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.

Wind speed forecasting based on wavelet decomposition and wavelet neural networks optimized by the Cuckoo search algorithm

Wind speed forecasting is of great importance for wind farm management and plays an important role in grid integration. Wind speed is volatile in nature and therefore it is difficult to predict with a single model. In this study, three hybrid multi-step wind speed forecasting models are developed and compared — with each other and with earlier proposed wind speed forecasting models. The three models are based on wavelet decomposition(WD), the Cuckoo search(CS) optimization algorithm, and a wavelet neural network(WNN). They are referred to as CS-WD-ANN(artificial neural network), CS-WNN, and CS-WD-WNN, respectively. Wind speed data from two wind farms located in Shandong, eastern China, are used in this study. The simulation result indicates that CS-WD-WNN outperforms the other two models, with minimum statistical errors. Comparison with earlier models shows that CS-WD-WNN still performs best, with the smallest statistical errors. The employment of the CS optimization algorithm in the models shows improvement compared with the earlier models.

Wind Speed Forecasting Based on ARMA-ARCH Model in Wind Farms

Wind speed forecasting is signif icant for wind farm planning and power grid operation. The research in this paper uses Eviews software to build the ARMA (autoregressive moving average) model of wind speed time series, and employs Lagrange multipliers to test the ARCH (autoregressive conditional heteroscedasticity) effects of the residuals of the ARMA model. Also, the corresponding ARMA-ARCH models are established, and the wind speed series are forecasted by using the ARMA model and ARMA-ARCH model respectively. The comparison of the forecasting accuracy of the above two models shows that the ARMA-ARCH model possesses higher forecasting accuracy than the ARMA model and has certain practical value.

ELMAN Neural Network with Modified Grey Wolf Optimizer for Enhanced Wind Speed Forecasting

The scope of this paper is to forecast wind speed. Wind speed, temperature, wind direction, relative humidity, precipitation of water content and air pressure are the main factors make the wind speed forecasting as a complex problem and neural network performance is mainly influenced by proper hidden layer neuron units. This paper proposes new criteria for appropriate hidden layer neuron unit’s determination and attempts a novel hybrid method in order to achieve enhanced wind speed forecasting. This paper proposes the following two main innovative contributions 1) both either over fitting or under fitting issues are avoided by means of the proposed new criteria based hidden layer neuron unit’s estimation. 2) ELMAN neural network is optimized through Modified Grey Wolf Optimizer (MGWO). The proposed hybrid method (ELMAN-MGWO) performance, effectiveness is confirmed by means of the comparison between Grey Wolf Optimizer (GWO), Adaptive Gbest-guided Gravitational Search Algorithm (GGSA), Artificial Bee Colony (ABC), Ant Colony Optimization (ACO), Cuckoo Search (CS), Particle Swarm Optimization (PSO), Evolution Strategy (ES), Genetic Algorithm (GA) algorithms, meanwhile proposed new criteria effectiveness and precise are verified comparison with other existing selection criteria. Three real-time wind data sets are utilized in order to analysis the performance of the proposed approach. Simulation results demonstrate that the proposed hybrid method (ELMAN-MGWO) achieve the mean square error AVG ± STD of 4.1379e-11 ± 1.0567e-15, 6.3073e-11 ± 3.5708e-15 and 7.5840e-11 ± 1.1613e-14 respectively for evaluation on three real-time data sets. Hence, the proposed hybrid method is superior, precise, enhance wind speed forecasting than that of other existing methods and robust.

Empirical Mode Decomposition-k Nearest Neighbor Models for Wind Speed Forecasting

Hybrid model is a popular forecasting model in renewable energy related forecasting applications. Wind speed forecasting, as a common application, requires fast and accurate forecasting models. This paper introduces an Empirical Mode Decomposition (EMD) followed by a k Nearest Neighbor (kNN) hybrid model for wind speed forecasting. Two configurations of EMD-kNN are discussed in details: an EMD-kNN-P that applies kNN on each decomposed intrinsic mode function (IMF) and residue for separate modelling and forecasting followed by summation and an EMD-kNN-M that forms a feature vector set from all IMFs and residue followed by a single kNN modelling and forecasting. These two configurations are compared with the persistent model and the conventional kNN model on a wind speed time series dataset from Singapore. The results show that the two EMD-kNN hybrid models have good performance for longer term forecasting and EMD-kNN-M has better performance than EMD-kNN-P for shorter term forecasting.

