Short-term wind speed forecasting bias correction in the Hangzhou area of China based on a machine learning model

准确的风速预报具有重要的社会意义.在本研究中,使用名为WSFBC-XGB的XGBoost机器学习模型对中国浙江省杭州市自动气象站的短期风速预报误差进行校正.WSFBC-XGB使用本地数值天气预报系统的产品作为输入.将WSFBC-XGB校正的结果与传统MOS(模型输出统计)方法校正的结果进行了比较.结果表明:WSFBC-XGB预报风速的均方根误差(RMSE)/准确率(ACC)分别比NWP和MOS降低/提高了26.1%和7.64%/35.6%和7.02%;对于90%的站点WSFBC-XGB的RMSE/ACC均小于/高于MOS.此外,采用平均杂质减少法对WSFBC-XGB的可解释性进行分析,以帮助用户增加对模型的信任.结果表明:10米风速(47.35%),10米风的经向分量(12.73%),日循环(9.97%)和1000百帕风的经向分量(7.45%)是前4个最重要的特征.WSFBC-XGB模型将有助于提高短期风速预报的准确性,为大型户外活动提供支持.

AdaBoosting Neural Network for Short-Term Wind Speed Forecasting Based on Seasonal Characteristics Analysis and Lag Space Estimation

High accurary in wind speed forcasting remains hard to achieve due to wind’s random distribution nature and its seasonal characteristics.Randomness,intermittent and nonstationary usually cause the portion problem of the wind speed forecasting.Seasonal characteristics of wind speed means that its feature distribution is inconsistent.This typically results that the persistence of excitation for modeling can not be guaranteed,and may severely reduce the possibilities of high precise forecasting model.In this paper,we proposed two effective solutions to solve the problems caused by the randomness and seasonal characteristics of the wind speed.(1)Wavelet analysis is used to extract the robust components of time series and reduce the influence of randomness.(2)Based on the energy distribution about the extracted amplitude and associated frequency,seasonal characteristics of wind speed are analyzed based on self-similarity in periodogram under scales range generated by wavelet transformation.Thus,the original dataset is reasonably divided into subsest which can effectively reflect the seasonal distribution characteristics of wind speed.In addition,two strategies are given to optimal model structure and improve the forecasting accuracy:(1)The forecasting model’s lag space is approximately estimated by the Lipschitz quotient to improve the generality ability of the feedforward neural network.(2)The forecasting accuracy and model robustness are further improved by the wavelet decomposition combined with AdaBoosting neural network.Finally,experimental evaluation based on the dataset from National Renewable Energy Laboratory(NREL)is given to demonstrate the performance of the proposed approach.

Short-term Wind Speed Prediction with a Two-layer Attention-based LSTM

Wind speed prediction is of great importance because it affects the efficiency and stability of power systems with a high proportion of wind power.Temporal-spatial wind speed features contain rich information;however,their use to predict wind speed remains one of the most challenging and less studied areas.This paper investigates the problem of predicting wind speeds for multiple sites using temporal and spatial features and proposes a novel two-layer attentionbased long short-term memory(LSTM),termed 2Attn-LSTM,a unified framework of encoder and decoder mechanisms to handle temporal-spatial wind speed data.To eliminate the unevenness of the original wind speed,we initially decompose the preprocessing data into IMF components by variational mode decomposition(VMD).Then,it encodes the spatial features of IMF components at the bottom of the model and decodes the temporal features to obtain each component's predicted value on the second layer.Finally,we obtain the ultimate prediction value after denormalization and superposition.We have performed extensive experiments for short-term predictions on real-world data,demonstrating that 2Attn-LSTM outperforms the four baseline methods.It is worth pointing out that the presented 2Atts-LSTM is a general model suitable for other spatial-temporal features.

