Search

Accurately forecasting wind and solar power output poses challenges for deeply decarbonized electricity systems. Grid operators must commit resources to provide reserves to ensure reliable operations in the face of forecast errors, a process which can increase fuel consumption and emissions. We apply neural network-based machine learning to expand the usefulness of median point forecast data by creating probabilistic distributions of short-term uncertainty in demand, wind, and solar forecasts that adapt to prevailing grid conditions. Machine learning derived estimates of forecast errors compare favorably to estimates based on incumbent methods. Reserves derived from machine learning are usually smaller than values derived using incumbent methods, which enables fuel savings during most hours. Machine learning reserves are generally larger than incumbent reserves during times of higher forecast error, potentially improving system reliability. Performance is tested using multistage production simulation modeling of the California Independent System Operator system. Machine learning reserves provide production cost and greenhouse gas (GHG) emission reductions of approximately 0.3% relative to historical 2019 requirements. Savings in the 2030 timeframe are highly dependent on battery storage capacity. At lower levels of battery capacity, savings of 0.4% from machine learning reserves are shown. Significant quantities of battery storage are expected to be added to meet California's resource adequacy needs and GHG reduction targets. The addition of these batteries saturates reserve needs and results in minimal within-hour balancing costs in 2030. [ABSTRACT FROM AUTHOR]

Photovoltaics (PV), the primary use of solar energy, is growing rapidly. However, the variable output of PV under changing weather conditions may hinder the large-scale deployment of PV. In this study, we propose a two-stage classification-prediction framework to predict contemporaneous PV power output from sky images (a so-called "nowcast"), and compare it with an end-to-end convolution neural network (CNN). The proposed framework first classifies input images into different sky conditions and then the classified images are sent to specific sub-models for PV output prediction. Two types of classifiers are developed and compared: (1) a CNN-based classifier trained on clear sky index (CSI)-labeled sky images and (2) a physics-based non-parametric classifier based on a threshold of fractional cloudiness of sky images. Different numbers of classification categories are also examined. The results suggest that the cloudiness-based classifier is more suitable than the CSI-based classifier for the framework, and the 3-class classification (i.e., sunny, cloudy, overcast) is found to be the optimal choice. We then fine-tune the cloudiness threshold for the non-parametric classifier and tailor the architecture for each sky-condition-specific sub-model. Under the best design, the proposed framework can achieve a root mean squared error (RMSE) of 2.20 kW (relative to a 30 kW rated PV array) on the test set comprising 18 complete days (9 sunny, RMSE = 0.69 kW; 9 cloudy, RMSE = 3.06 kW). Compared with the end-to-end CNN baseline model, the overall prediction performance can be improved by 6% (7% in sunny and 6% in cloudy), with 6% fewer trainable parameters needed in the architecture. [ABSTRACT FROM AUTHOR]

Cloud movement makes short-term forecasting of solar photovoltaic (PV) panel output challenging. A better PV forecast can realize value for both grid operators and commercial or industrial customers with solar assets. In this study, we build convolutional neural network (CNN) based models to forecast power output from PV panels 15 min into the future. Model inputs are the PV power output history and ground-based sky images for the past 15 min. The key challenge is ensuring that due importance is given to each type of input. We systematically explore 28 methods of "fusing" these heterogeneous inputs in our CNN. These methods of fusion (MoF) belong to 4 families. We also systematically explore the many hyperparameters related to model training and tuning. Limited resources preclude an exhaustive search. We apply a three-stage "funnel" approach instead, wherein we narrow our search to the most promising one of these 28 MoF. We find that a two-step autoregression-CNN MoF has the best performance followed closely by a "mix-in" MoF that performs feature expansion and reduction to give appropriate importance to the two types of inputs. The two-step autoregression-CNN model has a forecast skill (FS) of 17.1% relative to smart persistence on the test set comprising 20 complete days (9 sunny, FS = 22%; 11 cloudy, FS = 16.9%). This optimization results in the improvement of FS from 14.1% for a previously published nonoptimized "baseline" model, a CNN wherein the PV history was simply concatenated to the end of the image-sourced vector obtained after convolution, pooling, and flattening operations. [ABSTRACT FROM AUTHOR]