Quantifying short-term uncertainty in wind power plays a crucial role in power system decision-making. In recent years, the scenario generation community has conducted numerous studies employing generative models. Among these generative models, diffusion models have shown remarkable capabilities with excellent posterior representation. However, diffusion models are seldom used to quantify renewable energy uncertainty. To fill this research gap, this manuscript proposes a novel conditional latent diffusion model (CLDM) adapted for short-term scenario generation. CLDM decomposes the wind power scenario generation task into deterministic forecasting and forecast error scenario generation. The embedding network is used to regress deterministic forecasts, which reduces the denoising complexity of diffusion models. The denoising network generates forecast error scenarios in a latent space. Subsequently, the wind power scenarios are reconstructed by combining deterministic forecasts and forecast error scenarios. The case study compares with existing state-of-the-art methods, CLDM demonstrates superior evaluation metrics and enhances the denoising efficiency.

The rapidly increasing wind power penetration presents new challenges to the operation of power systems. Improving the accuracy of wind power forecasting is a possible solution under this circumstance. In the power forecasting of multiple wind farms, determining the spatio-temporal correlation of multiple wind farms is critical for improving the forecasting accuracy. This paper proposes a spatio-temporal convolutional network (STCN) that utilizes a directed graph convolutional structure. A temporal convolutional network is also adopted to characterize the temporal features of wind power. Historical data from 15 wind farms in Australia are used in the case study. The forecasting results show that the proposed model has higher accuracy than the existing methods. Based on the structure of the STCN, asymmetric spatial correlation at different temporal scales can be observed, which shows the effectiveness of the proposed model.

Power systems with high penetration of renewable energy contain various uncertainties. Scenario-based optimization problems need a large number of discrete scenarios to obtain a reliable approximation for the probabilistic model. It is important to choose typical scenarios and ease the computational burden. This paper presents a scenario reduction network model based on Wasserstein distance. Entropy regularization is used to transform the scenario reduction problem into an unconstrained problem. Through an explicit neural network structure design, the output of the scenario reduction network corresponds to the Sinkhorn distance function. The scenario reduction network can generate the typical scenario set through unsupervised learning. An efficient algorithm is proposed for continuous/discrete scenario reduction. The superiority of the scenario reduction network model is verified through case studies. The numerical results highlight the high accuracy and computational efficiency of the proposed model over state-of-the-art models, and it makes an ideal candidate for large-scale scenario reduction problems.

The extreme power output scenarios of renewable energy sources (RES) proposed new challenges to the safe and stable operation of the power system. Transmission expansion planning (TEP) with large‐scale RES grid integration needs considering the risk of extreme scenarios. In this paper, an adaptive decision‐making approach for the TEP problem based on planning‐risk assessment‐replanning iterative process is proposed. The method obtains massive temporal and spatial correlated wind‐photovoltaic (PV) power output scenarios by generalizing the historical data to describe the uncertainties. A data‐driven load loss risk assessment model (RAM) based on the power system's actual operating state is built, referring to the degree of extreme scenario risks on the balance of supply and demand, and the probability of extreme scenario occurrence. The planning decision is progressively revised according to the risk assessment result. The Garver's 6‐bus system and the IEEE RTS 24‐bus system are adopted as simulation cases. The results show that the optimal expansion plans achieve a balance between the economy and robustness, which verifies the effectiveness of the proposed method.

As the penetration rate of distributed photovoltaics (PVs) increases, behind-the-meter (PVs) amplify the impact of meteorological factors on net load predictions. Public weather forecasts (PWFs) can serve as an alternative meteorological covariate to ensure prediction accuracy when numerical weather prediction (NWP) data are unavailable owing to high costs or data delays. PWFs are freely accessible to the public and significantly influence resident behavior. However, they provide limited information and have inherent heterogeneity with historical net load data in terms of data structure and features, posing challenges in feature extraction and fusion. To address these challenges, this study proposes a day-ahead net load prediction model cross-modal contrastive learning network (CCL-Net) that incorporates PWFs. The CCL module employs contrastive learning to minimize the gap between the representations of temporal meteorological data and PWF features, enriching the semantic features of PWFs. Considering the heterogeneity in features and data structure between future weather forecasts and historical net load data, CCL-Net incorporates a cross-modal fusion block to fuse both features, enhancing prediction accuracy. The proposed model is compared with benchmark methods, demonstrating superior performance. In addition, the analysis and ablation experiments of cross-modal technologies validate the rationality and superiority of the proposed approach.