AIRG

The calculation of the power flow is based on the well-known Newton-Raphson method and Neural Network method. The power flow calculation aims to evaluate the grid performance parameters such as voltage bus magnitude, angle, real and reactive power flow in the system transmission lines under given load conditions. The standard test system used was a benchmark test system for Networked MGs with four MGs and 40 buses. The data for the entire system has been chosen as per the IEEE Standard 1547-2018. The results showed minimum losses and higher efficiency when performing OPF using NN than the Newton-Raphson method. The efficiency of the power system for the networked MG is 99.3% using Neural Network and 97% using the Newton-Raphson method. The Neural Network method, which mimics how the human brain works based on AI technologies, gave the best results and better efficiency in both cases (Battery as Load/Battery as Source) than the conventional method.

Click here to view the paper

Modelling and Load Frequency Control (LFC) Design of Microgrid Frame-worked As Technological Norm to Multi Energy System (MES)

Researcher: Dr. Abdulla Ismail

In today's world, the trend of using renewable energy sources as an alternative energy source is becoming more and more common due to various driving factors, such as energy scarcity (fossil fuel depletion, etc.) and environmental issues (carbon footprint). This trend is leading to a steady increase in the penetration of renewable energy sources (RES) through microgrids and DER, which consist of RES and are integrated into the power grid. As a result, the power system is becoming more complex and operations are subject to a less structured and uncertain environment. The uncertainties in the power system and the inherent high fluctuations of RES usually make conventional controllers less effective in providing adequate load frequency control (LFC) for diverse and wide operating conditions. In addition, the virtual inertia system, which has been thoroughly studied and verified in this work, cannot maintain and stabilize the frequency deviation within an acceptable range and cannot fully be relied on reducing the uncertainty in operation during disturbances due to the fluctuation of RES and abrupt load changes. Accordingly, detailed and continuously improved control techniques are required for stability and operation uncertainty reduction reasons. Therefore, an advanced control technique based on LQG (LQR with Kalman filter for state estimation) for an isolated microgrid structured as a multi-energy system (MES) is presented in this thesis. The introduction of MES into the microgrid structure helps to achieve a high level of synergy among the different subsystems/units and maximize the use of RES within the system, thereby contributing to frequency stability in MG. The performance of the proposed advanced controller is compared with conventional proportional-integral-derivative (PID) – PSO tuned controllers and interval type 1 (IT1) fuzzy controller design methods. The result shows that the LQG controller has better or lower settling time, improves the dynamic responses, and is highly resilient to severe load or active power disturbances for the subject MG.

Traditionally, this problem is addressed in a supervised manner, where a set of corresponding instance pairs across different KGs is available. However, in real-world applications involving billions of instances, such sets are often incomplete, adversely affecting the recall score. To address this challenge, our proposal explores an unsupervised approach by representing the instance matching problem as a language translation task. This is achieved through the use of a shared-latent space based on Generative Adversarial Networks (GANs), which allows for effective matching even in the absence of complete paired instances.

Machine Learning-Based Intrusion Detection System: A Novel Sinh Cosh Optimizer with XGBoost Approach for Attack Detection in Metaverse

Researcher: Dr. Abdulla Ismail

Networks that combine metaverse and Internet of Things (IoT) technologies are vulnerable to various privacy and security risks. This makes it necessary to develop intrusion detection systems as a very effective way to immediately detect these forms of attacks. The convergence of the metaverse with the IoT will create increasingly diverse virtual and intelligent networks. Integrating IoT networks into the metaverse will facilitate stronger connections between the virtual and physical world by enabling:

• Real-time data processing
• Retrieval
• iInspection

Due to the interactive nature of the metaverse and the numerous human interactions that occur in virtual worlds, a large amount of data is generated which leads to computational complexity in building an intrusion detection system. The proposed work presents a novel model for intrusion detection system that relies on machine learning (ML) and optimization algorithms to address this problem. This model uses a combination of two techniques, namely:

• k-nearest neighbors (KNN)
  to detect and classify attacks
• Sinh Cosh optimizer (SCHO)
  to identify the most important features and obtain the highest performance.

Its purpose is to effectively detect the bulk of attacks on Metaverse-IoT connections. The effectiveness of the proposed SCHO-XGBoost model is evaluated by using the widely recognized benchmark datasets:

• CIC-IDS2017
• CSE-CIC-IDS2018
• CICIoT2023

These datasets include a variety of IoT threats that have the potential to compromise IoT communications. Based on experimental evidence, the proposed model SCHO-XGBoost has proven effective in identifying 17 specific types of threats associated with Metaverse-IoT. It achieved the highest accuracy, recall, and f1-score rate of up to 99.9% on CSE-CIC-IDS2017, where it was compared with other benchmarked ML algorithms and proved superior to all algorithms. In addition, our SCHO-XGBoost model shows greater performance in accurately detecting attacks when compared to other models discussed in the existing literature.

Automated Detection of Cyber Attacks in Healthcare Systems: A Novel Scheme with Advanced Feature Extraction and Classification

Researcher: Dr. Abdulla Ismail

The growing incorporation of interconnected healthcare equipment, software, networks, and operating systems into the Internet of Medical Things (IoMT) poses a risk of security breaches. This is because the IoMT devices lack adequate safeguards against cyberattacks. To address this issue, this article presents a proposed framework for detecting anomalies and cyberattacks. The proposed model employs the 1) K-nearest neighbors (KNN) algorithm for classification, while 2) utilizing long-short term memory (LSTM) for feature extraction, and 3) applying Principal component analysis (PCA) to modify and reduce the features. PCA subsequently enhances the important temporal characteristics identified by the LSTM network. The parameters of the KNN classifier were confirmed by using fivefold cross-validation after making hyperparameter adjustments. The evaluation of the proposed model involved the use of four datasets: 1) telemetry data from the tested IoT network (TON-IoT), 2) Edith Cowan University-Internet of Health Things (ECU-IoHT) dataset, 3) intensive care unit (ICU) dataset, and 4) Washington University in St. Louis Enhanced Healthcare Surveillance System (WUSTL-EHMS) dataset. The proposed model achieved 99.9% accuracy, recall, F1 score, and precision on the WUSTL-EHMS dataset. The proposed technique efficiently mitigates cyber threats in healthcare environments.

Reliable decision tree model to detect Botnet traffic in the Internet of Vehicles (IoV)

Researcher: Dr. Ali Assi

Machine learning algorithms have offered unprecedented solutions for many real-world problems. These algorithms frequently involve using a large number of features. However, in many real-world applications, several of the used features could be not very informative due to uncertainties in the data, such as noise and residual variation. Decision trees are among the most preferred classification models. This is due to their simplicity, explainability, and readability. However, inaccuracies in the data could impact the construction stage of decision trees and thus hinder their results. Feature selection and feature construction present promising research direction to enhance the performance of decision tree models. In this project, a novel strategy including both feature selection and feature construction could be defined.  This strategy can be applied in multiple applications. Out of them, I am working on constructing a reliable decision tree model that can effectively detect Botnet traffic in the Internet of Vehicles (IoV).