Effects of wind speed,underlying surface,and seed morphological traits on the secondary seed dispersal in the Tengger Desert,China

The maintenance of sand-fixing vegetation is important for the stability of artificial sand-fixing systems in which seed dispersal plays a key role.Based on field wind tunnel experiments using 11 common plant species on the southeastern edge of the Tengger Desert,China,we studied the secondary seed dispersal in the fixed and semi-fixed sand dunes as well as in the mobile dunes in order to understand the limitations of vegetation regeneration and the maintenance of its stability.Our results indicated that there were significant variations among the selected 11 plant species in the threshold of wind speed(TWS).The TWS of Caragana korshinskii was the highest among the 11 plant species,whereas that of Echinops gmelinii was the lowest.Seed morphological traits and underlying surface could generally explain the TWS.During the secondary seed dispersal processes,the proportions of seeds that did not disperse(no dispersal)and only dispersed over short distance(short-distance dispersal within the wind tunnel test section)were significantly higher than those of seeds that were buried(including lost seeds)and dispersed over long distance(long-distance dispersal beyond the wind tunnel test section).Compared with other habitats,the mobile dunes were the most difficult places for secondary seed dispersal.Buried seeds were the easiest to be found in the semi-fixed sand dunes,whereas fixed sand dunes were the best sites for seeds that dispersed over long distance.The results of linear mixed models showed that after controlling the dispersal distance,smaller and rounder seeds dispersed farther.Shape index and wind speed were the two significant influencing factors on the burial of seeds.The explanatory power of wind speed,underlying surface,and seed morphological traits on the seeds that did not disperse and dispersed over short distance was far greater than that on the seeds that were buried and dispersed over long distance,implying that the processes and mechanisms of burial and long-distance dispersal are more complex.In summary,most seeds in the study area either did not move,were buried,or dispersed over short distance,promoting local vegetation regeneration.

Relationships between Terrain Features and Forecasting Errors of Surface Wind Speeds in a Mesoscale Numerical Weather Prediction Model

Numerical weather prediction(NWP)models have always presented large forecasting errors of surface wind speeds over regions with complex terrain.In this study,surface wind forecasts from an operational NWP model,the SMS-WARR(Shanghai Meteorological Service-WRF ADAS Rapid Refresh System),are analyzed to quantitatively reveal the relationships between the forecasted surface wind speed errors and terrain features,with the intent of providing clues to better apply the NWP model to complex terrain regions.The terrain features are described by three parameters:the standard deviation of the model grid-scale orography,terrain height error of the model,and slope angle.The results show that the forecast bias has a unimodal distribution with a change in the standard deviation of orography.The minimum ME(the mean value of bias)is 1.2 m s^(-1) when the standard deviation is between 60 and 70 m.A positive correlation exists between bias and terrain height error,with the ME increasing by 10%−30%for every 200 m increase in terrain height error.The ME decreases by 65.6%when slope angle increases from(0.5°−1.5°)to larger than 3.5°for uphill winds but increases by 35.4%when the absolute value of slope angle increases from(0.5°−1.5°)to(2.5°−3.5°)for downhill winds.Several sensitivity experiments are carried out with a model output statistical(MOS)calibration model for surface wind speeds and ME(RMSE)has been reduced by 90%(30%)by introducing terrain parameters,demonstrating the value of this study.

Experimental study on the variation of optical remote sensing imaging characteristics of internal solitary waves with wind speed

Optical remote sensing has been widely used to study internal solitary waves(ISWs).Wind speed has an important effect on ISW imaging of optical remote sensing.The light and dark bands of ISWs cannot be observed by optical remote sensing when the wind is too strong.The relationship between the characteristics of ISWs bands in optical remote sensing images and the wind speed is still unclear.The influence of wind speeds on the characteristics of the ISWs bands is investigated based on the physical simulation experiments with the wind speeds of 1.6,3.1,3.5,3.8,and 3.9 m/s.The experimental results show that when the wind speed is 3.9 m/s,the ISWs bands cannot be observed in optical remote sensing images with the stratification of h_(1)∶h_(2)=7∶58,ρ_(1)∶ρ_(2)=1∶1.04.When the wind speeds are 3.1,3.5,and 3.8 m/s,which is lower than 3.9 m/s,the ISWs bands can be obtained in the simulated optical remote sensing image.The location of the band’s dark and light extremum and the band’s peak-to-peak spacing are almost not affected by wind speed.More-significant wind speeds can cause a greater gray difference of the light-dark bands.This provided a scientific basis for further understanding of ISW optical remote sensing imaging.

