Two-Stage Attention Over LSTM With Bayesian Optimization for Day-Ahead Solar Power Forecasting

Abstract

The penetration of PVs into the power grid is increasing day by day, as they are more economical and environment-friendly. However, due to the intrinsic intermittency in solar radiation and other meteorological factors, the generated power from PVs is uncertain and unstable. Therefore, accurate forecasting of power generation is considered one of the fundamental challenges in power system. In this paper, a deep-learning model based on two-stage attention mechanism over LSTM is proposed to forecast a day-ahead PV power. In addition, the Bayesian optimization algorithm is applied to obtain the optimal combination of hyper-parameters for the proposed deep-learning model. Various input features that can affect the PV power generation such as solar radiation, temperature, humidity, snowfall, albedo etc. are considered and their impact with respect to the attention mechanisms on the forecasted PV power is analyzed. The input consists of data from 21 PVs installed at different geographical locations in Germany. The proposed model is compared with state-of-the-art models such as LSTM-Attention, CNN-LSTM, and Ensemble model for day-ahead forecasting. The model is also compared with various single attention mechanisms such as Inputattention, SNAIL, Raffel, and Hierarchical attention etc. The proposed model outperforms the traditional methods in terms of accuracy, hence proving its efficiency. Forecasting Skill (FS) score of the proposed model is 0.4813 whereas 0.4427 is for the Ensemble model, which is the best among other state-of-the-art models. Root Mean Square (RMSE) and Mean Absolute Error (MAE) of the proposed model is 0.0638 and 0.0324 respectively, whereas those of the Ensemble model are 0.0685 and 0.0369 respectively.

Prediction of antimicrobial minimal inhibitory concentrations for Neisseria gonorrhoeae using machine learning models

“… ( Aslam et al, 2021 )where o i and pi are the observed and predicted MIC, respectively, and n indicates the number of datasets.…”

Artificial Intelligence for Antimicrobial Resistance Prediction: Challenges and Opportunities towards Practical Implementation

“…DL/ML models are state-of-the-art tools that predict and interpret AMR [ 28 ]. These models map input features to the target labels in non-linear relationships [ 29 ]. The objective is to do regression or classification, or in some cases interpretation of the outcomes [ 28 , 30 ].…”

“…K-mers are also assigned labels with respective phenotypes and perform the encoding [ 32 ]. Data pre-processing and encoding and feature extraction all can easily be implemented using different Python packages [ 29 ]. Different machine-learning and statistical tools can also be used to generate important features.…”

“…Similarly, the authors in [ 30 , 39 ] used a deep-learning model, which consists of layers of artificial neurons mimicking the human brain [ 39 ]. LR, RF and SVMs are implemented by the scikit-learn Python library, whereas CNN and other deep-learning architectures can be applied with TensorFlow and Python [ 29 ]. The basic idea of all these models is to generate a mathematical relationship between input features and target labels based on available data.…”

An Ensemble Neural Network Based on Variational Mode Decomposition and an Improved Sparrow Search Algorithm for Wind and Solar Power Forecasting

“…Due to great advantages in solving complex nonlinear problems, artificial intelligence models have been favored by many researchers. Aslam et al proposed a deep-learning model based on a two-stage attention mechanism over long short-term memory networks (LSTM) for forecasting day-ahead solar power [10]. Although the introduction of an attention mechanism can improve the prediction performance of the neural network, its calculation is complicated, which may cause the model to converge slowly.…”