Short‐term wind power forecasting based on two‐stage attention mechanism

Li, Pengbo; Wang, Xiangwen; Yang, Junjie

doi:10.1049/iet-rpg.2019.0614

Cited by 46 publications

(12 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In summary, the main contributions of this approach are I) Despite the fact that there is numerous research on short-time, medium-time, and long-time wind power prediction [19][20][21][22][23][24][25][26][27][35][36], the only study related to the extremely short-time prediction was presented in [7]. The present study proposing a more efficient prediction method compared to [7].…”

Section: Time Horizon Range Referencesmentioning

confidence: 94%

“…Extremely short-time ∆ ≤ 0.02 Only [7] and [9] Very short-time [27] In the other hand, based on time periods, wind speed/power forecasting methods are classified in Table I to five categories; extremely short-time, very short-time [19], [20], short-time [21]- [23], medium-time [24], [25] and long-time forecasting [26], [27]. The first group of this division is long-time forecasting methods, which usually include forecasting from 1 day to 1 week.…”

Section: Time Horizon Range Referencesmentioning

confidence: 99%

See 1 more Smart Citation

TS fuzzy prediction-based SVC compensation of wind farms flicker: A dual-UKF approach

2021

CSEE JPES

View full text Add to dashboard Cite

Fast fluctuations in wind farm power produce voltage flicker in the network. One way to mitigate the flicker is to place a static VAr compensator (SVC). Due to the operating delay of SVCs, it is essential to predict the wind farm reactive power. Here, a novel fuzzy nonlinear modeling approach is suggested and used in the one-step-ahead prediction of the power characteristics. The base of the developed fuzzy modeling is the Takagi-Sugeno fuzzy representation and a dual-unscented Kalman filter (D-UKF). In other words, a nonlinear TS fuzzy system is trained online via the D-UKF. The forecasted value is used as the SVC's reference signal. A large number of the actual data gathered from a wind farm are used for the performance evaluation. These data are collected in winter and summer for different climate situations. Using the actual data a current source with changing amplitude and phase which are updated every half-cycle is used to model the wind farm. Numerical results, including the flicker indices, confirm the improvement in the SVC's performance.

show abstract

Section: Time Horizon Range Referencesmentioning

confidence: 94%

Section: Time Horizon Range Referencesmentioning

confidence: 99%

TS fuzzy prediction-based SVC compensation of wind farms flicker: A dual-UKF approach

2021

CSEE JPES

View full text Add to dashboard Cite

show abstract

“…Results show that AM can effectively improve the accuracy of load forecasting [26]. Li et al established a wind power forecasting model based on LSTM network two-stage attention mechanism and found that the attention mechanism is extensively employed to weight the input feature and strengthen the trend characteristic [27]. Wang et al proposed a novel short-term load forecasting method based on attention mechanism, rolling update, and bidirectional long short-term memory neural network.…”

Section: Literature Reviewmentioning

confidence: 99%

Short-Term Load Forecasting Using Recurrent Neural Networks With Input Attention Mechanism and Hidden Connection Mechanism

Zhang

2020

IEEE Access

View full text Add to dashboard Cite

Short-term load forecasting is a critical task in the smart grid, which can be used to optimize power deployment and reduce power losses. Recurrent neural networks (RNNs) are the most popular deep learning models for short-term load forecasting. However, despite of achieving better forecasting accuracy than the traditional models, the performance of the existing RNN-based load forecasting approaches is still unsatisfactory for practical usage. Therefore, in this work, we have proposed input attention mechanism (IAM) and hidden connection mechanism (HCM) to greatly enhance the accuracy and efficiency of RNNbased load forecasting models. Specifically, we use IAM to assign the importance weights on input layers, which have better performances in both efficiency and accuracy than traditional attention mechanisms. To further enhance the models' efficiency, HCM is then applied to utilize residual connections to enhance the model's converging speed. We have applied both IAM and HCM on four state-of-the-art RNN implementations, and then conducted extensive experimental studies on two public datasets. Experimental results show that the proposed RNNs with IAM and HCM models achieve much better performances than the state-of-the-art baselines in both accuracy and efficiency. Ablation studies show that both IAM and HCM are essential to achieve such superior performances.

show abstract

“…Forecasting and modelling the time variations of wind farm quantities such as wind speed or wind power is always a hot topic and many studies recently were published in this regard [20][21][22][23][24][25][26]. However none of them were about the extremely short time forecasting or modelling this kind of variations.…”

Section: Introductionmentioning

confidence: 99%

Non stationary‐ARMA flicker model for squirrel cage induction generators based wind farms

Samet

Ketabipour

Bagheri

2021

IET Renewable Power Gen

View full text Add to dashboard Cite

Time varying nature of the wind power is studied previously considering different time intervals from seconds to days. However for power quality problems such as flicker, a model which considers the extremely fast variations is essential. Here by using large number of actual records, a time-varying model is proposed which considers the extremely short-time variations of wind active and reactive powers. The wind farm is modelled as a current source with time varying amplitude and phase which change every 0.01 s. Autoregressive moving-average (ARMA) models are utilized to model the variations and ARMA coefficients are calculated for every record. Same to the actual behaviour, the proposed model is non-stationary as the ARMA order and coefficients are different at every run of the model. The proposed model is confirmed through utilizing several applications which need the power time series with extremely short sampling intervals. The following studies are performed by using the actual data and then the proposed model: power spectral density of active and reactive power variations, instantaneous flicker, short term flicker (Pst), estimating the Pst using the maximum value of the instantaneous flicker, estimation of cumulative Pst for multiple wind turbines, and the impact of SVC on flicker mitigation. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

show abstract

Short‐term wind power forecasting based on two‐stage attention mechanism

Cited by 46 publications

References 20 publications

TS fuzzy prediction-based SVC compensation of wind farms flicker: A dual-UKF approach

TS fuzzy prediction-based SVC compensation of wind farms flicker: A dual-UKF approach

Short-Term Load Forecasting Using Recurrent Neural Networks With Input Attention Mechanism and Hidden Connection Mechanism

Non stationary‐ARMA flicker model for squirrel cage induction generators based wind farms

Contact Info

Product

Resources

About