Article
Agronomy
Leila Forouhar, Wenyan Wu, Q. J. Wang, Kirsti Hakala
Summary: This study introduces and develops a hybrid framework that combines physical system understanding with data-driven models to improve the prediction of Irrigation Water Demand (IWD). Results show that the integration of physical system understanding improves the performance of the IWD forecasting models, particularly during high-demand periods. The hybrid framework also provides improved system understanding and increased capacity to support operational decisions.
AGRICULTURAL WATER MANAGEMENT
(2022)
Article
Green & Sustainable Science & Technology
Alessandro Incremona, Giuseppe De Nicolao
Summary: This paper addresses the challenges posed by the intermittent nature of renewable energy sources in meeting sustainability goals for increased usage of these sources. It discusses the improvement of short-term load forecasting models and their performance, with a focus on forecasting the 24-hour profile of electric load in Italy. The proposed predictors outperform the models used by the Italian Transmission System Operator, Terna, with significant improvements in mean absolute percentage error and mean absolute error.
SUSTAINABLE ENERGY TECHNOLOGIES AND ASSESSMENTS
(2022)
Article
Automation & Control Systems
Jian Chen, Yuan Gao, Jinyong Shan, Kai Peng, Chen Wang, Hongbo Jiang
Summary: Demand forecasting is crucial in supply chain management, but machine learning models are vulnerable to data poisoning attacks. This article focuses on the vulnerability of linear regression models to targeted poisoning attacks, where the attacker manipulates the forecast behavior on specific samples without compromising overall performance. A gradient-optimization framework for targeted regression poisoning in white-box settings and a regression value manipulation strategy for targeted poisoning in black-box settings are proposed. Countermeasures against these attacks are also discussed. Extensive experiments on real-world datasets demonstrate the effectiveness of the attacks, achieving high prediction deviation with control of less than 1% of the training samples.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Computer Science, Interdisciplinary Applications
Ariele Zanfei, Bruno Melo Brentan, Andrea Menapace, Maurizio Righetti
Summary: This study proposes a deep learning model based on LSTM neural networks for predicting hourly water demand. The model processes different temporal sequences of data with two modules and demonstrates potential in short-term water demand prediction.
JOURNAL OF HYDROINFORMATICS
(2022)
Article
Computer Science, Artificial Intelligence
Adnan Aktepe, Emre Yanik, Suleyman Ersoz
Summary: Various regression analysis methods and artificial intelligence technologies were utilized in the study to develop a forecasting model tailored to the construction machinery industry, aiming to accurately predict future customer demand.
JOURNAL OF INTELLIGENT MANUFACTURING
(2021)
Article
Computer Science, Artificial Intelligence
Faisal Mehmood Butt, Lal Hussain, Syed Hassan Mujtaba Jafri, Haya Mesfer Alshahrani, Fahd N. Al-Wesabi, Kashif Javed Lone, Elsayed M. Tag El Din, Mesfer Al Duhayyim
Summary: This study aims to provide an efficient load prediction system for projecting medium- and long-term load forecasting for different local feeders. By using a hybrid approach and optimizing model parameters, the accuracy of load forecasting is improved. The proposed methods demonstrate stable and better performance in load prediction.
APPLIED ARTIFICIAL INTELLIGENCE
(2022)
Article
Astronomy & Astrophysics
Jacob Coburn, Julian Arnheim, Sara C. Pryor
Summary: This study discusses the importance of short-term forecasting of wind gusts and develops predictive models using wind gust observations from airports in the United States. The results show that artificial neural networks with 3-5 hidden layers generally outperform logistic regression models in terms of accuracy, but deeper networks may lead to increased false alarms and prediction errors. The inclusion of an autoregressive term is critical for model skill, while wind speeds and lapse rates also contribute significantly.
EARTH AND SPACE SCIENCE
(2022)
Article
Computer Science, Information Systems
Shu-Rong Yan, Manwen Tian, Khalid A. A. Alattas, Ardashir Mohamadzadeh, Mohammad Hosein Sabzalian, Amir H. H. Mosavi
Summary: A neural network-based approach is designed for mid-term load forecasting, with the structure and hyperparameters tuned for the best accuracy one year ahead. The approach is practically applied in a region in Iran using real-world data sets of 10 years. The study investigates influential factors such as economic, weather, and social factors, and their impact on accuracy is numerically analyzed. The suggested approach also detects bad data and predicts the 24-hour load pattern, aiding in mid-term planning.
Article
Computer Science, Interdisciplinary Applications
Kien-Trinh T. Bui, Jose F. Torres, David Gutierrez-Aviles, Viet-Ha Nhu, Dieu Tien Bui, Francisco Martinez-Alvarez
Summary: This research proposes a new model, CVOA-LSTM, based on deep learning long short-term memory and the coronavirus optimization algorithm for forecasting the deformations of a hydropower dam. The efficacy of the model is assessed by comparing it with state-of-the-art benchmarks, showing high forecasting capability.
COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING
(2022)
Article
Computer Science, Artificial Intelligence
Nooriya A. Mohammed, Ammar Al-Bazi
Summary: This study introduces an improved ANN model with an Adaptive Backpropagation Algorithm (ABPA) for forecasting long-term electricity load demand, adjusting forecast values to account for behavioral differences between training and future datasets.
NEURAL COMPUTING & APPLICATIONS
(2022)
Article
Chemistry, Analytical
Lina Alhmoud, Ruba Abu Khurma, Ala' M. Al-Zoubi, Ibrahim Aljarah
Summary: Load forecasting is crucial for power system planning and expansion, leading to improved operation, security, stability, cost reduction, and emissions minimization. Utilizing actual data and optimization methods can result in accurate load forecasting, preventing wastage of resources.
Article
Energy & Fuels
Markel Eguizabal, Roberto Garay-Martinez, Ivan Flores-Abascal
Summary: This study utilizes a data-driven model to predict heat loads one hour in advance based on the current and recent history of weather and heat loads. A computationally inexpensive method is developed using existing data quality and resolution from smart meters for load forecasting. The optimal model formulation is discussed and optimized based on four-hour historical values. The model is trained and tested using synthetic data from a building energy simulation, resulting in an absolute error of less than 4% and R-2 values ranging from 0.92 to 0.94.
Article
Environmental Sciences
Sijing Lou, Li Mo, Jianzhong Zhou, Yongqiang Wang, Wenhao He
Summary: This study analyzed the variations in water supply and demand in the upstream Yangtze River over the next 85 years under the influence of climate change and human activity. The results showed that total water demand is projected to peak around 2028, with ecological pressure gradually increasing but not exceeding the stress threshold. The contradiction between monthly supply and demand becomes more pronounced under ecological restrictions.
Article
Engineering, Environmental
Hossein Namdari, Ali Haghighi, Seyed Mohammad Ashrafi
Summary: In this study, a One-Dimensional convolutional neural network (1D CNN) is implemented for the short-term forecast of urban water. The results show that the 1D CNN is superior to other deep learning models in terms of accuracy and training time.
STOCHASTIC ENVIRONMENTAL RESEARCH AND RISK ASSESSMENT
(2023)
Article
Computer Science, Information Systems
M. S. Balamurugan, R. Manojkumar
Summary: Weather forecasting is mostly reliant on statistical and numerical analysis globally, but the variability of climatic factors in different locations makes traditional methods unreliable. Machine learning, based on data-driven prediction, has shown promise in achieving more accurate weather forecasts compared to statistical methods.