Research on the IL-Bagging-DHKELM Short-Term Wind Power Prediction Algorithm Based on Error AP Clustering Analysis

With the continuous advancement of China’s“peak carbon dioxide emissions and Carbon Neutrality”process,the proportion of wind power is increasing.In the current research,aiming at the problem that the forecasting model is outdated due to the continuous updating of wind power data,a short-term wind power forecasting algorithm based on Incremental Learning-Bagging Deep Hybrid Kernel Extreme Learning Machine(IL-Bagging-DHKELM)error affinity propagation cluster analysis is proposed.The algorithm effectively combines deep hybrid kernel extreme learning machine(DHKELM)with incremental learning(IL).Firstly,an initial wind power prediction model is trained using the Bagging-DHKELM model.Secondly,Euclidean morphological distance affinity propagation AP clustering algorithm is used to cluster and analyze the prediction error of wind power obtained from the initial training model.Finally,the correlation between wind power prediction errors and Numerical Weather Prediction(NWP)data is introduced as incremental updates to the initial wind power prediction model.During the incremental learning process,multiple error performance indicators are used to measure the overall model performance,thereby enabling incremental updates of wind power models.Practical examples show the method proposed in this article reduces the root mean square error of the initial model by 1.9 percentage points,indicating that this method can be better adapted to the current scenario of the continuous increase in wind power penetration rate.The accuracy and precision of wind power generation prediction are effectively improved through the method.

A Study on Reconstruction of Surface Wind Speed in China Due to Various Climate Variabilities

Using European Centre for Medium-Range Weather Forecasts Reanalysis V5(ERA5)reanalysis data,this study investigated the reconstruction effects of various climate variabilities on surface wind speed in China from 1979 to 2022.The results indicated that the reconstructed annual mean wind speed and the standard deviation of the annual mean wind speed,utilizing various climate variability indices,exhibited similar spatial modes to the reanalysis data,with spatial correlation coefficients of 0.99 and 0.94,respectively.In the reconstruction of six major wind power installed capacity provinces/autonomous regions in China,the effects were notably good for Hebei and Shanxi provinces,with the correlation coefficients for the interannual regional average wind speed time series being 0.65 and 0.64,respectively.The reconstruction effects of surface wind speed differed across seasons,with spring and summer reconstructions showing the highest correlation with reanalysis data.The correlation coefficients for all seasons across most regions in China ranged between 0.4 and 0.8.Among the reconstructed seasonal wind speeds for the six provinces/autonomous regions,Shanxi Province in spring exhibited the highest correlation with the reanalysis,with a coefficient of 0.61.The large-scale climate variability indices showed good reconstruction effects on the annual mean wind speed in China,and could explain the interannual variability trends of surface wind speed in most regions of China,particularly in the main wind energy provinces/autonomous regions.

Wind Speed Prediction Based on Improved VMD-BP-CNN-LSTM Model

Amid the randomness and volatility of wind speed, an improved VMD-BP-CNN-LSTM model for short-term wind speed prediction was proposed to assist in power system planning and operation in this paper. Firstly, the wind speed time series data was processed using Variational Mode Decomposition (VMD) to obtain multiple frequency components. Then, each individual frequency component was channeled into a combined prediction framework consisting of BP neural network (BPNN), Convolutional Neural Network (CNN) and Long Short-Term Memory Network (LSTM) after the execution of differential and normalization operations. Thereafter, the predictive outputs for each component underwent integration through a fully-connected neural architecture for data fusion processing, resulting in the final prediction. The VMD decomposition technique was introduced in a generalized CNN-LSTM prediction model;a BPNN model was utilized to predict high-frequency components obtained from VMD, and incorporated a fully connected neural network for data fusion of individual component predictions. Experimental results demonstrated that the proposed improved VMD-BP-CNN-LSTM model outperformed other combined prediction models in terms of prediction accuracy, providing a solid foundation for optimizing the safe operation of wind farms.

