A Comparative Study of Reinforcement Learning Algorithms for Distribution Network Reconfiguration With Deep Q-Learning-Based Action Sampling

Gholizadeh, Nastaran; Kazemi, Nazli; Musilek, Petr

doi:10.1109/access.2023.3243549

Cited by 13 publications

(5 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While the reduction may not be substantial, it does contribute to lowering CO 2 emissions. Therefore, CO 2 emission can be computed by (15) after the convergence of the multi-objective function in (11), where C coe f is the CO 2 emission coefficient, NF is the number of feeders, and NSeg is the number of feeder segments.…”

Section: Numerical Results and Discussionmentioning

confidence: 99%

“…In the past decade, advancements in information and communication technology, coupled with the application of artificial intelligence algorithms, have enabled the realization of smart operations in DNs. Notably, in NR, the solution speed and efficiency have seen significant improvements in large-scale systems, utilizing machine learning, deep learning, and reinforcement learning algorithms [9][10][11][12], as well as metaheuristic algorithms [13][14][15][16][17][18][19].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Network Reconfiguration Framework for CO2 Emission Reduction and Line Loss Minimization in Distribution Networks Using Swarm Optimization Algorithms

Lin,

Hsiao,

Huang

et al. 2024

Sustainability

View full text Add to dashboard Cite

This paper presents the development of a generic active distribution network (ADN) operation simulation framework that incorporates selected swarm optimization algorithms (SOAs) for the purpose of reducing CO2 emissions and line loss minimization through network reconfiguration (NR). The framework has been implemented in the ADN of Taipower. Network data, provided by the Distribution Mapping Management System and Distribution Dispatch Control Center (DDCC) of Taipower, were converted into an OpenDSS script to create ADN models. The SOA is integrated into the framework and utilized to determine the statuses of both four-way and two-way switches in the planning and operating stages, in accordance with the proposed multi-objective function and operational constraints. The weightings for these decisions can be customized by distribution operators to meet their specific requirements. In this paper, the weighting for line loss reduction is set to one for minimizing CO2 emissions. The numerical results demonstrate that the proposed ADN framework can recommend a feeder switching scheme to distribution operators, aiming to balance feeder loading and minimize the neutral line current. Finally, this approach leads to reduced line losses and minimizes CO2 emissions. In contrast to relying solely on historical operational experience, this generic ADN reconfiguration framework offers a systematic approach that can significantly contribute to reducing CO2 emissions and enhancing the operational efficiency of ADNs.

show abstract

Section: Numerical Results and Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Network Reconfiguration Framework for CO2 Emission Reduction and Line Loss Minimization in Distribution Networks Using Swarm Optimization Algorithms

Lin,

Hsiao,

Huang

et al. 2024

Sustainability

View full text Add to dashboard Cite

show abstract

“…DQL, which integrates RL and DL, employs DNNs as function approximators to identify the optimal Q-values for actions. The Q-values are determined using DL techniques [49,50].…”

Section: Reward Stat Actionmentioning

confidence: 99%

Enhancing Hyper-Spectral Image Classification with Reinforcement Learning and Advanced Multi-Objective Binary Grey Wolf Optimization

