scholarly journals Utilizing Physics-Based Input Features within a Machine Learning Model to Predict Wind Speed Forecasting Error

2020 ◽  
Author(s):  
Daniel Vassallo ◽  
Raghavendra Krishnamurthy ◽  
Harindra J. S. Fernando
Keyword(s):  
Machine Learning ◽  
Wind Speed ◽  
Moving Average ◽  
Arima Model ◽  
Atmospheric Stability ◽  
Time Of Day ◽  
Turbulent Heat Flux ◽  
Turbulent Heat ◽  
Forecasting Error ◽  
Atmospheric Variables

Abstract. Machine learning is quickly becoming a commonly used technique for wind speed and power forecasting. Many of these methods utilize exogenous variables as input features, but there remains the question of which atmospheric variables provide the most predictive power, especially in handling non-linearities that lead to forecasting error. This investigation addresses this question via creation of a hybrid model that utilizes an autoregressive integrated moving average (ARIMA) model to make an initial wind speed forecast followed by a random forest model that attempts to predict the ARIMA forecasting error using knowledge of exogenous atmospheric variables. Variables conveying information about atmospheric stability and turbulence as well as inertial forcing are found to be useful in dealing with non-linear error prediction. Wind direction (θ) and temperature (T) are found to be the most beneficial individual input features. Streamwise wind speed (U), time of day (t), turbulence intensity (TI), turbulent heat flux (w'T'), θ, and T are found to be particularly useful when used in conjunction.The prediction accuracy of the ARIMA-RF hybrid is compared to that of the persistence and bias-corrected ARIMA models. The ARIMA-RF model is shown to improve upon these commonly employed modeling methods, reducing hourly forecasting error by approximately 30 % below that of the bias-corrected ARIMA model.

scholarly journals Utilizing physics-based input features within a machine learning model to predict wind speed forecasting error

2021 ◽  
Vol 6 (1) ◽  
pp. 295-309
Author(s):  
Daniel Vassallo ◽  
Raghavendra Krishnamurthy ◽  
Harindra J. S. Fernando
Keyword(s):  
Machine Learning ◽  
Wind Speed ◽  
Random Forest ◽  
Hybrid Model ◽  
Arima Model ◽  
Time Of Day ◽  
Random Forest Model ◽  
Forest Model ◽  
Forecasting Error ◽  
Atmospheric Variables

Abstract. Machine learning is quickly becoming a commonly used technique for wind speed and power forecasting. Many machine learning methods utilize exogenous variables as input features, but there remains the question of which atmospheric variables are most beneficial for forecasting, especially in handling non-linearities that lead to forecasting error. This question is addressed via creation of a hybrid model that utilizes an autoregressive integrated moving-average (ARIMA) model to make an initial wind speed forecast followed by a random forest model that attempts to predict the ARIMA forecasting error using knowledge of exogenous atmospheric variables. Variables conveying information about atmospheric stability and turbulence as well as inertial forcing are found to be useful in dealing with non-linear error prediction. Streamwise wind speed, time of day, turbulence intensity, turbulent heat flux, vertical velocity, and wind direction are found to be particularly useful when used in unison for hourly and 3 h timescales. The prediction accuracy of the developed ARIMA–random forest hybrid model is compared to that of the persistence and bias-corrected ARIMA models. The ARIMA–random forest model is shown to improve upon the latter commonly employed modeling methods, reducing hourly forecasting error by up to 5 % below that of the bias-corrected ARIMA model and achieving an R2 value of 0.84 with true wind speed.

scholarly journals Effects of clouds on surface melting of Laohugou glacier No. 12, western Qilian Mountains, China

2018 ◽  
Vol 64 (243) ◽  
pp. 89-99 ◽  
Author(s):  
JIZU CHEN ◽  
XIANG QIN ◽  
SHICHANG KANG ◽  
WENTAO DU ◽  
WEIJUN SUN ◽  
...  
Keyword(s):  
Wind Speed ◽  
Air Temperature ◽  
Meteorological Data ◽  
Water Vapor Pressure ◽  
Longwave Radiation ◽  
Turbulent Heat Flux ◽  
Qilian Mountains ◽  
Shortwave Radiation ◽  
Turbulent Heat ◽  
Glacier Melt

