Article
Engineering, Multidisciplinary
Fang Li, Tao Li
Summary: The tourism industry has experienced rapid growth in recent years, particularly in China. China has become the leading tourism country, with tourism revenue contributing significantly to the national income. The increasing number of tourists presents both opportunities and challenges to tourist attractions. In response, many attractions are exploring innovative applications of new technologies, such as speech big data analysis. This paper aims to study an effective model for predicting tourism consumer demand using big data analysis. By analyzing the demands of tourism consumers and establishing a demand forecast model based on key indicators, actionable advice can be provided for the tourism industry. The research contributes to accurate prediction of tourist numbers and formulation of targeted plans for scenic spots.
MATHEMATICAL PROBLEMS IN ENGINEERING
(2022)
Article
Energy & Fuels
Hyojeoung Kim, Sujin Park, Sahm Kim
Summary: This study presents a bottom-up prediction method for electricity consumption in a smart grid environment using time-series clustering and AMI data. The time-series clustering method outperforms the total amount of electricity demand in terms of forecast accuracy. Various models were used for demand forecasting based on clustering, and the model considering exogenous variables performed better than the one without exogenous variables.
Article
Computer Science, Artificial Intelligence
Guneet Singh Kohli, PrabSimran Kaur, Alamjeet Singh, Jatin Bedi
Summary: This research proposes a transfer learning model called TransLearn that combines clustering analysis with deep neural networks to address the challenge of data scarcity in time series prediction. The model creates larger datasets for training deep neural models by clustering time series groups with similar characteristics, trends, and seasonality patterns. The performance evaluation on an energy consumption dataset of Indian states shows that TransLearn outperforms conventional neural models in terms of prediction error.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Sushil Punia, Sonali Shankar
Summary: This article proposes a novel forecasting model that combines advanced sequence modeling and machine learning methods to handle both temporal and covariate variations in demand data, improving forecasting accuracy. Experimental results demonstrate the superior performance of the proposed method compared to benchmarking methods. Additionally, a demand sensing algorithm for real-time demand forecasting is also proposed.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Energy & Fuels
Ru-Guan Wang, Wen-Jen Ho, Kuei-Chun Chiang, Yung-Chieh Hung, Jen-Kuo Tai, Jia-Cheng Tan, Mei-Ling Chuang, Chi-Yun Ke, Yi-Fan Chien, An-Ping Jeng, Chien-Cheng Chou
Summary: In the context of energy conservation and carbon reduction, the deployment of smart meters is crucial in promoting electricity savings. In Taiwan, smart meters are actively installed to empower residents to monitor their electricity consumption and mitigate the risk of electrical fires. This study proposes an integrated approach using knowledge graphs and deep learning techniques to address data anomalies and identify hazardous usage behavior, aiming to improve household electrical safety and energy efficiency.
Article
Construction & Building Technology
Atharvan Dogra, Ashima Anand, Jatin Bedi
Summary: Energy load estimation is crucial for various activities, and deep learning and hybrid models have shown capabilities in capturing non-linear energy load variations. However, these models require a large amount of data and sharing it raises privacy concerns and communication costs. This study proposes a federated learning-based approach that combines time-series patterns and clustering to achieve privacy, cost benefits, and accuracy. Experimental results on a real-world dataset demonstrate that the proposed approach performs as well as the local learning approach while ensuring data privacy and reducing communication costs.
SUSTAINABLE CITIES AND SOCIETY
(2023)
Article
Hospitality, Leisure, Sport & Tourism
Jian-Wu Bi, Chunxiao Li, Hong Xu, Hui Li
Summary: Accurately forecasting daily tourism demand is a meaningful and challenging task, with this study introducing the use of big data, proposing an ensemble of LSTM networks, and providing an effective predictor selection algorithm in ensemble learning for this purpose.
