Article
Computer Science, Information Systems
Mehdi Abbasipour, Mosayeb Afshari Igder, Xiaodong Liang
Summary: Wind power, as a dominant form of renewable energy, is increasingly integrated into power grids with significant technical progress. Accurate wind speed forecasting is crucial for the proper planning and operation of high wind power penetration systems. This paper proposes a novel Neural Network-based method for day-ahead wind speed forecasting, evaluating five different algorithms and analyzing the impact of single- and multi-features on prediction accuracy.
Article
Computer Science, Artificial Intelligence
Khouloud Zouaidia, Salim Ghanemi, Mohamed Saber Rais, Lamine Bougueroua, Wgrzyn-Wolska Katarzyna
Summary: Wind power is considered one of the fastest growing alternative energies, expected to continue its rapid growth. A new hybrid architecture for wind speed forecasting showed superior adaptability and predictive performance, outperforming benchmark models in reducing error metrics values.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Thermodynamics
Mao Yang, Chaoyu Shi, Huiyu Liu
Summary: An improved Fuzzy C-means clustering algorithm is proposed to classify turbines with similar power output characteristics into several categories and select a representative power curve as the equivalent curve of the wind farm, aiming to improve the accuracy of wind power prediction and reduce model complexity.
Article
Energy & Fuels
Chao-Ming Huang, Shin-Ju Chen, Sung-Pei Yang, Hsin-Jen Chen
Summary: This paper proposes an optimal ensemble method for one-day-ahead hourly wind power forecasting. The method includes three stages, including data classification using the k-means method, preliminary forecast generation using five single prediction models, and weight optimization using swarm-based intelligence algorithms. The proposed method is applied to a 3.6 MW wind power generation system in Taiwan and shows improved accuracy compared to single prediction models and other ensemble methods.
Article
Energy & Fuels
Wei Wang, Bin Feng, Gang Huang, Chuangxin Guo, Wenlong Liao, Zhe Chen
Summary: With the rapid increase in wind power installed capacity, day-ahead wind power interval prediction has become increasingly important. This paper proposes a prediction method based on conformal asymmetric multi-quantile generative transformer to provide higher quality intervals. The experiments show that this method outperforms benchmarks by providing narrower prediction intervals with more accurate empirical coverage probability. The average width is reduced by 19.6% compared to symmetric prediction intervals given by common benchmark quantile long short term memory network.
Article
Chemistry, Multidisciplinary
Peihua Xu, Maoyuan Zhang, Zhenhong Chen, Biqiang Wang, Chi Cheng, Renfeng Liu
Summary: This paper proposes a day-ahead wind power short-term prediction model based on deep learning (DWT_AE_BiLSTM). The model utilizes discrete wavelet transform (DWT) for denoising, autoencoder (AE) technology for feature extraction, and bidirectional long short-term memory (BiLSTM) for prediction. Experimental analysis shows that the proposed model is more competitive in forecasting accuracy and stability compared to the shallow neural network model, achieving an increase of 3.86%, 3.22%, and 3.42% in three wind farms, respectively.
APPLIED SCIENCES-BASEL
(2023)
Article
Green & Sustainable Science & Technology
Feng Wu, Rui Jing, Xiao-Ping Zhang, Fei Wang, Yifan Bao
Summary: The paper proposes a combined model of day-ahead wave energy forecast based on an improved grey BP neural network and modified ensemble empirical mode decomposition-autoregressive integrated moving average model. By decomposing wind waves and swells, analyzing the correlation between wind waves and wind speed, and forecasting wave heights, the model effectively predicts and converts wave energy.
IEEE TRANSACTIONS ON SUSTAINABLE ENERGY
(2021)
Article
Computer Science, Artificial Intelligence
Siyi Zhang, Mingbo Liu, Mingde Liu, Zhenxing Lei, Guihua Zeng, Zirui Chen
Summary: In this study, a novel ensemble model considering multiple indicators and error correction was proposed to improve the accuracy of day-ahead wind power prediction. The effectiveness of the model was verified using data from a real wind farm. The results showed that the proposed ensemble model outperformed other models in terms of prediction errors, and further improvements were achieved through error correction.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Information Systems
Chun-Hung Liu, Jyh-Cherng Gu, Ming-Ta Yang
Summary: The development of renewable energies is posing new challenges to energy policy, technology, and business ecosystems, while the interconnection of distributed energy resources presents technical and economic issues for power systems. The simplified LSTM algorithm outperforms the MLP model in solar power generation forecasting, showing promise for short-term applications.
