ARIMA-Based Time Series Model of Stochastic Wind Power Generation

Wind power generation is one of the renewable energy generation methods which maintains good momentum of development at present. However, its extremely intense intermittences and uncertainties bring great challenges to wind power integration and the stable operation of wind power grids. To achieve accurate prediction of wind power generation in China, a hybrid prediction model based on the combination of Wavelet Decomposition (WD) and Long Short-Term Memory neural network (LSTM) is constructed. Firstly, the nonstationary time series is decomposed into multidimensional components by WD, which can effectively reduce the volatility of the original time series and make them more stable and predictable. Then, the components of the original time series after WD are used as input variables of LSTM to predict the national wind power generation. Forty points were used, 80% as training samples and 20% as testing samples. The experimental results show that the MAPE of WD-LSTM is 5.831, performing better than other models in predicting wind power generation in China. In addition, the WD-LSTM model was used to predict the wind power generation in China under different development trends in the next two years.

Download Full-text

Less Information, Similar Performance: Comparing Machine Learning-Based Time Series of Wind Power Generation to Renewables.ninja

Energies ◽

10.3390/en13092277 ◽

2020 ◽

Vol 13 (9) ◽

pp. 2277

Author(s):

Johann Baumgartner ◽

Katharina Gruber ◽

Sofia G. Simoes ◽

Yves-Marie Saint-Drenan ◽

Johannes Schmidt

Keyword(s):

Machine Learning ◽

Power Generation ◽

Time Series ◽

Wind Power ◽

Wind Power Generation ◽

Location Information ◽

Similar Performance ◽

Input Dataset ◽

Synthetic Time Series

Driven by climatic processes, wind power generation is inherently variable. Long-term simulated wind power time series are therefore an essential component for understanding the temporal availability of wind power and its integration into future renewable energy systems. In the recent past, mainly power curve-based models such as Renewables.ninja (RN) have been used for deriving synthetic time series for wind power generation, despite their need for accurate location information and bias correction, as well as their insufficient replication of extreme events and short-term power ramps. In this paper, we assessed how time series generated by machine learning models (MLMs) compare to RN in terms of their ability to replicate the characteristics of observed nationally aggregated wind power generation for Germany. Hence, we applied neural networks to one wind speed input dataset derived from MERRA2 reanalysis with no location information and two with additional location information. The resulting time series and RN time series were compared with actual generation. All MLM time series feature an equal or even better time series quality than RN, depending on the characteristics considered. We conclude that MLM models show a similar performance to RN, even when information on turbine locations and turbine types is unavailable.

Download Full-text

Multi-decadal offshore wind power variability can be mitigated through optimized European allocation

10.5194/egusphere-egu2020-9538 ◽

2020 ◽

Author(s):

Charlotte Neubacher ◽

Jan Wohland ◽

Dirk Witthaut

Keyword(s):

Power Generation ◽

Time Series ◽

Wind Power ◽

North Sea ◽

Wind Power Generation ◽

Offshore Wind ◽

Offshore Wind Power ◽

The North ◽

The North Sea

Wind power generation is a promising technology to reduce greenhouse gas emissions in line with the Paris Agreement. &#160;In the recent years, the global offshore wind market grew around 30% per year but the full potential of this technology is still not fully exploited. In fact, offshore wind power has the potential to generate more than the worldwide energy demand of today. The high variability of wind on many different timescales does, however, pose serious technical challenges for system integration and system security. &#160;With a few exceptions, little focus has been given to multi-decadal variability. Our research therefore focuses on timescales exceeding ten years.Based on detrended wind data from the coupled centennial reanalysis CERA-20C, we calculate long-term offshore wind power generation time series across Europe and analyze their variability with a focus on the North Sea, the Mediterranean Sea and the Atlantic Ocean. Our approach is based on two independent spectral analysis methods, namely power spectral density and singular spectrum analysis. The latter is particularly well suited for relatively short and noisy time series. In both methods an AR(1)-process is considered as a realistic model for the noisy background. The analysis is complemented by computing the 20yr running mean to also gain insight into long term developments and quantify benefits of large-scale balancing.We find strong indications for two significant multidecadal modes, which appear consistently independent of the statistical method and at all locations subject to our investigation. Moreover, we reveal potential to mitigate multidecadal offshore wind power generation variability via spatial balancing in Europe. In particular, optimized allocations off the Portuguese coast and in the North Sea allow for considerably more stable wind power generation on multi-decadal time scales.

