Photovoltaic power prediction method and system based on convolutional neural network and meta-learning

Abstract

The invention discloses a photovoltaic power prediction method based on convolutional neural network and meta-learning, which belongs to the technical field of renewable energy development and utilization. By establishing a deep convolutional neural network model, historical photovoltaic power data, historical meteorological data and numerical weather forecast data are used As the input of the model, the photovoltaic power of the day to be predicted is used as the output to form a model training sample; based on the radiation data in the numerical weather forecast, the weather type of the day to be predicted is analyzed, and the training samples of similar days are selected; based on the meta-learning strategy and the training samples of similar days , train the neural network model with eight loss function indicators, output eight prediction results, and realize point prediction and probability prediction of photovoltaic power; the invention also discloses its prediction system. The method of the invention can adapt to photovoltaic power prediction conditions in different seasons and different weathers, has extremely high prediction accuracy, and can effectively improve the operation stability of a photovoltaic grid-connected system.

Description

technical field [0001] The invention belongs to the technical field of renewable energy development and utilization, and in particular relates to a photovoltaic power prediction method and system based on convolutional neural network and meta-learning. Background technique [0002] As a widely distributed and easy-to-obtain renewable resource, solar energy can effectively deal with the current world's resource shortage and environmental pollution problems, so it has been fully utilized and developed in photovoltaic power generation and other fields. However, the output of photovoltaic power generation is affected by the weather and environment, and its output power fluctuates greatly, which has a certain impact on the photovoltaic grid-connected power system and threatens its safe and stable operation. Therefore, photovoltaic power prediction technology has received extensive attention. Among them, the day-ahead photovoltaic power prediction can provide information guidance...

AI Technical Summary

Problems solved by technology

The statistical method only inputs the historical photovoltaic power sequence, so it is more suitable for hourly forecasting, and the accuracy is poor in day-ahead forecasting; the satellite data method, that is, realizes photovoltaic power forecasting by analyzing satellite cloud images and satellite radiation data, but due to its high cost , there are few related research methods; the numerical weather forecast method is to use the weather forecast data such as radiation and temperature as the input of the model, and use the machine learning method to analyze the mapping relationship between the weather forecast data and the photovoltaic power of the day to realize the forecast of photovoltaic power

Images