Article
Engineering, Aerospace
Ding Yang, Qingfeng Li, Hanxian Fang, Zhendi Liu
Summary: This paper presents a 1-day forecasting of global total electron content (TEC) using the novel deep learning method pix2pixhd based on GAN. The model outperforms the traditional model IRI-2016 in predicting TEC at a global scale and particularly in low latitude regions.
ADVANCES IN SPACE RESEARCH
(2022)
Article
Energy & Fuels
Linfei Yin, Xinghui Cao, Dongduan Liu
Summary: This study proposes a weighted fully-connected regression network model, which reduces the photovoltaic power prediction errors by automatically selecting well-trained regression networks from multiple groups and configurations, without the need for additional sensors and data sources. Experimental results show that this method can reduce the mean absolute error of photovoltaic power prediction by at least 75.9954% compared to state-of-the-art methods and 68.2937% compared to 18 other famous convolutional neural network methods.
Article
Energy & Fuels
Bogdan Bochenek, Jakub Jurasz, Adam Jaczewski, Gabriel Stachura, Piotr Sekula, Tomasz Strzyzewski, Marcin Wdowikowski, Mariusz Figurski
Summary: In the Polish power system, renewable energy sources like wind and solar energy play a growing role with high variability and low dispatchability. This study explores the prediction of day-ahead wind power at the national level in Poland using machine learning methods, achieving accuracy with a mean absolute percentage error of 26.7% and root mean square error of 4.5% for 2020. Seasonal and daily variations in prediction errors were observed, with higher errors in summer and daytime.
Article
Thermodynamics
Jizhe Dong, Shunjie Han, Xiangxin Shao, Like Tang, Renhui Chen, Longfei Wu, Cunlong Zheng, Zonghao Li, Haolin Li
Summary: The study proposes a method to calculate spinning reserve requirements based on historical virtual wind power data to improve the local adaptability of unit commitment. Application and comparison studies on two systems demonstrate the effectiveness and cost benefits of the method, while sensitivity analyses of different parameters used in the method are also investigated.
Article
Computer Science, Information Systems
Ying Yi Hong, Jay Bhie D. Santos
Summary: This article proposes a novel hybrid model of quantum and residual long short-term memory (LSTM) optimized by particle swarm optimization (PSO) for day-ahead spatiotemporal wind speed forecasting. The proposed model outperforms numerous machine learning methods and deep learning algorithms in terms of accuracy.
IEEE SYSTEMS JOURNAL
(2023)
Article
Computer Science, Information Systems
Elisha C. C. Asiri, C. Y. Chung, Xiaodong Liang
Summary: This paper proposes an artificial neural network (ANN)-based model for regional-scale day-ahead PV power forecasts using weather variables from numerical weather predictions as inputs. The model divides a region into clusters, selects a representative site for each cluster, and generates solar irradiance forecasts and corresponding PV power simulations. The cluster power output is obtained using a linear upscaling model and summed to produce regional-scale power forecasts. The accuracy of the model is validated using power generation data of distributed systems in California, showing a 29% reduction in root mean square error compared to benchmarking models.
Review
Geography, Physical
Igor G. Rizaev, Oktay Karakus, S. John Hogan, Alin Achim
Summary: In this paper, a universal simulation framework for SAR imagery of the sea surface is presented, which includes the superposition of sea-ship waves. The study explores the impact of SAR parameters and hydrodynamic related parameters on the imaging of sea waves and ship wakes. The simulation results agree well with SAR imaging theory and provide a fuller understanding of radar imaging mechanisms for these phenomena.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2022)
Article
Computer Science, Information Systems
Yi Yan, Radwa Adel, Ercan Engin Kuruoglu
Summary: In this paper, an adaptive graph normalized least mean pth power (GNLMP) algorithm is introduced, which utilizes graph signal processing techniques to estimate sampled graph signals corrupted by impulsive noise. The GNLMP algorithm demonstrates better reconstruction ability and convergence performance compared to other algorithms.
JOURNAL OF SIGNAL PROCESSING SYSTEMS FOR SIGNAL IMAGE AND VIDEO TECHNOLOGY
(2023)
Article
Environmental Sciences
Wanli Ma, Oktay Karaku, Paul L. Rosin
Summary: Land cover mapping is a widely used technique in remote sensing computational imaging that provides spatial information on various classes of physical properties on the Earth's surface. It plays a crucial role in developing solutions to environmental problems and faces challenges in integrating complementary information from multi-modal remote sensing imagery.
