraktar
cihanbayraktar@karabuk.edu.tr
Ziya Karakaya
ziya.karakaya@atilim.edu.tr; ziya.karakaya@gidatarim.edu.tr
Hadi Gökçen
hgokcen@gazi.edu.tr
1 Department ofComputer Technologies, Karabuk University, Karabuk, Turkey
2 Department ofComputer Engineering, Atılım University, Ankara, Turkey
3 Department ofComputer Engineering, Konya Food andAgriculture University, Konya, Turkey
4 Department ofIndustrial Engineering, Gazi University, Ankara, Turkey
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
... This data is vital for grid operators to monitor and optimize grid performance. By analyzing data from IEDs, operators can pinpoint inefficiencies, predict potential issues, and enhance energy flow for improved reliability and cost-effectiveness [7]. IEDs also facilitate the integration of renewable energy sources by managing their variability to ensure grid stability. ...
... The objective value of the flow is given as flF(u) and the neighbor is offered as NiF(v) , the dimension issue is presented as j and the relation among the new position in the flow is given in Eq. (7). ...
Smart Grid Anomaly Detection Using MFDA and Dilated GRU-based Neural Networks
Article
Full-text available
  • Jan 2025
  • Smart Grids and Energy
The integration of Information and Communication Technologies (ICT) into conventional power grids has given rise to smart grids, which oversee electrical power distribution, generation, and utilization. Despite their benefits, smart grids face communication challenges due to various abnormalities. Detecting these anomalies is crucial for identifying power outages, energy theft, equipment failures, structural faults, power consumption irregularities, and cyber-attacks. While power systems handle natural disturbances adeptly, identifying anomalies caused by cyber-attacks remains complex. This paper introduces an intelligent Deep Learning (DL) approach for smart grid anomaly detection. Data is initially collected from standard sources such as smart meters, weather stations, and user behavior records. Optimal weighted feature selection is performed using the Modified Flow Direction Algorithm (MFDA) before inputting the selected features into an "Adaptive Residual Recurrent Neural Network with Dilated Gated Recurrent Unit (ARRNN-DGRU)" for anomaly identification. Simulation results confirm the model's superior performance, demonstrating a higher detection rate compared to existing methods and enhancing the robustness of the smart grid system.
View
The regression analysis of dry - wet wear outcomes and materials properties of biodegradable MgCu and MgZn, made by P/M, using machine learning models
Article
Full-text available
  • Mar 2025
  • APPL PHYS A-MATER
Incorporating Cu and Zn into Mg as a biomaterial offers a unique opportunity to exploit their antibacterial performance and biodegradability. The main challenge in this area is understanding the ratio and effects of these elements. To achieve this, the present work, based on two separate studies, aims to develop a regression model and apply machine learning (ML) to predict the wear behaviors using the effects of Cu and Zn elements doped into Mg matrix at low ratios on wear and micro and nanostructure properties (Grain size, density, hardness, Crystallite Size, microstrain, dislocation density). The wear behavior of the samples was investigated under 5–20 N loads at a constant sliding speed of 42 mm/s. Auto Sklearn library was used to generate training models that accurately predict the wear loss, friction coefficient, and specific wear rate values. The model showed satisfactory explanatory power and reliability in predicting the volume loss target. It also exhibited remarkable capability in predicting the friction coefficient and specific wear rate targets. The results of sample wear tests (MgZn2 under 15 N) conducted to generate data not included in the dataset showed a high degree of agreement with the ML results. Sensitivity analyses confirmed that Load, Environment, Hardness, and Grain Size are the most influential factors in predicting wear behavior, further validating the model’s reliability and interpretability.