Dividing the transit wind speeds into intervals as a favorable methodology for analyzing the relationship between wind speed and the aerodynamic impedance of vegetation in semiarid grasslands

In grassland ecosystems,the aerodynamic roughness(Z0)and frictional wind speed(u*)contribute to the aerodynamic impedance of the grassland canopy.Thus,they are often used in the studies of wind erosion and evapotranspiration.However,the effect of wind speed and grazing measures on the aerodynamic impedance of the grassland canopy has received less analysis.In this study,we monitored wind speeds at multiple heights in grazed and grazing-prohibited grasslands for 1 month in 2021,determined the transit wind speed at 2.0 m height by comparing wind speed differences at the same height in both grasslands,and divided these transit wind speeds at intervals of 2.0 m/s to analyze the effect of the transit wind speed on the relationship among Z0,u*,and wind speed within the grassland canopy.The results showed that dividing the transit wind speeds into intervals has a positive effect on the logarithmic fit of the wind speed profile.After dividing the transit wind speeds into intervals,the wind speed at 0.1 m height(V0.1)gradually decreased with the increase of Z0,exhibiting three distinct stages:a sharp change zone,a steady change zone,and a flat zone;while the overall trend of u*increased first and then decreased with the increase of V0.1.Dividing the transit wind speeds into intervals improved the fitting relationship between Z0 and V0.1 and changed their fitting functions in grazed and grazing-prohibited grasslands.According to the computational fluid dynamic results,we found that the number of tall-stature plants has a more significant effect on windproof capacity than their height.The results of this study contribute to a better understanding of the relationship between wind speed and the aerodynamic impedance of vegetation in grassland environments.

Bi-LSTM-Based Deep Stacked Sequence-to-Sequence Autoencoder for Forecasting Solar Irradiation and Wind Speed

Wind and solar energy are two popular forms of renewable energy used in microgrids and facilitating the transition towards net-zero carbon emissions by 2050.However,they are exceedingly unpredictable since they rely highly on weather and atmospheric conditions.In microgrids,smart energy management systems,such as integrated demand response programs,are permanently established on a step-ahead basis,which means that accu-rate forecasting of wind speed and solar irradiance intervals is becoming increasingly crucial to the optimal operation and planning of microgrids.With this in mind,a novel“bidirectional long short-term memory network”(Bi-LSTM)-based,deep stacked,sequence-to-sequence autoencoder(S2SAE)forecasting model for predicting short-term solar irradiation and wind speed was developed and evaluated in MATLAB.To create a deep stacked S2SAE prediction model,a deep Bi-LSTM-based encoder and decoder are stacked on top of one another to reduce the dimension of the input sequence,extract its features,and then reconstruct it to produce the forecasts.Hyperparameters of the proposed deep stacked S2SAE forecasting model were optimized using the Bayesian optimization algorithm.Moreover,the forecasting performance of the proposed Bi-LSTM-based deep stacked S2SAE model was compared to three other deep,and shallow stacked S2SAEs,i.e.,the LSTM-based deep stacked S2SAE model,gated recurrent unit-based deep stacked S2SAE model,and Bi-LSTM-based shallow stacked S2SAE model.All these models were also optimized and modeled in MATLAB.The results simulated based on actual data confirmed that the proposed model outperformed the alternatives by achieving an accuracy of up to 99.7%,which evidenced the high reliability of the proposed forecasting.

Effect of photovoltaic panel electric field on the wind speed required for dust removal from the panels

Methods to remove dust deposits by high-speed airflow have significant potential applications,with optimal design of flow velocity being the core technology.In this paper,we discuss the wind speed required for particle removal from photovoltaic(PV)panels by compressed air by analyzing the force exerted on the dust deposited on inclined photovoltaic panels,which also included different electrification mechanisms of dust while it is in contact with the PV panel.The results show that the effect of the particle charging mechanism in the electric field generated by the PV panel is greatly smaller than the effect of the Van der Waals force and gravity,but the effect of the particle charged by the contact electrification mechanism in the electrostatic field is very pronounced.The wind speed required for dust removal from the PV panel increases linearly with the PV panel electric field,so we suggest that the nighttime,when the PV electric field is relatively small,would be more appropriate time for dust removal.The above results are of great scientific importance for accurately grasping the dust distribution law and for achieving scientific removal of dust on PV panels.