Review
Acoustics
Tianqi Yang, Oktay Karakus, Nantheera Anantrasirichai, Alin Achim
Summary: This article reviews the application of medical ultrasound imaging in the examination of the lungs, discussing the basis of lung ultrasound examination and various methods for disease detection. It categorizes medical ultrasound image computing methods into model-based and data-driven approaches, with a focus on deep/machine learning techniques in the latter.
IEEE TRANSACTIONS ON ULTRASONICS FERROELECTRICS AND FREQUENCY CONTROL
(2023)
Article
Multidisciplinary Sciences
Maverick Lim Kai Rong, Ercan Engin Kuruoglu, Wai Kin Victor Chan
Summary: This study analyzes the mutations in the SARS-CoV-2 genome sequence by modeling them as a stochastic process in both the time-series and spatial domain of the gene sequence. The results show distinct asymmetries in mutation rate and propensities among different nucleotides and strains, with an average mutation rate of approximately 2 mutations per month. Additionally, the study reveals a characteristic distribution of mutation inter-occurrence distances, which displays a notable pattern similar to other natural diseases. These findings provide interesting insights into the underlying biological mechanism of SARS-CoV-2 mutations and could improve the accuracy of existing mutation prediction models.
Article
Engineering, Electrical & Electronic
Pengcheng Hao, Oktay Karaku, Alin Achim
Summary: Filtering in nonlinear state-space models is challenging due to intractable or complex posterior distribution. Particle filtering (PF) outperforms traditional filters but suffers from sample degeneracy. Stochastic map filter (SMF) solves this problem, but is limited by nonlinear map parameterization. To overcome these limitations, we propose a hybrid filter called PSMF that combines PF and SMF. PSMF updates the likelihood using PF and SMF separately, and incorporates systematic resampling and smoothing to address particle degeneracy caused by PF. The PSMF is compared to reference models on various nonlinear state-space models and shows improved performance with linear and nonlinear map variants.
Article
Multidisciplinary Sciences
Henry Booth, Wanli Ma, Oktay Karakus
Summary: The combination of multi-spectral satellite information and machine learning approaches has been suggested to effectively monitor plastic pollutants in the ocean. The article introduces the development and validation of a supervised machine learning marine debris detection model, the mapping of marine debris density using an automated tool called MAP-Mapper, and the evaluation of the system for out-of-distribution test locations. The proposed approach achieves high precision in detecting marine debris and suspected plastic, and the Marine Debris Map index provides efficient measurement of density mapping findings.
SCIENTIFIC REPORTS
(2023)
Article
Computer Science, Artificial Intelligence
Alessio Guerra, Oktay Karakus
Summary: This article proposes a novel lexicon-based unsupervised sentiment analysis method to measure the hope and fear for the 2022 Ukrainian-Russian Conflict. Reddit data is collected and analyzed to create a new dataset, and the analysis shows that hope strongly decreases after symbolic and strategic losses. Surprisingly, spikes in hope/fear are observed not only after important battles but also after non-military events.
FRONTIERS IN ARTIFICIAL INTELLIGENCE
(2023)
Article
Biology
Hengyuan Miao, Ercan Engin Kuruoglu, Tao Xu
Summary: Advances in sequencing technology have helped biologists uncover DNA cancer mutation process and mutation mechanisms. However, previous studies mainly focused on the type and frequency of mutations, ignoring the importance of spatial information. Therefore, we propose that DNA cancer mutations are location-dependent and that integrating location and inter-distance variables can provide more accurate insights into DNA cancer mutation processes.
COMPUTATIONAL BIOLOGY AND CHEMISTRY
(2023)
Article
Computer Science, Artificial Intelligence
Zidi Gao, Ercan Engin Kuruoglu
Summary: This paper investigates non-stationary time series analysis and forecasting techniques for financial datasets. A hybrid model, GARCH-ATT-LSTM, is proposed to improve the accuracy of price forecasting. Experimental results show that GARCH-ATT-LSTM outperforms other models, indicating the success of combining parametric models with neural network models.
Article
Environmental Sciences
Andrew Rowley, Oktay Karakus
Summary: Air pollution, a major driving force behind climate change and environmental issues, is often invisible and difficult to detect. However, the newly launched Sentinel-5P satellite by the European Space Agency has the potential to study the atmosphere and measure various pollutant information. This study aims to create a multi-modal machine learning model to predict air-quality metrics with high precision, making it applicable to areas without monitoring stations.
