Distributed photovoltaic power generation system short-term prediction method based on LSTM-Morlet model

Abstract

The invention relates to a distributed photovoltaic power generation system short-term prediction method based on an LSTM-Morlet model in the technical field of power system automation. The method comprises the steps of sample data normalization, short-term weather classification and establishment of a photovoltaic power generation prediction model. A photovoltaic power generation prediction modelis established by using a variant long-term and short-term memory neural network LSTM; a Morlet wavelet function is used as an activation function of an LSTM model, and experiments prove that the power generation prediction effect of the LSTM model is obviously improved by using the Morlet wavelet function as the activation function and adding weather index parameters; the method has the advantages of high convergence speed and high prediction precision.

Description

Technical field [0001] The invention relates to the technical field of power system automation, in particular to a short-term prediction method of a distributed photovoltaic power generation system based on an LSTM-Morlet model. Background technique [0002] The power generation of the distributed photovoltaic power generation system is affected by various weather factors, and the weather factors have certain volatility and discontinuity. Therefore, how to construct a suitable prediction model according to the output characteristics of photovoltaic power generation to achieve accurate prediction of power generation , To avoid unreasonable allocation and waste of resources. In order to accurately realize distributed photovoltaic power generation forecasting, relevant scholars at home and abroad have conducted extensive research on photovoltaic output forecasting models. Whether it is to establish a gray dynamic GM model, use the historical raw data of the past five months as the m...

AI Technical Summary

Problems solved by technology

In order to accurately realize the prediction of distributed photovoltaic power generation, relevant scholars at home and abroad have conducted extensive research on the prediction model of photovoltaic output. Whether it is to establish a gray dynamic GM model, using the historical raw data of the past five months as a modeling support, the photovoltaic The prediction of power generation is still based on the prediction model of the adaptive fuzzy time series method, and grid-connected photovoltaic power generation is used for short-term power prediction; the direct prediction method itself needs to discover the laws between historical data through mathematical statistics and then make relevant predictions. Applicability Not very good, and the introduction of artificial intelligence algorithms with higher complexity to improve the direct prediction method will increase the complexity of prediction, and the simple use of neural networks for photovoltaic prediction can only ensure the accuracy of prediction under small samples. Accuracy, the adaptability of many scenarios is also perfect, and it is easy to fall into the disadvantage of local minimum; therefore, it is very necessary to provide a short-term prediction method for distributed photovoltaic power generation systems based on the LSTM-Morlet model with fast convergence speed and high prediction accuracy of

Images