An extreme learning machine based very short-term wind power forecasting method for complex terrain

Açıkgöz, Hakan; Yıldız, Ceyhun; Şekkeli, Mustafa

doi:10.1080/15567036.2020.1755390

Cited by 29 publications

(13 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For SW-KRLS M2C-KRLS machines, the best results are usually 25% of the dictionary size with the largest dictionary size (70). For the KRLS-T M2C-KRLS, the best results are usually 25% of the dictionary size with the smallest dictionary size (20). Similar behavior is found when using any machine listed in Table 1.…”

Section: ) Test Casesupporting

confidence: 61%

“…In Fig. 3 and 4, it can be noted that the ALD-KRLS M2C-KRLS obtains the lowest RMSE (%), mainly when 75% of the dictionary size is set to start a new kernel and it uses dictionaries with 20 examples, the smallest one in the range analyzed by this study (20,30,40,50,60,70). For SW-KRLS M2C-KRLS machines, the best results are usually 25% of the dictionary size with the largest dictionary size (70).…”

Section: ) Test Casementioning

confidence: 79%

“…It is a matrix plot that produces a filled net of shaded rectangles, where each matrix position corresponds to one rectangle. The 'Dictionary size' axis represents the maximum number of examples (20,30,40,50,60,70) saved in each kernel dictionary. The 'Start new kernel' axis shows the percentage of the maximum number of examples reached by a kernel dictionary (25%, 50%, 75%, 100%) before a new dictionary is created.…”

Section: ) Test Casementioning

confidence: 99%

“…NWP input is optional for these models. They are faster during the development period and processing results than physical models, and many have been studied for this purpose, such as autoregressive moving average with exogenous input (ARMAX) [13], autoregressive integrated moving average (ARIMA) [14], neural networks (NNs) [15], support vector machines (SVMs) [16], fuzzy logic [17], [18], and extreme learning machines (ELM) [19], [20]. (iii) Hybrid models, in order to improve forecasting performance, combine different methodologies to take advantage of each method [21], such as weighting-based models, hybrid models with data preprocessing techniques, hybrid models with parameter selection, and optimization techniques and hybrid models with error processing techniques [22]- [26].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

A Self-Adaptive Multikernel Machine Based on Recursive Least-Squares Applied to Very Short-Term Wind Power Forecasting

2021

View full text Add to dashboard Cite

Wind power has contributed significantly to the increase in electricity generation, but a decision-making tool capable of dealing with its variability and limited predictability is necessary. For this purpose, a novel self-adaptive approach for kernel recursive least-squares machines named multiple challengers is introduced in this work, which is successfully used to produce very short-term wind power forecasts at eight wind farms in Australia. The proposed method is based on a competitive tracking method, and the algorithm deals with some common difficulties of kernel methods, e.g., the increasing kernel matrix size associated with time and memory complexities and the overfitting problem. The proposed method always considers the new information received by the model, thus identifying changes in the time series, avoiding abrupt loss of information and maintaining a controlled number of examples since there is an adaptive selection of the active kernel. It works with the smallest dictionary possible, reducing the probability of overfitting. Five minute-ahead wind power forecasts are produced and evaluated in terms of point forecast skill scores and calibration. The results of the proposed method are compared with those provided by other kernel-based versions of the recursive least-squares algorithm, an online version of the extreme learning machine method, and the persistence time series model. An increase in the number of kernels used in the ensemble system can lead to better results when compared with those of single-kernel models.INDEX TERMS Multiple kernel learning, online training, renewable energy, wind power forecasting.

show abstract

Section: ) Test Casesupporting

confidence: 61%

Section: ) Test Casementioning

confidence: 79%

Section: ) Test Casementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Self-Adaptive Multikernel Machine Based on Recursive Least-Squares Applied to Very Short-Term Wind Power Forecasting

2021

View full text Add to dashboard Cite

show abstract

“…However, the existing use of machine learning algorithms in the field of wind power generation focuses on prediction, optimization, and other problems. Reference (Acikgoz, Yildiz, and Sekkeli 2020) used the Extreme Learning Machine (ELM) for wind energy prediction. Reference (Bilgili and Sahin 2013) explained that the artificial neural network method can be used to measure data of surrounding stations to successfully predict the wind speed of any target station.…”

Section: Introductionmentioning

confidence: 99%

Application of machine learning algorithms in wind power: a review

Deng

Xue-hua

Zhou

et al. 2021

Energy Sources, Part A: Recovery, Utilization, and Environmenta

View full text Add to dashboard Cite

Literature review is an overview of existing research, it can also be used to understand research trends and directions. In recent years, the literature associated with wind power has grown rapidly, and it seems inadequate to rely on human resources to study all papers. Very few studies have used machine learning algorithms and visualization approaches to analyze the trends and directions in wind power. To explore these, we collected 50,579 articles (2012-2019) from Web of Science Core Collection (WoSCC) and 785 papers (2012-2019) from China National Knowledge Infrastructure (CNKI). We applied machine learning algorithms including text mining, word segmentation, T-Distributed Stochastic Neighbor Embedding (T-SNE), Auto-Encoder (AE), visual imagery and other methods to analyze and visually display literature in the field of wind power via analysis of the trends with time-sequence, hotspots in abstracts and keywords, and spatial distribution. China, the United States, and Iran are the top three countries in the field of wind power. Through analyzing the trends between 2012-2019, we find that research hotspots have changed. The usage rate of terms such as Power Generation Control, Power Grids, Wind Power Plants, and Wind Turbines has significantly increased, and the corresponding growth rates are 10.91%, 7.06%, 6.28% and 4.33%, respectively. This study also provides information on the relationship of words of abstracts in papers, which shows that these words are mainly divided into four categories: forecasting, optimization, investment, energy and equipment. An implication of this study is that machine learning algorithms may play an important role in the analysis of wind power literature.

show abstract

Improved extreme learning machine with AutoEncoder and particle swarm optimization for short-term wind power prediction

Bourakadi

Yahyaouy

Boumhidi

2021

Neural Comput & Applic

View full text Add to dashboard Cite

An extreme learning machine based very short-term wind power forecasting method for complex terrain

Cited by 29 publications

References 29 publications

A Self-Adaptive Multikernel Machine Based on Recursive Least-Squares Applied to Very Short-Term Wind Power Forecasting

A Self-Adaptive Multikernel Machine Based on Recursive Least-Squares Applied to Very Short-Term Wind Power Forecasting

Application of machine learning algorithms in wind power: a review

Improved extreme learning machine with AutoEncoder and particle swarm optimization for short-term wind power prediction

Contact Info

Product

Resources

About