REMOTE SENSING OF ENVIRONMENT
(2023)
Article
Engineering, Electrical & Electronic
Junping Hong, Yi Yan, Ercan Engin Kuruoglu, Wai Kin Chan
Summary: This article proposes a graph-based framework for multivariate Generalized Autoregressive Conditional Heteroscedasticity (GARCH) models and applies it to weather prediction and wind power forecasting. The framework decomposes multivariate GARCH models into a linear combination of univariate GARCH processes in the graph spectral domain, reducing parameters and estimation cost. Experimental results demonstrate that the proposed graph models outperform non-graph GARCH models and a Graph Vector Autoregressive Moving Average model in multi-step predictions.
IEEE TRANSACTIONS ON SIGNAL AND INFORMATION PROCESSING OVER NETWORKS
(2023)
Proceedings Paper
Acoustics
Mutong Li, Ercan E. Kuruoglu
Summary: SAR technology is widely used for geo-sensing and mapping due to its advantages. Modelling SAR image data is important, especially for impulsive signal features in urban areas. The Cauchy-Rician model shows potential but requires significant computational power for parameter estimation. This work proposes a new analytical parameter estimation method based on algebraic moments to improve computation speed and accuracy.
2023 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING WORKSHOPS, ICASSPW
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Yi Yan, Ercan E. Kuruoglu
Summary: In this paper, we propose an adaptive Graph-Sign Diffusion (GSD) algorithm to predict the time-varying wind speed in real time, which is crucial for applications like renewable energy generation and weather prediction. The GSD algorithm, formulated on a combination of adaptive graph filtering, graph diffusion, and l(1)-norm optimization, outputs a fast and robust prediction of time-varying graph signals under impulsive noise in an online manner. Experimental results demonstrate the accurate predictions of the GSD algorithm for wind speed at multiple sensor locations.
2023 IEEE CONFERENCE ON ARTIFICIAL INTELLIGENCE, CAI
(2023)
Article
Computer Science, Artificial Intelligence
Wenhui Zhou, Lili Lin, Yongjie Hong, Qiujian Li, Xingfa Shen, Ercan Engin Kuruoglu
Summary: Although learning-based light field disparity estimation has made great progress recently, unsupervised light field learning still faces challenges from occlusions and noises. By analyzing the unsupervised methodology and the light field geometry implied in epipolar plane images (EPIs), a new occlusion-aware unsupervised framework is proposed to address the issue of photometric consistency conflict. This framework includes a geometry-based light field occlusion model and two occlusion-aware unsupervised losses to improve estimation accuracy and preserve occlusion boundaries.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Green & Sustainable Science & Technology
Seyed Majid Hashemzadeh, Mohammed A. Al-Hitmi, Hadi Aghaei, Vafa Marzang, Atif Iqbal, Ebrahim Babaei, Seyed Hossein Hosseini, Shirazul Islam
Summary: This article proposes an interleaved high step-up DC-DC converter topology with an ultra-high voltage conversion ratio for renewable energy applications. The converter utilizes an interleaved structure to reduce the input source current ripple, which is advantageous for solar PV sources. By employing voltage multiplier cells and coupled inductor techniques, the topology enhances the output voltage. The article provides comprehensive operation modes and steady-state analyses, compares the proposed structure with other similar converter topologies, and validates the mathematical analysis with experimental results.
IET RENEWABLE POWER GENERATION
(2024)
Article
Green & Sustainable Science & Technology
Gang Xu, Zixuan Guo
Summary: This paper proposes a two-stage resilience enhancement strategy for the recovery of critical loads after disasters. The first stage utilizes a heuristic algorithm to determine the post-disaster topology, while the second stage incorporates user demand response to maximize the socio-economic value of the recovery.
IET RENEWABLE POWER GENERATION
(2024)
Article
Green & Sustainable Science & Technology
Faruk Oral
Summary: This study investigates the wind characteristics and electricity generation potential from wind energy in the Bitlis-Rahva region in eastern Turkey. Wind data from the Bitlis meteorological station is analyzed using the WindPRO program to determine the wind speed distribution and predict turbine performance. The results show that the region has low wind energy capacity factor, indicating it is not efficient for wind energy investments. However, it is suggested that higher altitudes in the region may have better wind energy utilization.
IET RENEWABLE POWER GENERATION
(2024)
Article
Green & Sustainable Science & Technology
Yingjie Tang, Zheren Zhang, Zheng Xu
Summary: This paper investigates the modular multilevel matrix converter with symmetrically integrated energy storage for low frequency AC system. An evaluation method for the minimum required number of active submodules is presented, and the influences of operating conditions on the minimum required number of active submodules are studied. Issues about the converter control system are also discussed in this paper.
IET RENEWABLE POWER GENERATION
(2024)
