Short-term prediction of wind power and its ramp events based on semi-supervised generative adversarial network

Bin Zhou,Haoran Duan,Qiuwei Wu,Huaizhi Wang,Siu Wing Or,Ka Wing Chan,Yunfan Meng

DOI: https://doi.org/10.1016/j.ijepes.2020.106411

2021-02-01

Abstract:<p>Short-term predictions of wind power and its ramp events play a critical role in economic operation and risk management of smart grid. This paper proposes a hybrid forecasting model based on semi-supervised generative adversarial network (GAN) to solve the short-term wind power outputs and ramp event forecasting problems. In the proposed model, the original time series of wind energy data can be decomposed into several sub-series characterized by intrinsic mode functions (IMFs) with different frequencies, and the semi-supervised regression with label learning is employed for data augmentation to extract non-linear and dynamic behaviors from each IMF. Then, the GAN generative model is used to obtain unlabeled virtual samples for capturing data distribution characteristics of wind power outputs, while the discriminative model is redesigned with a semi-supervised regression layer to perform the point prediction of wind power. These two GAN models form a min-max game so as to improve the sample generation quality and reduce forecasting errors. Moreover, a self-tuning forecasting strategy with multi-label classifier is proposed to facilitate the forecasting of wind power ramp events. Finally, the real data of a wind farm from Belgium is collected in the case study to demonstrate the superior performance of the proposed approach compared with other forecasting algorithms.</p>

engineering, electrical & electronic

What problem does this paper attempt to address?

The paper aims to address the issues of short-term wind power forecasting and ramp events prediction. Specifically, it proposes a hybrid prediction model based on semi-supervised Generative Adversarial Networks (GAN) to improve the accuracy of short-term wind power output and ramp events prediction.

### Main Problems Addressed by the Paper:
1. **Short-term Wind Power Forecasting**: Accurately predicting the output of wind power in the near future.
2. **Ramp Events Prediction in Wind Power**: Identifying and predicting sudden increases or decreases in wind power output within a short period, usually caused by extreme weather conditions.

### Background and Motivation:
- Wind energy, as a clean energy source, has developed rapidly in recent years, but its volatility and intermittency pose challenges to the economic operation, reliability, and real-time control of the power grid.
- Accurate and reliable wind energy forecasting technology is crucial for scheduling renewable energy generation and balancing the demand between large-scale wind farms and the power system.
- Ramp events in wind power refer to situations where wind power output changes dramatically within a short period, requiring mitigation through energy storage technologies and scheduling strategies.

### Main Contributions:
1. Proposes a wind power forecasting method based on the GAN framework, generating virtual samples to address the underfitting problem in traditional GAN training processes.
2. Combines GAN and semi-supervised regression to propose a hybrid point prediction method, utilizing the generative model to expand training samples and improving prediction accuracy through an enhanced discriminative model.
3. Introduces a multi-step self-optimizing prediction strategy, combining rolling time-domain programs to adjust future ramp event prediction results, and uses a multi-label classifier to analyze ramp characteristics in wind power, thereby improving prediction accuracy.