Search

Satellite cloud image data plays an important role in the ultra-short-term photovoltaic (PV) power forecasting. By calculating the cloud displacement vectors of two adjacent cloud images, it is possible to calculate the cloud cluster's occlusion of the PV power plant for a period of time in the future, thereby forecasting the PV power. The satellite cloud image data provided by FY-4A Satellite has large and unequal sampling time intervals, which brings difficulties to cloud displacement vector calculation. This paper proposes a method for calculating the displacement vector of the cloud edge block based on the cosine similarity and the adaptive adjustment of the matching spatial scale. The parameter equation is used to establish the relationship between the relative position of the cloud cluster and the PV plant over time, and to predict the future position. On this basis, this paper proposes an ultra-short-term forecasting method of PV power based on cloud displacement vector. The performance of the proposed model is verified through case analysis.

With the increasing of wind power installed capacity, the fluctuation and uncertainty of wind bring huge risks and challenges to the safe and steady operation of power system, while ultra-short-term wind power forecasting is an effective mitigation measure. In the last ten years, a multitude of studies has grown up around the theme of prediction accuracy, while most of them have ignored the effect of input time step on the prediction results, which makes the prediction accuracy have certain geographical limitations. For the sake of solving these problems and ensure that prediction model can make a more accurate judgment on the changes in wind power output, an iterative dynamic integration weighting method is proposed. First, the input step size of the rolling forecast is selected to determine the input with promising performance. Then, compared with other general models by using several error indicators. We have carried out several sets of experiments to verify the availability of this model. The consequence of the experiment demonstrate that the proposed model has better performance, especially in the case of rapid fluctuations of wind power.

Regional photovoltaic (PV) power forecasting provides a foundation for grid management and trading in the power markets. To tackle the deficiency of conventional regional PV power modeling methods, such as the problem of challenging to select useful meteorological information and the problem that a single model cannot fully learn the complex and diverse fluctuation characteristics of power curves, an ultra-short-term regional PV power forecasting framework assembled by fusing fluctuation pattern recognition (FPR) and deep learning modeling under a fluctuation pattern prediction (FPP) model was proposed. First of all, the fluctuation characteristics of the regional PV power curves were extracted, including sampling loss area, mean value, standard deviation, and third derivative. Then, a K-means algorithm based FPR model was established to obtained three kinds of patterns. Moreover, each pattern of PV data was trained by long-short-term memory network (LSTM) for different forecasting models, separately. Finally, the FPP model based on convolutional neural network (CNN) was proposed to predict the fluctuation pattern of the future PV power with historical satellite images as input. Simulation results show that the proposed forecasting method is more accurate than other traditional models.

The randomness and volatility of wind speed will weaken the power system's stability after the wind power is connected to the grid. It is prone to deviations from voltage and frequency. Therefore, it is imperative to predict wind speed. However, many factors affect wind speed (such as temperature, air pressure, terrain, etc.), and the mechanism of action is complicated, so it is challenging to model wind speed directly. Existing research directly uses all training sets to train the model and does not consider the different wind patterns of wind speed time series. In different wind patterns, the wind speed will show different characteristics. Therefore, it is not appropriate to use all historical data to train a single model for predicting wind speed. To achieve better forecasting performance, our forecasting method should be based on wind patterns. Current research focuses on the wind speed prediction of a single wind farm, ignoring the temporal and spatial correlation with nearby wind farms. There is no wind speed prediction method of wind farm cluster. The measurement data of wind farm cluster is not fully utilized. Combined with the graph neural network of deep learning in recent years, we propose a pattern-based wind speed prediction method of wind farm cluster. Using the wind speed and wind direction measurement data of many wind farms in Northwest China, the graph structure data were constructed. The graph neural network model is established for training. Compared with other methods, better wind speed prediction results are obtained.

With the increasing proportion of wind power generation in the power grid, the randomness and volatility of wind power generation have a more and more serious impact on the power grid operation. Therefore, accurate wind power forecasting is an essential guarantee for the power system’s safe and stable operation. Wind power is mainly affected by wind speed, so precise forecasting of wind speed is imperative. Given the insufficient use of information by current wind speed forecasting methods, this paper makes full use of information in two dimensions of time and space. It proposes a method of deep learning based visualized wind speed matrix forecasting model for wind power forecasting. First, the wind speed matrix is formed by extracting the wind speed data under different coordinates. Secondly, the wind speed matrix is converted into a gray image of wind speed by a visualization method. The mapping relationship between the pixels in the gray image and the actual wind speed is established. The convolutional neural network (CNN) algorithm inputs a series of wind speed images and makes forecasting to obtain the forecasted wind speed images and uses the mapping relationship between pixels and wind speed to get the forecasted wind speed data. Finally, the results are compared with the Long short-term memory (LSTM) method. The simulation results show that the method proposed in this paper has higher forecasting accuracy and better robustness in the period when wind speed fluctuates wildly.