Ultra-short-term photovoltaic power prediction method

Abstract

The invention relates to an ultra-short-term photovoltaic power prediction method. The method includes using a longitude and latitude correction method for correcting a fisheye cloud picture; quicklyextracting feature points from the foundation cloud atlas sequence by adopting an HSVSURF algorithm; quickly matching and correcting the feature point pairs, and extracting and predicting a cloud cluster motion track; finely extracting the cloud cluster by using an improved threshold segmentation method, and proposing an irradiation coefficient for visually representing the irradiance condition atthe next moment; establishing an ultra-short-term photovoltaic power prediction model based on the improved IAM-CNN-LSTM hybrid neural network; initializing the weight of the hybrid neural network, and setting the maximum number of iterations; establishing a convolutional neural network, and inputting the feature matrix into the model; optimizing a network structure by using a control variable method; if the maximum number of iterations is reached, iterating to stop outputting the network parameters; and performing ultra-short-term power prediction by using the trained composite network to obtain predicted power of the to-be-predicted time point. Compared with the prior art, the invention has the advantages of improving the prediction precision and the like.

Description

technical field [0001] The invention relates to the technical field of ultra-short-term photovoltaic power forecasting for centralized photovoltaic power plants, in particular to an ultra-short-term photovoltaic power forecasting method. Background technique [0002] With the continuous improvement of solar energy development and utilization, the proportion of photovoltaics connected to the grid is increasing. Due to the instability of large-scale photovoltaic output, its grid connection is likely to cause fluctuations in grid voltage, current and frequency, affecting the power quality of the grid. In order to eliminate the above adverse effects, it is particularly important to improve the prediction accuracy of photovoltaic power. Accurate centralized photovoltaic power prediction is an important means to improve the stability of power system operation and the ability to accommodate photovoltaic power. The power generation of centralized photovoltaic power plants shows str...

AI Technical Summary

Problems solved by technology

In terms of forecasting methods, the main methods used in the prior art are: Propose an ultra-short-term photovoltaic power forecasting method based on variational mode decomposition combined with deep echo state network hybrid model to predict photovoltaic power, but its forecasting steps and network are complex is too high, leading to multi-feature sample sets and forecasting efficiency under different scenarios to be improved; using a hybrid model short-term load forecasting method based on convolutional neural network (CNN) and long-short-term memory network (LSTM) can improve forecasting to a certain extent Accuracy, but this method simply combines CNN and LSTM, which destroys the time structure of the feature matrix and weakens the inherent correlation between each feature time series, resulting in poor overall prediction accuracy.

Images