Scenario Generation for Wind Power Using Improved Generative Adversarial Networks

Jiang, Congmei; Mao, Yongfang; Chai, Yi; Yu, Mi; Tao, Songbing

doi:10.1109/access.2018.2875936

“…The time-series modeling of renewable energy sources can be trained using step 1. The network structure we used was based on our previous work [21]. In this system, a generator with three deconvolutional layers is used to find a function that transforms a well-defined noise distribution z to generate realistic time series.…”

Section: B Training Algorithmmentioning

confidence: 99%

“…Bayesian information has also been incorporated into these GANs to produce scenarios with different variances and mean values that capture different salient modes in the data [20]. In a previous work, we utilized an improved GAN to generate wind-power scenarios by applying alternative training techniques to improve the performance of the models [21]. Additionally, VAEs have been used for the generation of renewable scenarios [22,23].…”

mentioning

confidence: 99%

Day-ahead renewable scenario forecasts based on generative adversarial networks

Jiang¹,

Mao²,

Chai³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

<p>With the increasing penetration of renewable resources such as wind and solar, the operation and planning of power systems, especially in terms of large-scale integration, are faced with great risks due to the inherent stochasticity of natural resources. Although this uncertainty can be anticipated, the timing, magnitude, and duration of fluctuations cannot be predicted accurately. In addition, the outputs of renewable power sources are correlated in space and time, and this brings further challenges for predicting the characteristics of their future behavior. To address these issues, this paper describes an unsupervised method for renewable scenario forecasts that considers spatiotemporal correlations based on generative adversarial networks (GANs), which have been shown to generate high-quality samples. We first utilized an improved GAN to learn unknown data distributions and model the dynamic processes of renewable resources. We then generated a large number of forecasted scenarios using stochastic constrained optimization. For validation, we used power-generation data from the National Renewable Energy Laboratory wind and solar integration datasets. The experimental results validated the effectiveness of our proposed method and indicated that it has significant potential in renewable scenario analysis.</p>

show abstract

“…The spatial dependence is imperative for joint uncertainty modeling, especially for power flow optimizations and transmission risk assessments. To overcome above issues, based on the previous research in [21], [23] and [26], we proposed a novelty method to generate trajectories for uncertainty forecasting of renewable power generation. Fig.…”

Section: Introductionmentioning

confidence: 99%

“…Besides, the proposed deep neural networks are less prone to overfitting for cases where there is an insufficient amount of training data. More details can be found in the previous work in [21], [23] and [26].…”

Section: Introductionmentioning

confidence: 99%

Day-ahead renewable scenario forecasts based on generative adversarial networks

Jiang¹,

Mao²,

Chai³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

<p>With the increasing penetration of renewable resources such as wind and solar, the operation and planning of power systems, especially in terms of large-scale integration, are faced with great risks due to the inherent stochasticity of natural resources. Although this uncertainty can be anticipated, the timing, magnitude, and duration of fluctuations cannot be predicted accurately. In addition, the outputs of renewable power sources are correlated in space and time, and this brings further challenges for predicting the characteristics of their future behavior. To address these issues, this paper describes an unsupervised method for renewable scenario forecasts that considers spatiotemporal correlations based on generative adversarial networks (GANs), which have been shown to generate high-quality samples. We first utilized an improved GAN to learn unknown data distributions and model the dynamic processes of renewable resources. We then generated a large number of forecasted scenarios using stochastic constrained optimization. For validation, we used power-generation data from the National Renewable Energy Laboratory wind and solar integration datasets. The experimental results validated the effectiveness of our proposed method and indicated that it has significant potential in renewable scenario analysis.</p>

show abstract

“…Variational inference was also incorporated into GAN by Hu et al to produce wind and solar scenarios that capture different salient characteristics in the data 26 . In a previous work, we utilized an improved GAN to create scenarios for wind power by applying alternative training techniques to enforce Lipschitz constraints on the discriminator 27 . As another popular framework for generative models, VAE learns the data distribution by calculating the mean square error between samples.…”

Section: Introductionmentioning

confidence: 99%

Day‐ahead renewable scenario forecasts based on generative adversarial networks

Jiang

¹

,

Mao

²

,

Chai

³

et al. 2020

Self Cite

View full text Add to dashboard Cite

Summary With the increasing penetration of renewable resources such as wind and solar, especially in terms of large‐scale integration, the operation and planning of power systems are faced with great risks due to the inherent stochasticity of natural resources. Although this uncertainty can be anticipated, the timing, magnitude, and duration of fluctuations cannot be predicted accurately. In addition, the outputs of renewable power sources are correlated in space and time, and this brings further challenges for predicting the characteristics of their future behavior. To address these issues, this paper describes an unsupervised distribution learning method for renewable scenario forecasts that considers spatiotemporal correlation based on generative adversarial network (GAN), which has been shown to generate realistic time series for stochastic processes. We first utilize an improved GAN to learn unknown data distributions and model the dynamic processes of renewable resources. We then generate a large number of forecasted scenarios using stochastic constrained optimization. For validation, we use power generation data from the National Renewable Energy Laboratory wind and solar integration datasets. The simulation results show that the generated trajectories not only reflect the future power generation dynamics, but also correctly capture the temporal, spatial, and fluctuant characteristics of the real power generation processes. The experimental comparisons verify the superiority of the proposed method and indicate that it can reduce at least 50% of the training iterations of the generative model for scenario forecasts.

show abstract

Scenario Generation for Wind Power Using Improved Generative Adversarial Networks

Cited by 83 publications

References 25 publications

Day-ahead renewable scenario forecasts based on generative adversarial networks

Day-ahead renewable scenario forecasts based on generative adversarial networks

Day-ahead renewable scenario forecasts based on generative adversarial networks

Day‐ahead renewable scenario forecasts based on generative adversarial networks

Contact Info

Product

Resources

About