Electronics, Journal Year: 2025, Volume and Issue: 14(5), P. 887 - 887

Published: Feb. 24, 2025

In recent wind power forecasting studies, deep neural networks have shown powerful performance in estimating future from data. this paper, a pseudo-twin network model of full multi-layer perceptron is proposed for farms. model, the input data are divided into physical attribute and historical These two types processed separately by feature extraction module structure to obtain features features. To ensure comprehensive integration establish connection between extracted features, mixing introduced cross-mix After mixing, set perceptrons used as regression forecast power. simulation research carried out based on measured The compared with mainstream models such CNN, RNN, LSTM, GRU, hybrid network. results show that MAE RMSE single-step reduced up 21.88% 16.85%, respectively. Additionally, 1 h rolling (six steps ahead) 31.58% 40.92%,

Language: Английский