ABSTRACTWe analyzed a 2-year time series of meteorological data (January 2011–December 2012) from three automatic weather stations on Laohugou glacier No. 12, western Qilian Mountains, China. Air temperature, humidity and incoming radiation were significantly correlated between the three sites, while wind speed and direction were not. In this work, we focus on the effects of clouds on other meteorological parameters and on glacier melt. On an average, ~18% of top-of-atmosphere shortwave radiation was attenuated by the clear-sky atmosphere, and clouds attenuated a further 12%. Most of the time the monthly average increases in net longwave radiation caused by clouds were larger than decreases in net shortwave radiation but there was a tendency to lose energy during the daytime when melting was most intense. Air temperature and wind speed related to turbulent heat flux were found to suppress glacier melt during cloudy periods, while increased water vapor pressure during cloudy days could enhance glacier melt by reducing energy loss by latent heat. From these results, we have increased the physical understanding of the significance of cloud effects on continental glaciers.

scholarly journals Coordination Scheduling of Wind-Hydropower Generation and Profit Allocation Based on Shapley Value Method

2020 ◽  
Vol 2020 ◽  
pp. 1-20
Author(s):  
Yi Quan ◽  
Min Jiang ◽  
Hongjun Sun ◽  
Xiaohui Yuan ◽  
Li He ◽  
...  
Keyword(s):  
Wind Speed ◽  
Wind Power ◽  
Shapley Value ◽  
Power Plants ◽  
Moving Average ◽  
Arima Model ◽  
Water Inflow ◽  
Spinning Reserve ◽  
Reservoir Volume ◽  
Two Parameters

The volatility of wind makes the forecasting of wind speed unreliable. The inaccurate forecast in wind speed always leads to generation imbalance and causes Wind Generating Companies’ (WGenCOs) losses in the intrahour market. In contrast to wind power, Hydrogenerating Companies (HGenCOs) can utilize the reservoir volume to settle the fluctuation of water inflow easily. When treated as a specialized Spinning Reserve (SR) unit for wind power, hydropower can help to settle the generation imbalance and obtain more profit in the power market for both power plants. In this paper, the author establishes a coordination scheduling model of wind-hydro alliance which covers the day-ahead market and the intrahour market. First, to evaluate the deviation of the wind-hydro generation in the intrahour market, an imbalance charge rule considering each period of schedule horizon is constructed. Second, the author introduces two parameters to control the resources that hydropower can use to coordinate with wind power. Finally, the author introduces the Shapley value method to allocate the profit of the alliance which comprises several independent entities fairly. For the simulation of uncertainties, the scenario-based approach is used to simulate the water inflow of a reservoir considering the Monte Carlo (MC) method. The wind speed for the intrahour market is forecasted with the Autoregressive Integrated Moving Average (ARIMA) model. Simulations are implemented, and the results show that when treated as an SR unit for wind power, hydropower can diminish the imbalance charges significantly and will improve the revenue of the wind-hydro alliance. Furthermore, the coordination operation also helps reduce the spillage of the reservoir and the curtailment of the wind power to achieve better utilization of renewable energy.

Short-term wind speed forecasting using ARIMA model

Energetika ◽  
2016 ◽  
Vol 62 (1-2) ◽  
Author(s):  
Ernesta Grigonytė ◽  
Eglė Butkevičiūtė
Keyword(s):  
Time Series ◽  
Wind Speed ◽  
Moving Average ◽  
Arima Model ◽  
Model Structure ◽  
Optimal Model ◽  
Short Term ◽  
Wind Speed Forecasting ◽  
Auto Correlation ◽  
Auto Correlation Function

The massive integration of wind power into the power system increasingly calls for better short-term wind speed forecasting which helps transmission system operators to balance the power systems with less reserve capacities. The  time series analysis methods are often used to analyze the  wind speed variability. The  time series are defined as a sequence of observations ordered in time. Statistical methods described in this paper are based on the prediction of future wind speed data depending on the historical observations. This allows us to find a sufficiently good model for the wind speed prediction. The paper addresses a short-term wind speed forecasting ARIMA (Autoregressive Integrated Moving Average) model. This method was applied for a number of different prediction problems, including the short term wind speed forecasts. It is seen as an early time series methodology with well-known limitations in wind speed forecasting, mainly because of insufficient accuracies of the hourly forecasts for the second half of the day-ahead forecasting period. The authors attempt to find the maximum effectiveness of the model aiming to find: (1) how the identification of the optimal model structure improves the forecasting results and (2) what accuracy increase can be gained by reidentification of the structure for a new wind weather season. Both historical and synthetic wind speed data representing the sample locality in the Baltic region were used to run the model. The model structure is defined by rows p, d, q and length of retrospective data period. The structure parameters p (Autoregressive component, AR) and q (Moving Average component, MA) were determined by the Partial Auto-Correlation Function (PACF) and Auto-Correlation Function (ACF), respectively. The model’s forecasting accuracy is based on the root mean square error (RMSE), mean absolute percentage error (MAPE) and mean absolute error (MAE). The results allowed to establish the optimal model structure and the length of the input/retrospective period. The  quantitative study revealed that identification of the  optimal model structure gives significant accuracy improvement against casual structures for 6–8 h forecast lead time, but a season-specific structure is not appropriate for the entire year period. Based on the conducted calculations, we propose to couple the ARIMA model with any more effective method into a hybrid model.