Shoeibi,

Nevisi,

Salehi

et al. 2024

CMC

View full text Add to dashboard Cite

Hyperspectral (HS) image classification plays a crucial role in numerous areas including remote sensing (RS), agriculture, and the monitoring of the environment. Optimal band selection in HS images is crucial for improving the efficiency and accuracy of image classification. This process involves selecting the most informative spectral bands, which leads to a reduction in data volume. Focusing on these key bands also enhances the accuracy of classification algorithms, as redundant or irrelevant bands, which can introduce noise and lower model performance, are excluded. In this paper, we propose an approach for HS image classification using deep Q learning (DQL) and a novel multi-objective binary grey wolf optimizer (MOBGWO). We investigate the MOBGWO for optimal band selection to further enhance the accuracy of HS image classification. In the suggested MOBGWO, a new sigmoid function is introduced as a transfer function to modify the wolves' position. The primary objective of this classification is to reduce the number of bands while maximizing classification accuracy. To evaluate the effectiveness of our approach, we conducted experiments on publicly available HS image datasets, including Pavia University, Washington Mall, and Indian Pines datasets. We compared the performance of our proposed method with several state-of-the-art deep learning (DL) and machine learning (ML) algorithms, including long short-term memory (LSTM), deep neural network (DNN), recurrent neural network (RNN), support vector machine (SVM), and random forest (RF). Our experimental results demonstrate that the Hybrid MOBGWO-DQL significantly improves classification accuracy compared to traditional optimization and DL techniques. MOBGWO-DQL shows greater accuracy in classifying most categories in both datasets used. For the Indian Pine dataset, the MOBGWO-DQL architecture achieved a kappa coefficient (KC) of 97.68% and an overall accuracy (OA) of 94.32%. This was accompanied by the lowest root mean square error (RMSE) of 0.94, indicating very precise predictions with minimal error. In the case of the Pavia University dataset, the MOBGWO-DQL model demonstrated outstanding performance with the highest KC of 98.72% and an impressive OA of 96.01%. It also recorded the lowest RMSE at 0.63, reinforcing its accuracy in predictions. The results clearly demonstrate that the proposed MOBGWO-DQL architecture not only reaches a highly accurate model more quickly but also maintains superior performance throughout the training process.

show abstract

“…The major contribution of the proposed mechanism is the utilization of federated learning to train the decentralized nodes, however, required extensive computational capabilities to perform computations in the blockchain that may reduce throughput [19]. In [20], another trust-driven approach is proposed that uses a reinforcement selection strategy with a double deep Q-Learning algorithm [21] along with federated learning for the identification of malicious and compromised nodes. The proposed model consists of an edge server that process and creates the global model, the weights uploading and model updating process, and the local layers in which IoT devices are used to train the model.…”

Section: Background Studymentioning

confidence: 99%

Securing IoT With Deep Federated Learning: A Trust-Based Malicious Node Identification Approach

et al. 2023

View full text Add to dashboard Cite

The Internet of Things (IoT) has revolutionized the world with its diverse applications and smart connected devices. These IoT devices communicate with each other without human intervention and make life easier in many ways. However, the independence of these devices raises several significant concerns, such as security and privacy preservation due to malicious and compromised nodes within the network. Trust management has been introduced as a less computationally intensive alternative to traditional approaches such as cryptography. The proposed FedTrust approach addresses these challenges by designing a method for identifying malicious and compromised nodes using federated learning. FedTrust trains edge nodes with a provided dataset and forms a global model to predict the abnormal behavior of IoT nodes. The proposed approach utilizes a novel trust dataset consisting of 19 trust parameters from three major components: knowledge, experience, and reputation. To reduce the computational burden, FedTrust employs the concept of communities with dedicated servers to divide the dataset into smaller parts for more efficient training. The proposed approach is extensively evaluated in comparison to existing approaches in terms of accuracy, precision, and other metrics to validate its performance in IoT networks. Simulation results demonstrate the effectiveness of FedTrust by achieving a higher rate of detection and prediction of malicious and compromised nodes.

show abstract

A Comparative Study of Reinforcement Learning Algorithms for Distribution Network Reconfiguration With Deep Q-Learning-Based Action Sampling

Cited by 13 publications

References 35 publications

Network Reconfiguration Framework for CO2 Emission Reduction and Line Loss Minimization in Distribution Networks Using Swarm Optimization Algorithms

Network Reconfiguration Framework for CO2 Emission Reduction and Line Loss Minimization in Distribution Networks Using Swarm Optimization Algorithms

Enhancing Hyper-Spectral Image Classification with Reinforcement Learning and Advanced Multi-Objective Binary Grey Wolf Optimization

Securing IoT With Deep Federated Learning: A Trust-Based Malicious Node Identification Approach

Contact Info

Product

Resources

About