학술논문
A Hybrid Deep Neural Network Model for Photovoltaic Generation Power Prediction
Document Type
Conference
Source
2022 25th International Conference on Electrical Machines and Systems (ICEMS) Electrical Machines and Systems (ICEMS), 2022 25th International Conference on. :1-5 Nov, 2022
Subject
Language
ISSN
2642-5513
Abstract
This paper presents a hybrid deep neural network (DNN) model for predicting the power of a photovoltaic generation (PV) system. The proposed model consists of multilayer architecture by synthesizing a DNN model and a gated recurrent unit (GRU) model. This architecture enhances prediction accuracy by reflecting the nonlinearity and time-series characteristics of the PV power. The performance of the proposed model is verified by comparative simulation with the DNN model and the GRU model. As a simulation result, the proposed model improved the prediction accuracy by up to 98% compared to the DNN model. Therefore, the proposed model can accurately predict the PV power by reflecting the time-series characteristics.