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.

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.

Multi-model ensemble forecasting of 10-m wind speed over eastern China based on machine learning optimization

风对人类活动和电力运行有重大影响,准确预报短期风速具有深远的社会和经济意义.基于中国东部100个站点,本研究首先评估了5个业务模式对10米风速的预报能力,日本气象厅JMA模式在减少预报误差方面表现最好.进一步,利用5种数值模式和多种机器学习方法,将动力和统计相结合,对每个站点分别进行了特征工程和机器学习算法优选,建立了10米风速多模式集成预报模型。针对24至96小时预报时长,将该方法的预报性能与基于岭回归的多模式集成和JMA单模式进行比较.结果表明,基于机器学习优选的多模型集成方法可以将JMA模式的预报误差降低39%以上,预报效果的提升在11月最明显.此外,该方法优于基于岭回归的多模式集成方法.

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.

Coupling framework for a wind speed forecasting model applied to wind energy

Wind energy is the burgeoning renewable energy. Accurate wind speed prediction is necessary to ensure the stability and reliability of the power grid for wind energy. This study focuses on developing a novel hybrid forecasting model to tackle adverse effects caused by strong variability and abrupt changes in wind speed. The hybrid model combines data decomposition and error correction strategy for a wind speed forecasting model applied to wind energy. First, wavelet packet decomposition is applied to wind speed series to obtain stationary subseries. Next, outlier robust extreme learning machine is implemented to predict subseries. Finally, an error correction strategy coupled with data decomposition is designed to repair preliminary prediction results. In addition, four measured datasets from China and USAwind farms with different time intervals are used to evaluate the performance of the proposed approach. Experimental analysis indicates that the proposed model outperforms the compared models. Results show that(1) the prediction accuracy of the proposed model is remarkably improved compared with other conventional models;(2) the proposed model can reduce the influence of the end effect in the decomposition-based forecasting model;(3) the coupling framework is successful for enhancing performance of hybrid forecasting model.

Bootstrapped Multi-Model Neural-Network Super-Ensembles for Wind Speed and Power Forecasting

The bootstrap resampling method is applied to an ensemble artificial neural network (ANN) approach (which combines machine learning with physical data obtained from a numerical weather prediction model) to provide a multi-ANN model super-ensemble for application to multi-step-ahead forecasting of wind speed and of the associated power generated from a wind turbine. A statistical combination of the individual forecasts from the various ANNs of the super-ensemble is used to construct the best deterministic forecast, as well as the prediction uncertainty interval associated with this forecast. The bootstrapped neural-network methodology is validated using measured wind speed and power data acquired from a wind turbine in an operational wind farm located in northern China.

Improving the Forecasts of Coastal Wind Speeds in Tianjin,China Based on the WRF Model with Machine Learning Algorithms

Characterized by sudden changes in strength,complex influencing factors,and significant impacts,the wind speed in the circum-Bohai Sea area is relatively challenging to forecast.On the western side of Bohai Bay,as the economic center of the circum-Bohai Sea,Tianjin exhibits a high demand for accurate wind forecasting.In this study,three machine learning algorithms were employed and compared as post-processing methods to correct wind speed forecasts by the Weather Research and Forecast(WRF)model for Tianjin.The results showed that the random forest(RF)achieved better performance in improving the forecasts because it substantially reduced the model bias at a lower computing cost,while the support vector machine(SVM)performed slightly worse(especially for stronger winds),but it required an approximately 15 times longer computing time.The back propagation(BP)neural network produced an average forecast significantly closer to the observed forecast but insufficiently reduced the RMSE.In regard to wind speed frequency forecasting,the RF method commendably corrected the forecasts of the frequency of moderate(force 3)wind speeds,while the BP method showed a desirable capability for correcting the forecasts of stronger(force>6)winds.In addition,the 10-m u and v components of wind(u_(10)and v_(10)),2-m relative humidity(RH_(2))and temperature(T_(2)),925-hPa u(u925),sea level pressure(SLP),and 500-hPa temperature(T_(500))were identified as the main factors leading to bias in wind speed forecasting by the WRF model in Tianjin,indicating the importance of local dynamical/thermodynamic processes in regulating the wind speed.This study demonstrates that the combination of numerical models and machine learning techniques has important implications for refined local wind forecasting.

