Photothermal electric field solar direct normal radiation prediction method based on convolutional neural network

Abstract

The invention discloses a photothermal electric field solar direct normal radiation intensity prediction method based on a convolutional neural network. A DNI prediction method of a CSP electric fieldis designed by using the CNN in deep learning so as to overcome the defects of the traditional prediction method and accurately obtain the prediction value, so that the CSP power station is easy to dispatch, and the impact on the existing power system during new energy power generation grid connection is further reduced. Firstly, the characteristics of direct normal radiation of the sun are analyzed, a convolutional neural network is selected according to the obtained characteristics, parameters in the network are modified and debugged, and finally, a prediction method is obtained so as to reduce the negative influence caused when the photo-thermal power station is connected to the power grid. The prediction method can accurately predict the solar direct normal radiation intensity of thephoto-thermal electric field.

Description

technical field [0001] The present invention relates to a direct normal irradiance (Direct Normal Irradiance, DNI) prediction technology of the sun, in particular to a direct normal irradiance prediction technology of the sun based on a convolutional neural network (Convolutional Neural Network, CNN). Background technique [0002] In recent years, a new type of power generation using solar energy—Concentrating Solar Power (CSP) has appeared on the stage of history. With its unique thermal energy storage subsystem, it can quickly adjust the output of the system and reduce the cost of new energy power generation. The characteristics of network time's impact on the existing power system have become a research hotspot at the present stage. Because the CSP power station needs a large amount of solar energy - DNI, it needs to be built in the northwest of my country with the above characteristics. The complex and changeable climatic conditions in Northwest my country have caused c...

AI Technical Summary

Problems solved by technology

However, in the past, when the traditional shallow prediction network was working, it would cause gradient loss and fall into local minimum problems, which reduced the accuracy of the entire network.

Images