Performance of the CMA-GD Model in Predicting Wind Speed at Wind Farms in Hubei, China

This study assesses the predictive capabilities of the CMA-GD model for wind speed prediction in two wind farms located in Hubei Province,China.The observed wind speeds at the height of 70m in wind turbines of two wind farms in Suizhou serve as the actual observation data for comparison and testing.At the same time,the wind speed predicted by the EC model is also included for comparative analysis.The results indicate that the CMA-GD model performs better than the EC model in Wind Farm A.The CMA-GD model exhibits a monthly average correlation coefficient of 0.56,root mean square error of 2.72 m s^(-1),and average absolute error of 2.11 m s^(-1).In contrast,the EC model shows a monthly average correlation coefficient of 0.51,root mean square error of 2.83 m s^(-1),and average absolute error of 2.21 m s^(-1).Conversely,in Wind Farm B,the EC model outperforms the CMA-GD model.The CMA-GD model achieves a monthly average correlation coefficient of 0.55,root mean square error of 2.61 m s^(-1),and average absolute error of 2.13 m s^(-1).By contrast,the EC model displays a monthly average correlation coefficient of 0.63,root mean square error of 2.04 m s^(-1),and average absolute error of 1.67 m s^(-1).

Short‐time wind speed prediction based on Legendre multi‐wavelet neural network

As one of the most widespread renewable energy sources,wind energy is now an important part of the power system.Accurate and appropriate wind speed forecasting has an essential impact on wind energy utilisation.However,due to the stochastic and un-certain nature of wind energy,more accurate forecasting is necessary for its more stable and safer utilisation.This paper proposes a Legendre multiwavelet‐based neural network model for non‐linear wind speed prediction.It combines the excellent properties of Legendre multi‐wavelets with the self‐learning capability of neural networks,which has rigorous mathematical theory support.It learns input‐output data pairs and shares weights within divided subintervals,which can greatly reduce computing costs.We explore the effectiveness of Legendre multi‐wavelets as an activation function.Mean-while,it is successfully being applied to wind speed prediction.In addition,the appli-cation of Legendre multi‐wavelet neural networks in a hybrid model in decomposition‐reconstruction mode to wind speed prediction problems is also discussed.Numerical results on real data sets show that the proposed model is able to achieve optimal per-formance and high prediction accuracy.In particular,the model shows a more stable performance in multi‐step prediction,illustrating its superiority.

Short-Term Wind Power Prediction Based on ICEEMDAN-SE-LSTM Neural Network Model with Classifying Seasonal

Wind power prediction is very important for the economic dispatching of power systems containing wind power.In this work,a novel short-term wind power prediction method based on improved complete ensemble empirical mode decomposition with adaptive noise(ICEEMDAN)and(long short-term memory)LSTM neural network is proposed and studied.First,the original data is prepossessed including removing outliers and filling in the gaps.Then,the random forest algorithm is used to sort the importance of each meteorological factor and determine the input climate characteristics of the forecast model.In addition,this study conducts seasonal classification of the annual data where ICEEMDAN is adopted to divide the original wind power sequence into numerous modal components according to different seasons.On this basis,sample entropy is used to calculate the complexity of each component and reconstruct them into trend components,oscillation components,and random components.Then,these three components are input into the LSTM neural network,respectively.Combined with the predicted values of the three components,the overall power prediction results are obtained.The simulation shows that ICEEMDAN-SE-LSTM achieves higher prediction accuracy ranging from 1.57%to 9.46%than other traditional models,which indicates the reliability and effectiveness of the proposed method for power prediction.

Deep Learning for Wind Speed Forecasting Using Bi-LSTM with Selected Features

Wind speed forecasting is important for wind energy forecasting.In the modern era,the increase in energy demand can be managed effectively by fore-casting the wind speed accurately.The main objective of this research is to improve the performance of wind speed forecasting by handling uncertainty,the curse of dimensionality,overfitting and non-linearity issues.The curse of dimensionality and overfitting issues are handled by using Boruta feature selec-tion.The uncertainty and the non-linearity issues are addressed by using the deep learning based Bi-directional Long Short Term Memory(Bi-LSTM).In this paper,Bi-LSTM with Boruta feature selection named BFS-Bi-LSTM is proposed to improve the performance of wind speed forecasting.The model identifies relevant features for wind speed forecasting from the meteorological features using Boruta wrapper feature selection(BFS).Followed by Bi-LSTM predicts the wind speed by considering the wind speed from the past and future time steps.The proposed BFS-Bi-LSTM model is compared against Multilayer perceptron(MLP),MLP with Boruta(BFS-MLP),Long Short Term Memory(LSTM),LSTM with Boruta(BFS-LSTM)and Bi-LSTM in terms of Root Mean Square Error(RMSE),Mean Absolute Error(MAE),Mean Square Error(MSE)and R2.The BFS-Bi-LSTM surpassed other models by producing RMSE of 0.784,MAE of 0.530,MSE of 0.615 and R2 of 0.8766.The experimental result shows that the BFS-Bi-LSTM produced better forecasting results compared to others.