View
Real-Time Anomaly Detection in Low-Light Environments for Enhanced Cybercrime Mitigation
Chapter
  • Feb 2025
With increasing security threats in public spaces, automatic detection of dangerous weapons and fire accidents in surveillance videos is an important capability for pre-emptive safety measures. Anomaly detection in computer vision has become an important research problem with applications in surveillance, industrial inspection, and medical imaging. Recent advances in deep learning provide powerful techniques for analysing visual data and identifying irregularities. The proposed system uses edge detection to isolate potential hazards in video frames, followed by a convolutional neural networks (CNN) classifier and other techniques to identify anomalies such as weapons, fire, and violence in the cropped image patches, and thus provides a survey of deep learning techniques for anomaly detection in CCTV surveillance videos. A key advantage of deep learning in video analysis is its ability to learn feature representations directly from visual data and CNN can capture spatial patterns, while recurrent networks can model temporal dynamics in video. Techniques including pre-trained models such as feature extractors, combining convolutional and recurrent architectures, generating region proposals to focus on anomalous areas, and training models on labelled anomaly data are used. However, challenges remain, including limited training data, difficulty modelling all normal patterns, and reducing false alarm rates. Deep CNNs are well-suited for weapon detection across images and videos, capturing visual features associated with firearms and violent intent. The experimental results validate the proposed anomaly detection system, demonstrating real-time efficiency and impressive accuracy in tasks like violence (91%), weapon (98%), and fire (95%) detection.
View
Optimizing Anomaly Detection in Smart Grids with Modified FDA and Dilated GRU-based Adaptive Residual RNN
Preprint
Full-text available
  • Jan 2024
The integration of Information and Communication Technologies (ICT) into the conventional power grid defines a smart grid, overseeing electrical power distribution, generation, and utilization. Despite its benefits, the smart grid encounters communication challenges due to various abnormalities. Detecting these anomalies is crucial for identifying power outages, energy theft, equipment failure, structural faults, power consumption irregularities, and cyber-attacks. While power systems adeptly handle natural disturbances, discerning cyber-attack-induced anomalies proves complex. This paper introduces an intelligent deep learning approach for smart grid anomaly detection. Initially, data is collected from standard smart meter, weather, and user behavior sources. Optimal weighted feature selection, utilizing the Modified Flow Direction Algorithm (MFDA), precedes inputting selected features into the "Adaptive Residual Recurrent Neural Network with Dilated Gated Recurrent Unit (ARRNN-DGRU)" for anomaly identification. Simulation results affirm the model's superior performance, with a heightened detection rate compared to existing methods, bolstering the smart grid system's robustness.
View
Federated Learning-Based Explainable Anomaly Detection for Industrial Control Systems
Article
Full-text available
  • Jan 2022
We are now witnessing the rapid growth of advanced technologies and their application, leading to Smart Manufacturing (SM). The Internet of Things (IoT) is one of the main technologies used to enable smart factories, which is connecting all industrial assets, including machines and control systems, with the information systems and the business processes. Industrial Control Systems of smart IoT-based factories are one of the top industries attacked by numerous threats, especially unknown and novel attacks. As a result, with the distributed structure of plenty of IoT front-end sensing devices in SM, an effectively distributed anomaly detection (AD) architecture for IoT-based ICSs should: achieve high detection performance, train and learn new data patterns in a fast time scale, and have lightweight to be deployed on resource-constrained edge devices. To date, most solutions for anomaly detection have not fulfilled all of these requirements. In addition, the interpretability of why an instance is predicted to be abnormal is hardly concerned. In this paper, we propose the so- called FedeX architecture to address those challenges. The experiments show that FedeX outperforms 14 other existing anomaly detection solutions on all detection metrics with the liquid storage data set. And with Recall of 1 and F1-score of 0.9857, it also outperforms those solutions on the SWAT data set. FedeX is also proven to be fast in terms of training time of about 7.5 minutes and lightweight in terms of hardware requirement with memory consumption of 14%, allowing us to deploy anomaly detection tasks on top of edge computing infrastructure and in real-time. Besides, FedeX is considered as one of the frameworks at the forefront of interpreting the predicted anomalies by using XAI, which enables experts to make quick decisions and trust the model more.