JOURNAL OF TRAVEL RESEARCH
(2022)
Review
Energy & Fuels
Jayanthi Devaraj, Rajvikram Madurai Elavarasan, G. M. Shafiullah, Taskin Jamal, Irfan Khan
Summary: This study delves into the utilization of big data and deep learning models to enhance energy forecasting for improved planning and decision-making in the electric grid. By exploring AI and ML technologies, different DL models and their applications in renewable energy forecasting are examined.
INTERNATIONAL JOURNAL OF ENERGY RESEARCH
(2021)
Article
Energy & Fuels
Kuan-Cheng Lee, Hong-Tzer Yang, Wenjun Tang
Summary: This paper proposes a method to simultaneously determine bidding and purchasing strategies, aiming to address the issue of unawareness of residential and commercial consumers about the electricity market. The method utilizes historical bidding experiences and online learning with the latest bidding experiences to ensure the robustness and adaptability of the model.
Article
Computer Science, Theory & Methods
Kiran Chaudhary, Mansaf Alam, Mabrook S. Al-Rakhami, Abdu Gumaei
Summary: This study analyzed consumer behavior on social media platforms using data collected from various sources. Predictive modeling and machine learning techniques were used to predict consumer behavior on social media platforms. Data preprocessing and mathematical modeling were essential for improving prediction accuracy.
JOURNAL OF BIG DATA
(2021)
Article
Computer Science, Hardware & Architecture
Simona-Vasilica Oprea, Adela Bara, Bogdan George Tudorica, Maria Irene Calinoiu, Mihai Alexandru Botezatu
Summary: The consumption data from smart meters and complex questionnaires can reveal electricity consumers' willingness to adjust their behavior to reduce electricity usage peak and release stress in the power grid. An innovative methodology is proposed to extract valuable information from the increasing data volume to support the enforcement of tariff and demand response strategies.
COMPUTERS & ELECTRICAL ENGINEERING
(2021)
Article
Management
Maxime C. Cohen, Renyu Zhang, Kevin Jiao
Summary: This study discusses how retailers can improve demand prediction using data aggregation and clustering. A practical method called data aggregation with clustering (DAC) is proposed to balance the tradeoff between data aggregation and model flexibility. The DAC algorithm shows consistent and improved prediction accuracy compared to decentralized benchmarks. It also helps retailers uncover meaningful managerial insights.
OPERATIONS RESEARCH
(2022)
Article
Computer Science, Information Systems
Negin Alemazkoor, Mazdak Tootkaboni, Roshanak Nateghi, Arghavan Louhghalam
Summary: This paper proposes an efficient reduced model approach that leverages a hierarchical dimension reduction algorithm to enable scalable load prediction for smart-meter data from millions of customers. The results show significant improvements in forecast accuracy compared to existing approaches.
Article
Chemistry, Analytical
Jenniffer S. Guerrero-Prado, Wilfredo Alfonso-Morales, Eduardo F. Caicedo-Bravo
Summary: The paper introduces a Data Analytics/Big Data framework applied to AMI data in Smart Cities, encompassing architectural view, methodological view, and human expertise as a binding element. This framework aims to support optimal decision-making and transform knowledge into wisdom efficiently.
Article
Environmental Studies
Gang Xie, Yatong Qian, Shouyang Wang
Summary: The growth rate of cruise tourists in China is slowing down, and accurate forecasting of demand is crucial for investment decision-making and planning.
TOURISM MANAGEMENT
(2021)
Article
Energy & Fuels
Chuyi Li, Kedi Zheng, Hongye Guo, Qixin Chen
Summary: With the development of the smart grid, non-intrusive load monitoring (NILM) is a low-cost solution that provides appliance-level load information without installing extra sensors. This paper focuses on industrial NILM and proposes a mixed-integer programming NILM approach for flowline industries. The approach models equipment with different load features separately, exploits temporal dependencies, and introduces a pulse width constraint to improve performance.
