Robust Load Forecasting Towards Adversarial Attacks via Bayesian Learning

Electric load forecasting is an essential problem for the power industry, which has a significant impact on power system operation. Currently, deep learning is proved to be an effective tool for load forecasting. However, those learning-based models are vulnerable towards adversarial attacks, which raises concerns about the robustness of load forecasting models. In this study, we propose a Bayesian training method to enhance the robustness of deep learning-based load forecasting models towards adversarial attacks. We theoretically prove that the proposed method can improve the load forecasting robustness against various attacking objectives without compromising the prediction performance. An approximation-based training scheme is applied to reduce the computing burden so as to make the method better applied in practice. The experimental results show that such an approximation still yields higher robustness compared to four recently proposed benchmark robust forecasting methods while maintaining the prediction performance under no attack.

Automated summary

In this study, a Bayesian training method is proposed to enhance the robustness of deep learning-based load forecasting models against adversarial attacks. The experimental results demonstrate that this method maintains good prediction performance under no attack and achieves higher robustness compared to four other benchmark methods.