View
Trends in Discrete Event Simulation
Article
Full-text available
  • Jan 2008
Discrete event simulation technologies have been up and down as global manufacturing industries went through radical changes. The changes have created new problems, challenges and opportunities to the discrete event simulation. On manufacturing applications, it is no longer an isolated model but the distributed modeling and simulation along the supply chain. In order to study the hybrid manufacturing systems, it is critical to have capability to model human performance with different level of skills and under various working conditions. On service applications, the most critical part is to model knowledge workers and their decision making process. The speaker reviews the discrete event simulation technologies; discusses challenges and opportunities presented by both global manufacturing and the knowledge economy.
View
A Comparison of AutoML Tools for Machine Learning, Deep Learning and XGBoost
Conference Paper
Full-text available
  • Jul 2021
This paper presents a benchmark of supervised Automated Machine Learning (AutoML) tools. Firstly, we analyze the characteristics of eight recent open-source AutoML tools (Auto-Keras, Auto-PyTorch, Auto-Sklearn, AutoGluon, H2O AutoML, rminer, TPOT and TransmogrifAI) and describe twelve popular OpenML datasets that were used in the benchmark (divided into regression, binary and multi-class classification tasks). Then, we perform a comparison study with hundreds of computational experiments based on three scenarios: General Machine Learning (GML), Deep Learning (DL) and XGBoost (XGB). To select the best tool, we used a lexicographic approach, considering first the average prediction score for each task and then the computational effort. The best predictive results were achieved for GML, which were further compared with the best OpenML public results. Overall, the best GML AutoML tools obtained competitive results, outperforming the best OpenML models in five datasets. These results confirm the potential of the general-purpose AutoML tools to fully automate the Machine Learning (ML) algorithm selection and tuning.
View
Detecting Cyberattacks using Anomaly Detection in Industrial Control Systems: A Federated Learning approach
Article
Full-text available
  • Nov 2021
  • COMPUT IND
In recent years, the rapid development and wide application of advanced technologies have profoundly impacted industrial manufacturing, leading to smart manufacturing (SM). However, the Industrial IoT (IIoT)-based manufacturing systems are now one of the top industries targeted by a variety of attacks. In this research, we propose detecting Cyberattacks in Industrial Control Systems using Anomaly Detection. An anomaly detection architecture for the IIoT-based SM is proposed to deploy one of the top most concerned networking technique - a Federated Learning architecture - that can detect anomalies for time series data typically running inside an industrial system. The architecture achieves higher detection performance compared to the current detection solution for time series data. It also shows the feasibility and efficiency to be deployed on top of edge computing hardware of an IIoT-based SM that can save 35% of bandwidth consumed in the transmission link between the edge and the cloud. At the expense, the architecture needs to trade off with the computing resource consumed at edge devices for implementing the detection task. However, findings in maximal CPU usage of 85% and average Memory usage of 37% make this architecture totally realizable in an IIoT-based SM.
View
Using Autoencoders for Anomaly Detection and Transfer Learning in IoT
Article
Full-text available
  • Jul 2021
With the development of Internet of Things (IoT) technologies, more and more smart devices are connected to the Internet. Since these devices were designed for better connections with each other, very limited security mechanisms have been considered. It would be costly to develop separate security mechanisms for the diverse behaviors in different devices. Given new and changing devices and attacks, it would be helpful if the characteristics of diverse device types could be dynamically learned for better protection. In this paper, we propose a machine learning approach to device type identification through network traffic analysis for anomaly detection in IoT. Firstly, the characteristics of different device types are learned from their generated network packets using supervised learning methods. Secondly, by learning important features from selected device types, we further compare the effects of unsupervised learning methods including One-class SVM, Isolation forest, and autoencoders for dimensionality reduction. Finally, we evaluate the performance of anomaly detection by transfer learning with autoencoders. In our experiments on real data in the target factory, the best performance of device type identification can be achieved by XGBoost with an accuracy of 97.6%. When adopting autoencoders for learning features from the network packets in Modbus TCP protocol, the best F1 score of 98.36% can be achieved. Comparable performance of anomaly detection can be achieved when using autoencoders for transfer learning from the reference dataset in the literature to our target site. This shows the potential of the proposed approach for automatic anomaly detection in smart factories. Further investigation is needed to verify the proposed approach using different types of devices in different IoT environments.
