Photovoltaic power generation big data prediction method based on feature conversion multi-label learning

Abstract

The invention requests to protect a photovoltaic power generation big data prediction method based on feature conversion multi-label learning. The method comprises the following steps: 101, carrying out preprocessing operation on data; 102, dividing training set data and verification set data according to historical data and time; 103, performing feature engineering operation on the historical data of the photovoltaic power station; 104, performing feature selection based on the AUC maximum value on the data set with the constructed features; 105, establishing a plurality of machine learning models, and constructing an algorithm model based on feature conversion multi-label learning; 106, accurately predicting the power generation condition of the photovoltaic panel according to the data of the photovoltaic power station based on an algorithm model of feature conversion multi-label learning. According to the historical data of the photovoltaic power station, whether the power generation of the photovoltaic panel reaches the standard or not every day in the future week is predicted, the performance of the photovoltaic panel body is effectively guaranteed, and therefore data supportand decision support are provided for national power input.

Description

technical field [0001] The invention belongs to the technical fields of machine learning, multi-label learning, feature conversion learning, and big data processing, in particular, photovoltaic power generation big data prediction based on multiple models. Background technique [0002] In recent years, we have implemented the national strategy of technological innovation and development, built a world-class enterprise with global competitiveness, promoted the deep integration of big data, artificial intelligence and traditional business, and at the same time implemented the innovation-driven development strategy, innovated management models, and aggregated innovative resources. To further stimulate the innovative vitality of employees and enhance the core competitiveness of enterprises, according to the "SPIC Big Data Construction Overall Plan" and "SPIC's Promoting Mass Entrepreneurship and Innovation Industry Plan", photovoltaic power stations are becoming more and more imp...

AI Technical Summary

Problems solved by technology

[0004] Most of the existing methods and photovoltaic power generation prediction technologies only focus on modeling meteorological conditions and historical data, while ignoring the influence of photovoltaic panel body performance and actual operating conditions on power generation efficiency, and the existing technologies only use single-label learning (i.e. two Classification) model for training prediction

Therefore, the disadvantage of the existing technology is that the prediction accuracy of power generation cannot be guaranteed due to the neglect of the performance of the photovoltaic panel body and the actual operating conditions.

Images