15 research outputs found
Bayesian optimized autoencoder for predictive maintenance of smart packaging machines.
Get PDFSmart packaging machines incorporate various components (blades, motors, films) to accomplish the packaging process and are involved in almost all types of the manufacturing industry. Proper maintenance and monitoring of the components over time can help industries to maintain a sustainable production environment. On the contrary, a faulty system may degrade production efficiency and increase the cost. Smart packaging machines comprising several sensors can generate time series data and leverage data driven condition monitoring models to overcome faulty conditions. In this work, we have studied the application of Autoencoder as a data driven condition monitoring tool for the predictive maintenance of packaging machines. The trained Autoencoder on the new system's data can detect worn or degraded components over time. We have also used the Bayesian optimization algorithm to tune the hyper-parameters of the Autoencoder for better predictive performance. Moreover, the reconstruction error is analyzed to identify the worn components in the packaging machine
Topology for preserving feature correlation in tabular synthetic data.
Get PDFTabular synthetic data generating models based on Generative Adversarial Network (GAN) show significant contributions to enhancing the performance of deep learning models by providing a sufficient amount of training data. However, the existing GAN-based models cannot preserve the feature correlations in synthetic data during the data synthesis process. Therefore, the synthetic data become unrealistic and creates a problem for certain applications like correlation-based feature weighting. In this short theoretical paper, we showed a promising approach based on the topology of datasets to preserve correlation in synthetic data. We formulated our hypothesis for preserving correlation in synthetic data and used persistent homology to show that the topological spaces of the original and synthetic data have dissimilarity in topological features, especially in 0th and 1st Homology groups. Finally, we concluded that minimizing the difference in topological features can make the synthetic data space locally homeomorphic to the original data space, and the synthetic data may preserve the feature correlation under homeomorphism conditions
Temporal graph convolutional autoencoder based fault detection for renewable energy applications.
Get PDFDetecting faults in energy generation systems is a challenging task due to the complex nature of the system, measurement noise, and outliers. Recently, researchers have shown an increasing interest in using data-driven models that utilize sensor data for fault detection and diagnosis. However, the nonlinearities, spatial and temporal dependencies in timeseries sensor data make it difficult to develop an effective datadriven fault detection model. To address this issue, we propose an autoencoder model that uses a temporal graph convolutional layer to detect faults in the energy generation process. The proposed model has exceptional spatiotemporal feature learning capabilities, making it ideal for fault detection applications. In addition, we have included a data processing module to reduce noise and eliminate outliers from sensor data. We evaluated the model's performance using wind turbine blades and photovoltaic microgrid datasets. Experimental results have demonstrated that the proposed model outperforms other fault detection models based on graph convolutional autoencoders
HEADS: hybrid ensemble anomaly detection system for Internet-of-Things networks.
Get PDFThe rapid expansion of Internet-of-Things (IoT) devices has revolutionized connectivity, facilitating the exchange of extensive data within IoT networks via the traditional internet. However, this innovation has also increased security concerns due to the presence of sensitive nature of data exchanged within IoT networks. To address these concerns, network-based anomaly detection systems play a crucial role in ensuring the security of IoT networks through continuous network traffic monitoring. However, despite significant efforts from researchers, these detection systems still suffer from lower accuracy in detecting new anomalies and often generate high false alarms. To this end, this study proposes an efficient Hybrid Ensemble learning-based Anomaly Detection System (HEADS) to secure an IoT network from all types of anomalies. The proposed solution is based on a novel hybrid approach to improve the voting strategy for ensemble learning. The ensemble prediction is assisted by a Random Forest-based model obtained through the best F1 score for each label through dataset subset selection. The efficiency of HEADS is evaluated using the publicly available CICIoT2023 dataset. The evaluation results demonstrate an F1 score of 99.75% and a false alarm rate of 0.038%. These observations signify an average 4% improvement in the F1 score while a reduction of 0.7% in the false alarm rate comparing other anomaly detection-based strategies
Enhancing gas-pipeline monitoring with graph neural networks: a new approach for acoustic emission analysis under variable pressure conditions.