View
A Novel Anomaly Detection Method for Digital Twin Data Using Deconvolution Operation With Attention Mechanism
Article
  • Jan 2022
In recent years, industrial control systems have evolved toward stability and efficiency, increasing industrial control systems interconnected with the Internet, which means that industrial control systems are facing more serious cyber threats. Thus, it is critical for enterprises to consider issues related to data privacy and network security. Digital twin enables real-time synchronization and simulation of data from various physical components of industrial control systems. However, anomaly detection of twin data is still challenging because existing methods are usually multi-stage with tedious training and detection steps. Therefore, we propose a method called end-to-end anomaly detection with the aim to accomplish real-time anomaly detection quickly and accurately. In order to seek key features, multidimensional deconvolutional network and attention mechanism are applied to our model. The results of this study indicate that our method performs well on precision and F1 score in comparison to the state-of-art methods.
View
View
TSMAE: A Novel Anomaly Detection Approach for Internet of Things Time Series Data Using Memory-Augmented Autoencoder
Article
  • Jan 2022
With the development of the Internet of Things, it has been widely studied and deployed in industrial manufacturing, intelligent transportation, and healthcare systems. The time-series feature of the IoT makes the data density and the data dimension higher, where anomaly detection is important to ensure hardware and software security. However, the traditional anomaly detection algorithm has difficulty meeting this demand, not only in complexity but also accuracy. Sometimes the anomaly can be well reconstructed, resulting in a low reconstruction error. In this paper, we propose a memory-augmented autoencoder approach for detecting anomalies in IoT data, which aims to use reconstruction errors to determine data anomalies. First, a memory mechanism is introduced to suppress the generalization ability of the model, and a memory-augmented autoencoder TSMAE is designed for time-series data anomaly detection. Second, by adding penalties and derivable rectifier functions to loss to make the addressing vector sparse, memory modules are encouraged to extract typical normal patterns, thus inhibiting model generalization ability. Finally, through experiments on ECG and Wafer datasets, the validity of TSMAE is verified, and the rationality of hyperparameter setting is discussed through visualizing the memory module addressing vector.
View
Cognitive analytics platform with AI solutions for anomaly detection
Article
  • Jan 2022
  • COMPUT IND
This work presents a cognitive analytics platform for anomaly detection which is capable to handle, analyze and exploit resourcefully machine data from a shop-floor of factory, so as to support the emerging and growing needs of manufacturing industry. The introduced system contributes to industrial digitization and creation of smart factories by providing a generic platform which is a complete solution supporting standards-based factory connectivity, data management, various AI models training and comparisons, live predictions and real-time visualizations. The proposed system is built on a micro-service architecture, in order to be extendable and adaptive over time, and contains three core modules, the Data Acquisition, the Knowledge Management and the Predictive maintenance, which contribute to machine failure prediction and activate predictive maintenance procedures, to efficient production schemes and decision making, to monitor anomalies and handle unforeseen conditions, to predict future behaviours on time series etc. The proposed platform utilizes continuous re-training mechanisms enabling a self learning approach for the delivery of AI solutions, usable also for various production data, guaranteeing the quality of results without continuous monitoring and human-resources allocation for AI models’ retraining. This cognitive platform is supported by machine learning techniques and deep learning architectures in order to achieve the desired performance in the management of factory processes and needs. All the information generated by the proposed platform is provided to the end user through a user interface that utilizes advanced visualization techniques.
View
View