An Experiment on the Prediction of the Surface Wind Speed in Chongli Based on the WRF Model:Evaluation and Calibration

In this study,the ability of the Weather Research and Forecasting(WRF)model to generate accurate near-surface wind speed forecasts at kilometer-to subkilometer-scale resolution along race tracks(RTs)in Chongli during the wintertime is evaluated.The performance of two postprocessing methods,including the decaying-averaging(DA)and analogy-based(AN)methods,is tested to calibrate the near-surface wind speed forecasts.It is found that great uncertainties exist in the model’s raw forecasts of the near-surface wind speed in Chongli.Improvement of the forecast accuracy due to refinement of the horizontal resolution from kilometer to subkilometer scale is limited and not systematic.The RT sites tend to have large bias and centered root mean square error(CRMSE)values and also exhibit notable underestimation of high-wind speeds,notable overestimation or underestimation of the near-surface wind speed at high altitudes,and notable underestimation during daytime.These problems are not resolved by increasing the horizontal resolution and are even exacerbated,which leads to great challenges in the accurate forecasting of the near-surface wind speed in the competition areas in Chongli.The application of postprocessing methods can greatly improve the forecast accuracy of near-surface wind speed.Both methods used in this study have comparable abilities in reducing the(positive or negative)bias,while the AN method is also capable of decreasing the random error reflected by CRMSE.In particular,the large biases for high-wind speeds,wind speeds at high-altitude stations,and wind speeds during the daytime at RT stations can be evidently reduced.

A Literature Review of Wind Forecasting Methods

In this paper, an overview of new and current developments in wind forecasting is given where the focus lies upon principles and practical implementations. High penetration of wind power in the electricity system provides many challenges to the power system operators, mainly due to the unpredictability and variability of wind power generation. Although wind energy may not be dispatched, an accurate forecasting method of wind speed and power generation can help the power system operators reduce the risk of unreliability of electricity supply. This paper gives a literature survey on the categories and major methods of wind forecasting. Based on the assessment of wind speed and power forecasting methods, the future development direction of wind forecasting is proposed.

Very Short-Term Generating Power Forecasting for Wind Power Generators Based on Time Series Analysis

In recent years, there has been introduction of alternative energy sources such as wind energy. However, wind speed is not constant and wind power output is proportional to the cube of the wind speed. In order to control the power output for wind power generators as accurately as possible, a method of wind speed estimation is required. In this paper, a technique considers that wind speed in the order of 1 - 30 seconds is investigated in confirming the validity of the Auto Regressive model (AR), Kalman Filter (KF) and Neural Network (NN) to forecast wind speed. This paper compares the simulation results of the forecast wind speed for the power output forecast of wind power generator by using AR, KF and NN.

基于IBES-XGBoost的高速铁路沿线风速预测模型

为保证高速铁路沿线风灾预警信息具有较高时效性,需要进行高速铁路沿线超短期风速的提前多步预测。针对众多预测模型在预测中可能存在较大误差的问题,采用Tent混沌映射和BFGS拟牛顿法对秃鹰搜索算法进行改进,并用改进的秃鹰搜索算法(IBES)对XGBoost的初始参数进行优化。在构建IBES-XGBoost模型时,加入风速以外的其他气象特征,以提高预测精度。实验结果表明:(1)改进的秃鹰算法相比其他智能优化算法有更好的寻优能力,与其他模型相比IBES-XGBoost在超短期风速的提前多步预测上有着较高的精度和较好的拟合效果。(2)Tent混沌映射和BFGS拟牛顿法对秃鹰算法有着较好的改进效果。(3)IBES-XGBoost能为高速铁路规范下的大风预警提供可靠的提前多步预测结果。