Get PDFTraditional machine learning (ML) and deep learning (DL)-based acoustic emission (AE) data-driven condition monitoring models face several reliability issues due to factors such as fluid pressure changes, flange vibrations, inconsistent leak lengths, and noise in AE signals, which vary with pipeline conditions. Additionally, the noise, and variable pressure conditions complicate the interpretation of sensor data, especially in multivariate setups where understanding spatial relationships between sensors is challenging. In response, we have introduced Graph Convolutional Networks (GCNs) to overcome these challenges in AE-based pipeline monitoring for the first time. Our proposed method utilizes a publicly available pipeline monitoring dataset, named GPLA-12, which comprises AE signals to train and evaluate the GCN-based model. This innovative graph construction technique is designed to decipher and comprehend the subtleties in AE signals gathered under various pressure conditions from a multi-variate sensor setup. This approach can potentially establish a new standard in pipeline monitoring research and applications
Salsa20 based lightweight security scheme for smart meter communication in smart grid
Get PDFThe traditional power gird is altering dramatically to a smart power grid with the escalating development of information and communication technology (ICT). Among thousands of electronic devices connected to the grid through communication network, smart meter (SM) is the core networking device. The consolidation of ICT to the electronic devices centered on SM open loophole for the adversaries to launch cyber-attack. Therefore, for protecting the network from the adversaries it is required to design lightweight security mechanism for SM, as conventional cryptography schemes poses extensive computational cost, processing delay and overhead which is not suitable to be used in SM. In this paper, we have proposed a security mechanism consolidating elliptic curve cryptography (ECC) and Salsa20 stream cipher algorithm to ensure security of the network as well as addressing the problem of energy efficiency and lightweight security solution. We have numerically analyzed the performance of our proposed scheme in case of energy efficiency and processing time which reveals that the suggested mechanism is suitable to be used in SM as it consumes less power and requires less processing time to encrypt or decrypt
Time series analysis of electric energy consumption using autoregressive integrated moving average model and Holt Winters model
Get PDFWith the increasing demand of energy, the energy production is not that much sufficient and that’s why it has become an important issue to make accurate prediction of energy consumption for efficient management of energy. Hence appropriate demand side forecasting has a great economical worth. Objective of our paper is to render representations of a suitable time series forecasting model using autoregressive integrated moving average (ARIMA) and Holt Winters model for the energy consumption of Ohio/Kentucky and also predict the accuracy considering different periods (daily, weekly, monthly). We apply these two models and observe that Holt Winters model outperforms ARIMA model in each (daily, weekly and monthly observations) of the cases. We also make a comparison among few other existing analyses of time series forecasting and find out that the mean absolute percentage error (MASE) of Holt Winters model is least considering the monthly data
A comparative study of novelty detection models for zero day intrusion detection in industrial Internet of Things.
No full textThe detection of zero-day attacks in the IoT network is a challenging task due to unknown security vulnerabilities. Also, the unavailability of the data makes it difficult to train a machine learning (ML) model about new vulnerabilities. The existing supervised ML-based Intrusion Detection Systems (IDS) are trained to detect only known attacks. On the contrary, the unsupervised ML-based IDSs show a high false-positive rate. In this paper, we experimented on three novelty detection algorithms named One-Class SVM (OCSVM), Local Outlier Factor (LOF), and Isolation Forest (IF), which follow the one-vs-all strategy for zero-day-intrusion detection for IoT datasets. UNSW-NB15 and IoTID20 datasets are considered for the experiment. Experimental results show that OCSVM outperformed the other two models for zero-day intrusion or unseen anomaly detection in IoT domain
Automated microsegmentation for lateral movement prevention in industrial Internet of Things (IIoT).
No full textThe integration of the IoT network with the Operational Technology (OT) network is increasing rapidly. However, this incorporation of IoT devices into the OT network makes the industrial control system vulnerable to various cyber threats. Hacking an IoT device at the network edge, an attacker can move laterally to compromise the main control server and manipulate the whole control system of the industrial infrastructure. In this paper, we have proposed an automated Micro-segmentation (MS) model based on Machine Learning (ML) algorithms to reduce the lateral movement of an attacker or malware. The proposed model generates the micro-segments based on network traffic and blocks the malicious traffic at each segment. We have taken UNSW-NB15 and IoTID20 datasets for our experiments. Experimental results show that after generating micro-segments and separating the normal traffic, the model limits redundant links and blocks malicious traffic. Limiting the usage of redundant links reduces the lateral movement or spreading of malware. We also considered the deterministic epidemic model to analyze the device infection rate due to lateral movement or malware propagation
Gated recurrent unit autoencoder for fault detection in penicillin fermentation process.
No full textThe penicillin fermentation process is a fed-batch system to generate industrial-scale penicillin for antibiotic production. Any fault in the fermentation tank can lead to low-quality penicillin products, which may cause a severe impact on final antibiotic production. In this paper, we have developed a Gated Recurrent Unit-based Autoencoder deep learning model to detect faults in the batch data of the penicillin fermentation process. In particular, we have used the data shuffling strategy to minimize distribution discrepancy from different batches generated under various controlling conditions for training the deep learning model. We have also compared the model with the Feedforward Autoencoder and Long short-term memory Autoencoder model for fault detection. Experimental results show that our model trained on shuffled data from different batches outperformed the Feedforward and Long short-term memory Autoencoder model with an avergae fault detection rate of 94.74%