Download Full-text

Study on Forecast of Time Series of Wind Velocity for Wind Power Generation by Using Wide Meteorological Data

IEEJ Transactions on Power and Energy ◽

10.1541/ieejpes.128.416 ◽

2008 ◽

Vol 128 (2) ◽

pp. 416-422 ◽

Cited By ~ 8

Author(s):

Kengo Taniguchi ◽

Katsuhiro Ichiyanagi ◽

Kazuto Yukita ◽

Yasuyuki Goto

Keyword(s):

Power Generation ◽

Time Series ◽

Wind Power ◽

Wind Velocity ◽

Meteorological Data ◽

Wind Power Generation

Download Full-text

Comparison of deep learning models for multivariate prediction of time series wind power generation and temperature

Energy Reports ◽

10.1016/j.egyr.2019.11.009 ◽

2020 ◽

Vol 6 ◽

pp. 273-286 ◽

Cited By ~ 3

Author(s):

Sambeet Mishra ◽

Chiara Bordin ◽

Kota Taharaguchi ◽

Ivo Palu

Keyword(s):

Power Generation ◽

Time Series ◽

Deep Learning ◽

Wind Power ◽

Wind Power Generation ◽

Learning Models ◽

Multivariate Prediction

Download Full-text

Time-Series-Based Maximization of Distributed Wind Power Generation Integration

IEEE Transactions on Energy Conversion ◽

10.1109/tec.2007.914180 ◽

2008 ◽

Vol 23 (3) ◽

pp. 968-974 ◽

Cited By ~ 87

Author(s):

L.F. Ochoa ◽

A. Padilha-Feltrin ◽

G.P. Harrison

Keyword(s):

Power Generation ◽

Time Series ◽

Wind Power ◽

Wind Power Generation

Download Full-text

The role of spatial resolution of climate data for the quality of simulated wind power generation – A multi-country analysis

10.5194/egusphere-egu2020-9961 ◽

2020 ◽

Author(s):

Katharina Gruber ◽

Johann Baumgartner ◽

Claude Klöckl ◽

Peter Regner ◽

Johannes Schmidt

Keyword(s):

Power Generation ◽

Time Series ◽

Wind Power ◽

Spatial Resolution ◽

Bias Correction ◽

Wind Power Generation ◽

Specific Power ◽

Wind Speeds ◽

Coarse Spatial Resolution

Integration of a high share of renewables into the energy system comes with its implications. In order to study long and short-term effects on the electrical system, long time series of power generation with high spatial resolution are necessary. In recent years, reanalysis data have become a popular resource for obtaining these power generation datasets, however with the drawback of a rather coarse spatial resolution of several kilometres (MERRA-2: approx. 50km, ERA5: approx. 31 km). In order to overcome this limitation, reanalysis datasets can be combined with other datasets with a higher spatial resolution.In the present study, we assess whether applying the Global Wind Atlas (GWA) developed by the Technical University of Denmark with a spatial resolution of 1 km can improve wind power generation simulated from two reanalysis (MERRA-2 and ERA5) &#160;datasets when compared to observed power generation. Furthermore, these two reanalysis datasets are compared to determine how different spatial resolution of underlying reanalysis datasets affects the resulting time series. Wind power generation is simulated from reanalysis wind speeds based on a physical model. For wind speed bias correction to specific locations, mean wind speeds are approximated to GWA wind speeds. A turbine-specific power curve model scaled by the turbine specific power is applied to account for different technical performance. The analysis is conducted for different regions of the world (USA, Brazil, Austria, South Africa) and for different spatial and temporal levels, to determine how different datasets perform on different spatio-temporal scales.Preliminary results show that bias correction with the GWA has a positive impact on simulation results for MERRA-2, the dataset with lower spatial resolution, while the effect for ERA5 is ambiguous. The error between simulated and observed wind power generation time series can be decreased by spatial and temporal aggregation and a positive, but not very strong correlation between system size (defined by a wind-correlation indicator) and simulation quality (higher correlation, lower error measures) could be identified.Based on these results, we recommend applying additional wind speed bias correction on datasets with rather coarse spatial resolution, while the quality of newer datasets with high spatial resolution may be sufficient to be used without additional bias correction.

Download Full-text

Driving factors of the growth of US wind power generation - A decomposition of historical on-shore wind power generation data

10.5194/egusphere-egu21-11266 ◽

2021 ◽

Author(s):

Peter Regner ◽

Katharina Gruber ◽

Sebastian Wehrle ◽

Johannes Schmidt

Keyword(s):

Power Generation ◽

Time Series ◽

Wind Power ◽

Wind Turbines ◽

Time Series Data ◽

Wind Power Generation ◽

Driving Factors ◽

Series Data ◽

Data Set ◽

Swept Area

US Wind power generation has grown significantly over the last decades, driven by more and larger turbines being installed. However, less is known about how other factors affect the expansion of wind power. In this study, we use historical wind power generation time series, data on installed wind turbines and wind speed time series from the ERA5 data set to quantify driving factors of the growth of US wind power generation. By use of index-decomposition techniques and a regression analysis, we show how different factors affect the output of wind power generation in the US. These include changes in the number of installed turbines, average swept area, park efficiency, location choice, and hub height. Based on this, we discuss potential consequences for the future expansion of wind energy. As expected, the total rotor swept area is responsible for the largest part of the increase in generated power, due to a larger number of installed turbines and larger rotor sizes in particular. Unexpectedly, turbine efficiency slightly declined in the last decades. Wind speeds available to wind turbines have slightly increased. This is a result of larger hub heights, but also of new wind turbines being installed at windier locations.

Download Full-text