Direct dispersion separation modeling prediction control method under multiple weather types based on GRA-LMBP weights

Abstract

The invention relates to a direct dispersion separation modeling prediction control method under multiple weather types based on GRA-LMBP weights. Firstly, performing statistics and collection on meteorological radiation data of many years in Beijing area, selecting typical meteorological years according to a normal fitting method and using a revised definition index to classify the weather types;secondly, using support vector machine (SVM), gray relational analysis (GRA) based support vector machine GRA-SVM and principal component analysis (PCA) based support vector machine (PCA-SVM) prediction models to obtain prediction results of three single models under each weather type; then using the gray relational analysis to obtain the weight coefficient of each model under each weather type;then fitting an LMBP neural network prediction model according to each single model prediction result and prediction weight coefficient under each weather type; and finally, using the weight coefficients obtained by the LMBP model to calculate the combined prediction model prediction results of each weather type. Compared with the prior art, the direct dispersion separation modeling prediction control method under multiple weather types based on the GRA-LMBP weights has the advantages of high precision, good calculation stability and the like.

Description

technical field [0001] The invention relates to one, in particular to a GRA-LMBP weight-based direct-dispersive separation modeling predictive control method under multi-weather types. Background technique [0002] The National Energy Administration pointed out at a press conference in early 2019 that by the end of 2018, my country's installed capacity of renewable energy power generation had reached 728 million kilowatts, a year-on-year increase of 12%, of which photovoltaic power generation installed capacity was 174 million kilowatts, a year-on-year increase of 34%, ranking first in the world. 1 person. The new installed capacity of photovoltaic power generation in the whole year was 44.26 million kilowatts, second only to the newly installed capacity in 2017, and the second highest in history. In 2019, my country will focus on promoting the high-quality development of photovoltaic power generation, continue to promote the development of the photovoltaic industry, and mai...

AI Technical Summary

Problems solved by technology

[0006] The "straight-scatter separation" forecasting model is divided into daily scale, monthly scale and hourly scale. The current research on models of different scales is applicable to all weather types. Although it is easy to use, the accuracy of the model is relatively low

Images