A Neural Network Wind Speed ​​Prediction Method Based on Time Series Data Analysis

Abstract

The invention discloses a neural network wind speed prediction method based on time series data analysis, which includes acquiring multiple sets of time series data of wind farms; establishing regression models of the meteorological data respectively, and recording meteorological attribute values ​​after regression; establishing BP neural network Network model, using differential operation and firefly hybrid algorithm to train the neural network model in groups first, and then conduct global training; through the latest historical meteorological data, calculate the latest meteorological attribute value after regression; input the latest meteorological attribute value into the neural network, Calculate the predicted wind speed value. The present invention adopts the trend data after the regression to replace the fixed data of a certain date to train the model, utilizes the difference algorithm and the firefly algorithm to improve the BP neural network, and adopts the method of combining group training and global training, so that the present invention is more traditional The prediction accuracy of the method is higher.

Description

technical field [0001] The invention belongs to the field of wind speed prediction of wind farms, and in particular relates to a neural network wind speed prediction method based on time series data analysis. Background technique [0002] Wind power is one of the most important ways that society utilizes renewable energy today. However, in the process of wind power generation, wind energy is affected by natural factors in a complex manner, with very strong random fluctuations, and the output wind power energy also has strong fluctuations. In the process of utilizing wind power, two problems need to be solved , the first is the problem of wind power energy dispatching, the peak-to-valley difference of wind power power is large, and the energy regulation between peak and valley needs to be planned in advance, and the second is the stability of the transmission network. If there is a large disturbance, the carrying capacity and stability of the power grid must also be designed...

AI Technical Summary

Problems solved by technology

However, in the process of wind power generation, wind energy is affected by natural factors in a complex manner, with very strong random fluctuations, and the output wind power energy also has strong fluctuations. In the process of utilizing wind power, two problems need to be solved , the first is the problem of wind power energy dispatching, the peak-to-valley difference of wind power power is large, and the energy regulation between peak and valley needs to be planned in advance, and the second is the stability of the transmission network. large disturbances, the carrying capacity and stability of the power grid must also be designed in advance

Images