基于ViT和LSTM的风速多步预测

精确的风速预测对风力发电具有指导作用,据此提出一种多维时间序列下Vision Transformer(ViT)和长短时记忆网络(LSTM)的风速预测方法,实现对风速的超前一步和超前多步预测。结合斯皮尔曼系数(Spearman)和变分模态分解将风速分解为多维时间序列,多维时间序列能更好地表征原始风速的周期性和波动性;采用ViT提取多维时间序列中的特征以及隐藏信息,在保持ViT模型自注意力机制优势的同时,用LSTM进一步建立所提取特征和风速之间的关系,从而提高ViT-LSTM模型的泛化性和预测准确性。使用内蒙古某风场记录的数据进行试验分析,在超前1步、超前6步、超前12步和超前24步预测时,该方法的平均绝对误差分别比LSTM模型减少了15.15%、34.41%、68.32%和81.71%,结果表明该模型在超前多步风速预测方面具有较好的效果。

计及误差信息的自适应超短期风速预测模型

为提升超短期风速预测精度,提出一种计及误差信息的自适应混合预测模型。应用自适应噪声的完备集合经验模态分解模型与鲸鱼优化的变分模态分解模型分别对风速样本数据与预测误差进行分解,同时计算各子序列的模糊熵以判断序列复杂程度。在此基础上,应用鲸鱼优化的长短期网络预测复杂程度较高的序列,差分自回归移动平均模型预测复杂程度较低的序列。最后,将初始风速预测结果和风速误差预测值相加得到基于误差修正的超短期风速预测值。结果表明,修正预测误差与考虑分解策略可有效提升点预测的性能,与基准模型相比,所提模型在多场景下均具备优良的预测精度。

Forecasting wind speed using a reinforcement learning hybrid ensemble model:a high-speed railways strong wind signal prediction study in Xinjiang,China

Considering the application of wind-forecasting technology along the railway,it becomes an effective means to reduce the risk of tain more reliable wind-speed prediction results,this study proposes an intelligent ensemble forecasting method for strong winds train derailment and overturning.Accurate prediction of crosswinds can provide scientific guidance for safe train operation.To obalong the high-speed railway.The method consists of three parts:the data preprocessing module,the hybrid prediction module and original wind speed data.Then,Broyden-Fletcher-Goldfarb-Shanno(BFGS)method,non-linear autoregressive network with exoge-the reinforcement learing ensemble module.First,fast ensemble empirical model decomposition(FEEMD)is used to process the prediction models for all the sublayers of decomposition.Finally,Q-learning is utilized to iteratively calculate the combined weights nous inputs(NARX)and deep belief network(DBN),three benchmark predictors with different characteristics are employed to build of the three models,and the prediction results of each sublayer are superimposed to obtain the model output.The real wind speed data of two railway stations in Xinjiang are used for experimental comparison.Experiments show that compared with the single benchmark model,the hybrid ensemble model has better accumacy and robustness for wind speed prediction along the railway.The 1-step forecasting results mean absolute error(MAE),mean absolute percentage error(MAPE)and root mean square error(RMSE)of Q-leaming-FEEMD-BFGS-NARX-DBN in site #1 and site #2 are 0.0894 m/s,0.6509%,0.1146 m/s,and 0.0458 m/s.0.2709%,0.0616 m/s.respectively.The proposed ensemble model is a promising method for railway wind speed prediction.

基于属性约简与加权最优层次聚类的短期风速混合预测

准确的风速预测是提高风功率预测精度的重要保障。为此,提出一种基于互信息(mutualinformation,MI)属性约简与加权最优层次聚类(weighting optimal hierarchy clustering,WOHC)的离群鲁棒极限学习机(outlier robust extreme learning machine,ORELM)风速混合预测方法。首先,计算32维风速属性特征与风速时间序列间的MI,分析不同特征与风速的相关性。在此基础上,分别采用最大相关最小冗余(maximum correlation minimum redundancy,MRMR)算法和WOHC算法实现风速属性特征的约简及风速样本数据的聚类划分,并通过最优化聚类预处理(clusters optimizationonpreprocessingstage,COPS)确定最优聚类数。然后,采用ORELM对不同样本集分别进行训练,构建ORELM风速混合预测模型。计算待预测点约简后的属性特征与每个聚类中心的欧式距离,选择匹配的ORELM模型进行风速预测。最后,结合东北某风电场实测数据对所提预测方法的有效性和准确性进行验证,结果表明所提方法具有较好的预测精度,能够满足实际风电场风速预测的需要。