Article
Thermodynamics
Yuxuan Gu, Jianxiao Wang, Yuanbo Chen, Wei Xiao, Zhongwei Deng, Qixin Chen
Summary: The rapid increase in the penetration of lithium-ion batteries (LIBs) in transport, energy, and communication systems has prompted the search for a meticulous but simplified LIB model for non-uniform internal state monitoring and online control. A simplified electro-chemical model for LIBs based on the pseudo-two-dimensional (P2D) model is proposed, which includes a rigorous model of non-uniform reaction rates inside the battery and sub-models that capture the non-uniformity of current densities, potentials, and concentrations. The proposed model shows significant improvements in speed, estimation accuracy, and correction speed and accuracy compared to existing models.
Article
Automation & Control Systems
Yuxuan Gu, Yuanbo Chen, Jianxiao Wang, Wei Xiao, Qixin Chen
Summary: By considering internal states and using optimization algorithms and linearization methods, the dispatch of lithium-ion batteries can be optimized to increase energy conversion efficiency and reduce degradation and heat generation.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Engineering, Electrical & Electronic
Kedi Zheng, Hongye Guo, Qixin Chen
Summary: This paper examines the pool strategy for price-makers when faced with imperfect information. Due to the lack of essential transmission parameters, market participants must estimate market outcomes based on historical information. The linear programming model of economic dispatch is analyzed using the rim multi-parametric linear programming theory. The characteristics of system patterns are revealed, and a support vector machine-based multi-class classification model is trained to map offer curves to system patterns. The proposed method is validated on three different power systems.
IEEE TRANSACTIONS ON POWER SYSTEMS
(2023)
Article
Engineering, Electrical & Electronic
Jean-Francois Toubeau, Fei Teng, Thomas Morstyn, Leandro Von Krannichfeldt, Yi Wang
Summary: This paper presents a new privacy-preserving framework for short-term probabilistic forecasting of nodal voltages in local energy communities. The framework utilizes federated learning to keep individuals' data decentralized and incorporates differential privacy to ensure sensitive local information cannot be inferred. The approach also employs cross-series learning to smoothly integrate new clients without data scarcity issues, achieving improved performance compared to non-collaborative models.
IEEE TRANSACTIONS ON SMART GRID
(2023)
Article
Engineering, Electrical & Electronic
Dawei Wang, Kedi Zheng, Fei Teng, Qixin Chen
Summary: This study proposes a solution framework based on quantum annealing to solve the grid partitioning problem. By using the integer slack and binary expansion methods, the problem is transformed into a quadratic unconstrained binary optimization problem, avoiding complex iteration processes and obtaining accurate feasible solutions in a short computation time.
IEEE TRANSACTIONS ON POWER SYSTEMS
(2023)
Article
Engineering, Electrical & Electronic
Masoud Sobhani, Pu Wang, Tao Hong
Summary: Power distribution companies often face inaccurately logged load transfers, resulting in anomalies in load data at the distribution level. This paper proposes two methods, a model-free method and a model-based method, to detect load transfers between two meters. The methods aggregate meters with load transfers and use load profiles to screen for abnormal shapes. The evaluation using data from a U.S. utility company shows the effectiveness of both methods, with the model-based method detecting up to 81% of simulated load transfers with low false positives.
