Application of DLRM Deep Learning Model in Photovoltaic Power Generation Prediction
The predicted value of photovoltaic power generation can serve as a reference for relevant departments to dispatch or reserve electricity.The prediction of photovoltaic power generation has always been a challenge,as photovoltaic power generation has characteristics such as intermittency,randomness,and volatility.The deep learning recommendation model(DLRM)input layer takes continuous features and discrete features as inputs,and uses different feature extraction methods for feature extraction,afterwards,learn the interaction features between features in the middle layer.In the characteristics of photovoltaic power generation prediction models,there are also corresponding continuous and discrete features,such as solar radiation,temperature,rainfall,humidity,etc.,which are continuous features,while weather,season,month,etc.are discrete features.Moreover,the interaction features of some features,such as temperature and weather,weather and season,have practical significance for predicting photovoltaic power generation.Based on this,the model architecture of DLRM was constructed and modified from a model for predicting click through rates to a model for predicting continuous values,in order to achieve the goal of predicting photovoltaic power generation.The model proposed has been validated on actual data and can effectively predict photovoltaic power generation.
DLRM(deep learning recommendation model)deep learningphotovoltaic power generationforecast
万方数据
维普
