Renewable energy power generation and load hybrid prediction method

Abstract

The invention discloses a renewable energy power generation and load hybrid prediction method. Hybrid prediction comprises a prediction model formed by data mining, variational mode decomposition anda self-adaptive evolutionary extreme learning machine. Firstly, clustering features are selected according to the characteristics of power data and meteorological data, and K-means clustering is carried out; the acquired clusters and to-be-predicted day-ahead one-day data are subjected to similarity discrimination, and clusters most related to the current prediction are selected as training samples; the sample sequence is then subjected to variational mode decomposition to obtain multiple sub sequences with different center frequencies; each sub sequence then adopts the self-adaptive evolutionary extreme learning machine for prediction respectively; and the prediction results of the sub sequences are finally stacked to obtain a final prediction result. The method has the advantages of highprediction precision and quick speed.

Description

technical field [0001] The invention belongs to the technical field of new energy power generation and smart grid, and specifically relates to a hybrid prediction method based on data mining, variational mode decomposition and adaptive differential evolution learning machine to predict renewable energy power generation and load. Background technique [0002] With the depletion of traditional fossil energy, renewable energy (such as wind energy and solar energy) is increasingly penetrated into the power system, especially in the microgrid. A microgrid is a small grid that integrates renewable energy such as light energy and wind energy in the form of micro sources, and supplies power to loads. However, due to the randomness, volatility and load uncertainty of renewable energy generation, it is difficult to guarantee the reliability and stability of the microgrid operation process. Therefore, finding an accurate renewable energy generation and load forecasting method is the k...

AI Technical Summary

Problems solved by technology

[0005] In the selection of prediction tools, the generalization ability of the current artificial neural network method is not strong, and it is easy to fall into local minimization; the support vector machine is sensitive to the selection of parameters, and the training speed is slow

Images