IEEE TRANSACTIONS ON SMART GRID
(2023)
Article
Computer Science, Information Systems
Fangyuan Si, Ning Zhang, Yi Wang, Peng-Yong Kong, Wenjie Qiao
Summary: This article introduces a method of using secure multiparty computation (SMPC) to enhance the security of integrated energy systems (IES) in distributed energy resource integration. By utilizing standardized modeling and privacy-preserving distributed optimization algorithm, this method can achieve optimization control of IES without disclosing sensitive information.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Computer Science, Information Systems
Peng-Yong Kong, Yi Wang
Summary: We use symmetric cryptography to secure communications with smart grid control devices. We propose using unmanned aerial vehicles (UAVs) as couriers to distribute secret keys generated at the control center. An optimization problem is formulated to find the lowest attack risk flight route for key distribution considering key deficiency and battery capacity. We propose an efficient algorithm that can find a low-risk flight route in less than a second, even for large power grid systems.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Energy & Fuels
Dalin Qin, Guobing Liu, Zengxiang Li, Weicheng Guan, Shubao Zhao, Yi Wang
Summary: Accurate mid-term gas demand forecasting is crucial for gas companies and policymakers to meet increasing gas demand. However, data paucity and heterogeneous consumption patterns pose challenges. This paper proposes a novel method, FedCon-LCF, that integrates federated learning, deep contrastive learning, and clustering approaches to address these challenges. The method achieves high-performance forecasting by utilizing data from multiple gas companies and considering heterogeneous patterns. Evaluation on a dataset from 11 Chinese cities shows significant improvements over the benchmark LSTM model.
Article
Engineering, Electrical & Electronic
Dawei Qiu, Jianhong Wang, Zihang Dong, Yi Wang, Goran Strbac
Summary: With the increasing importance of advanced energy management schemes in multi-energy systems, incorporating these systems into the existing energy market is promising for future power systems. The continuous double auction (CDA) market is ideal for enabling peer-to-peer (P2P) energy trading due to its transparency and efficiency. However, modeling the CDA market is challenging due to stochastic and dynamic behaviors of participants. This study proposes a novel multi-agent reinforcement learning method to address these challenges and outperforms existing methods in terms of policy performance, scalability, and computational performance in P2P energy trading.
IEEE TRANSACTIONS ON POWER SYSTEMS
(2023)
Article
Engineering, Electrical & Electronic
Yinxiao Li, Yuxuan Gu, Guannan He, Qixin Chen
Summary: With the rapid development of distributed generation, battery energy storage systems (BESSs) are crucial in supporting the integration of renewable energy into distribution networks. This paper proposes an optimal dispatch model that considers the electrothermal-aging coupling relationship of BESSs. The proposed method effectively captures operational characteristics and optimizes the operational cost and aging cost, leading to an extension of lifetime and increase in profitability of BESSs.
IEEE TRANSACTIONS ON SMART GRID
(2023)
Article
Engineering, Electrical & Electronic
Cheng Feng, Qixin Chen, Yi Wang, Peng-Yong Kong, Hongchao Gao, Songsong Chen
Summary: The growth of variable renewable generations will impact synchronous inertia and increase the need for contingency frequency support. Distributed energy resources (DERs) are able to provide this support through a virtual power plant (VPP). This study proposes the use of equivalent aggregation models for the VPP to participate in contingency reserve services. The models accurately restore the DER aggregation's ability to prevent system frequency from falling and are used to construct a performance-to-cost map for the VPP.
IEEE TRANSACTIONS ON SMART GRID
(2023)
Article
Engineering, Electrical & Electronic
Cheng Feng, Kedi Zheng, Yangze Zhou, Peter Palensky, Qixin Chen
Summary: This paper proposes an online partial-update algorithm based on the alternating direction method of multipliers (ADMM) to solve the communication congestion issue in energy sharing in the electricity market. By restricting update connections and designing a fair and efficient prosumer update scheduling policy, the algorithm can reduce negotiation waiting time and improve energy sharing effectiveness.
IEEE TRANSACTIONS ON SMART GRID
(2023)
Article
Computer Science, Information Systems
Shiguang Song, Victor O. K. Li, Jacqueline C. K. Lam, Yi Wang
Summary: Timely and high-density air quality monitoring is crucial for the development of smart cities. Image-based air pollution estimation using widely deployed stationary cameras and the Internet of Things (IoT) can provide real-time estimation of pollution levels. To overcome limitations of limited samples and scenes from individual cameras, a global and personalized method is proposed to improve the estimation model's generalization and preserve local characteristics. Evaluation results show that the personalized model outperforms the local model in reducing error and improving accuracy.
IEEE INTERNET OF THINGS JOURNAL
(2023)