Short-term forecasting optimization algorithms for wind speed along Qinghai-Tibet railway based on different intelligent modeling theories

To protect trains against strong cross-wind along Qinghai-Tibet railway, a strong wind speed monitoring and warning system was developed. And to obtain high-precision wind speed short-term forecasting values for the system to make more accurate scheduling decision, two optimization algorithms were proposed. Using them to make calculative examples for actual wind speed time series from the 18th meteorological station, the results show that: the optimization algorithm based on wavelet analysis method and improved time series analysis method can attain high-precision multi-step forecasting values, the mean relative errors of one-step, three-step, five-step and ten-step forecasting are only 0.30%, 0.75%, 1.15% and 1.65%, respectively. The optimization algorithm based on wavelet analysis method and Kalman time series analysis method can obtain high-precision one-step forecasting values, the mean relative error of one-step forecasting is reduced by 61.67% to 0.115%. The two optimization algorithms both maintain the modeling simple character, and can attain prediction explicit equations after modeling calculation.

Variation in cross-shore wind speeds influenced by the morphology of nourished beach

The cross-shore variation in wind speeds influenced by beach nourishment,especially the dramatic changes at the nourished berm,is important for understanding the aeolian sand transport processes that occur after beach nourishment,which will contribute to better beach nourishment project design on windy coasts.In this paper,the influencing factors and potential mechanism of wind speed variation at the edge of a nourished berm were studied.Field observations,together with the Duna model,were used to study the cross-shore wind speed distribution for different nourishment schemes.The results show that the nourished berm elevation and beachface slope are the main factors controlling the increase in wind speed at the berm edge.When the upper beach slope is constant,the wind speed at the berm edge has a positive linear correlation with the berm elevation.When the berm elevation remains constant,the wind speed at the berm edge is also proportional to the upper beach slope.Considering the coupling effects of nourished berm elevation and beachface slope,a model for predicting the wind speed amplification rate at the nourished berm edge was established,and the underlying coupling mechanism was illustrated.

Joint Return Value Estimation of Significant Wave Heights and Wind Speeds with Bivariate Copulas

The joint design criteria of significant wave heights and wind speeds are quite important for the structural reliability of fixed offshore platforms.However,the design method that regards different ocean environmental variables as independent is conservative.In the present study,we introduce a bivariate sample consisting of the maximum wave heights and concomitant wind speeds of the threshold by using the peak-over-threshold and declustering methods.After selecting the appropriate bivariate copulas and univariate distributions and blocking the sample into years,the bivariate compound distribution of annual extreme wave heights and concomitant wind speeds is constructed.Two joint design criteria,namely,the joint probability density method and the conditional probability method,are applied to obtain the joint return values of significant wave heights and wind speeds.Results show that(28.5±0.5)m s^(-1)is the frequently obtained wind speed based on the Atlantic dataset,and these joint design values are more appropriate than those calculated by univariate analysis in the fatigue design.

The Influence of Sea Sprays on Drag Coefficient at High Wind Speed

Field and laboratory observations indicate that the variation of drag coefficient with wind speed at high winds is different from that under low-to-moderate winds.By taking the effects of wave development and sea spray into account,a new parameterization of drag coefficient applicable from low to extreme winds is proposed.It is shown that,under low-to-moderate wind conditions so that the sea spray effects could be neglected,the nondimensional aerodynamic roughness first increases and then decreases with the increasing wave age;whereas under high wind conditions,the drag coefficient decreases with the increasing wind speed due to the modification of the logarithmic wind profile by the effect of sea spray droplets produced by bursting bubbles or wind tearing breaking wave crests.The drag coefficients and sea surface aerodynamic roughnesses reach their maximum values vary under different wave developments.Correspondingly,the reduction of drag coefficient under high winds reduces the increasing rate of friction velocity with increasing wind speed.