Day-ahead renewable scenario forecasts based on generative adversarial networks

Abstract

<p>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.<br></p>

Generative Adversarial Networks (GANs)

“…Our approach is iterative and allows the designer control over the degree of attribute presence. Jehanian et al, [9] highlight that a common criticism of generative models is that they simply interpolate between data points failing to generate anything novel. Their results demonstrate that contrary to this, it is possible to achieve distributional shift using their steerability method which performs image transformations such as "zoom" "shift" and "luminance."…”

GAN-Generated Faces Detection: A Survey and New Perspectives

“…have been developed to synthesis and create realistic-looking face images with diversity from random noise input. These GANs can effectively encode rich semantic information in the intermediate features [4] and latent space [31,45,102] for high-quality face image generation. Moreover, these GANs can generate fake face images with various attributes, including various ages, expressions, backgrounds, and viewing angles.…”

VIVE3D: Viewpoint-Independent Video Editing using 3D-Aware GANs

“…Once an appropriate latent space representation of an input image has been recovered, semantic edits can be applied by navigating the latent space manifold surrounding the inverted latent code. Unsupervised techniques attempt to find interesting edits without labeled data [22,24,41,49]. InterfaceGAN [39,40] is a simple and robust supervised technique that is highly recommended for practical applications, and as such we also employ it in our work.…”