A Method for Predicting Photovoltaic Power Generation Based on Branched Evolutionary Neural Network

Abstract

The invention discloses a method for predicting photovoltaic power generation power based on a branched evolutionary neural network, which includes obtaining time series data of power generation and power influencing factors of a photovoltaic power field; obtaining time series data of power generation and power influencing factors of a photovoltaic power field; establishing A cyclic neural network model, using the training set to train the model; obtaining real power impact factor data, using the cyclic neural network model to predict the power generation value; obtaining the real value corresponding to the predicted power generation value, and continuing to model Incremental training. Aiming at the characteristics of photoelectric power prediction, the present invention proposes to use cyclic neural network for learning, so that the model has stronger predictive ability, and uses particle swarm algorithm to optimize the parameters in the cyclic neural network model, and at the same time adopts the combination of branch evolution and global evolution , making the model more robust and easier to obtain the global optimal value.

Description

technical field [0001] The invention belongs to the field of prediction of photovoltaic power generation, and in particular relates to a method for prediction of photovoltaic power generation based on a branched evolutionary neural network. Background technique [0002] The principle of photovoltaic power generation is the photovoltaic effect. The so-called photovoltaic effect refers to a phenomenon in which light causes a potential difference between different parts of an inhomogeneous semiconductor or a combination of metal and semiconductor. On the one hand, it is a phenomenon in which photons are converted into electrons and light energy is converted into electrical energy; on the other hand, it forms a voltage, and with voltage, if the two are connected, a current loop can be formed. It can be seen that photovoltaic power generation is closely related to the activities of the sun and is greatly affected by the climate environment. For the research on the prediction of ...

AI Technical Summary

Problems solved by technology

The BP neural network can set the number and structure of neurons and the activation function according to the needs, so that it has a very powerful learning and memory ability. However, the BP neural network also has some natural defects, such as the learning speed when training parameters. It may be very slow, or may oscillate, and it is easy to fall into a local optimum during the optimization process. At the same time, the BP neural network lacks the memory function because it has no feedback mechanism.

Especially for such prediction problems that the value obtained at the previous moment will actually affect the predicted value at the next moment, the BP neural network cannot realize this function well.

Images