scholarly journals The use of bulk and profile methods for determining surface heat fluxes in the presence of glacier winds

2000 ◽  
Vol 46 (154) ◽  
pp. 445-452 ◽  
Author(s):  
Bruce Denby ◽  
W. Greuell
Keyword(s):  
Wind Speed ◽  
Turbulent Fluxes ◽  
Turbulent Heat Flux ◽  
Heat Fluxes ◽  
Katabatic Wind ◽  
Turbulent Heat ◽  
Glacier Surface ◽  
Surface Heat Fluxes ◽  
Wind Speed Maximum

AbstractA one-dimensional second-order closure model and in situ observations on a melting glacier surface are used to investigate the suitability of bulk and profile methods for determining turbulent fluxes in the presence of the katabatic wind-speed maximum associated with glacier winds. The results show that profile methods severely underestimate turbulent fluxes when a wind-speed maximum is present. The bulk method, on the other hand, only slightly overestimates the turbulent heat flux in the entire region below the wind-speed maximum and is thus much more appropriate for use on sloping glacier surfaces where katabatic winds dominate and wind-speed maxima are just a few meters above the surface.

Short-term wind speed forecasting based on autoregressive moving average with echo state network compensation

2019 ◽  
Vol 44 (2) ◽  
pp. 152-167 ◽  
Author(s):  
Zhongda Tian ◽  
Gang Wang ◽  
Yi Ren
Keyword(s):  
Wind Speed ◽  
Moving Average ◽  
Autoregressive Moving Average ◽  
Echo State Network ◽  
Linear Component ◽  
Short Term ◽  
Autoregressive Moving Average Model ◽  
Forecasting Error ◽  
Forecasting Method ◽  
Moving Average Model

In order to improve the forecasting accuracy of short-term wind speed, a forecasting method based on autoregressive moving average with echo state network compensation is proposed in this article. First, the linear and nonlinear characteristics of short-term wind speed can be determined by Brock–Dechert–Scheinkman statistics method. Then, autoregressive moving average model is used for modeling and to forecast the linear component of short-term wind speed. The linear component of short-term wind speed sequence is obtained. Artificial bee colony algorithm–optimized echo state network model is used as the forecasting model of forecasting error sequences with the nonlinear characteristic. Finally, the final forecasting value is obtained by adding forecasting values of autoregressive moving average model and forecasting error values of echo state network model. k-fold cross-validation is used to improve the generalization ability of the forecasting model. The simulation comparison results show that the proposed forecasting method has higher prediction accuracy with the smaller prediction error. The forecasting indicators are also better than other forecasting methods.

scholarly journals A Machine Learning Approach for Determining the Turbulent Diffusivity in Film Cooling Flows

2017 ◽  
Vol 140 (2) ◽  
Author(s):  
Pedro M. Milani ◽  
Julia Ling ◽  
Gonzalo Saez-Mischlich ◽  
Julien Bodart ◽  
John K. Eaton
Keyword(s):  
Machine Learning ◽  
Film Cooling ◽  
Turbulent Heat Flux ◽  
Navier Stokes ◽  
Data Sets ◽  
Turbulent Heat ◽  
Turbulent Diffusivity ◽  
Film Cooling Effectiveness ◽  
Cooling Flows ◽  
Novel Approach

In film cooling flows, it is important to know the temperature distribution resulting from the interaction between a hot main flow and a cooler jet. However, current Reynolds-averaged Navier–Stokes (RANS) models yield poor temperature predictions. A novel approach for RANS modeling of the turbulent heat flux is proposed, in which the simple gradient diffusion hypothesis (GDH) is assumed and a machine learning (ML) algorithm is used to infer an improved turbulent diffusivity field. This approach is implemented using three distinct data sets: two are used to train the model and the third is used for validation. The results show that the proposed method produces significant improvement compared to the common RANS closure, especially in the prediction of film cooling effectiveness.

Sign in / Sign up

Export Citation Format

Share Document