Photovoltaic Power Forecasting With a Hybrid Deep Learning Approach

Li, Gangqiang; Xie, Siyu; Wang, Bozhong; Xin, Jiantao; Li, Yunfeng; Du, Shengnan

doi:10.1109/access.2020.3025860

Cited by 124 publications

(64 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Speculating again that a higher number of steps prediction would follow the same trend of Figure 16, our approach would obtain a much higher R 2 value. Work [32] uses a direct mode to supply multi-step ahead PV forecasts with a prediction horizon of 3 h, with steps of 15 min. Different models are designed for each season.…”

Section: Discussionmentioning

confidence: 99%

“…Li and co-workers [32] propose a hybrid deep learning approach based on CNN and LSTM for PV output power forecasting. The CNN model is intended to discover the non-linear features and invariant structures exhibited in the previous output power data, thereby facilitating PV power prediction.…”

Section: Literature Reviewmentioning

confidence: 99%

“…This type of forecasts can be achieved in a direct mode, by having several one-step-ahead forecasting models, each providing the prediction of each-step ahead within PH. This is the approach followed in [27,32]. An alternative, which is the one followed in this work, is to use a recursive version.…”

Section: Literature Reviewmentioning

confidence: 99%

See 2 more Smart Citations

Short-Term Forecasting Photovoltaic Solar Power for Home Energy Management Systems

Bot

Ruano

2021

Inventions

View full text Add to dashboard Cite

Accurate photovoltaic (PV) power forecasting is crucial to achieving massive PV integration in several areas, which is needed to successfully reduce or eliminate carbon dioxide from energy sources. This paper deals with short-term multi-step PV power forecasts used in model-based predictive control for home energy management systems. By employing radial basis function (RBFs) artificial neural networks (ANN), designed using a multi-objective genetic algorithm (MOGA) with data selected by an approximate convex-hull algorithm, it is shown that excellent forecasting results can be obtained. Two case studies are used: a special house located in the USA, and the other a typical residential house situated in the south of Portugal. In the latter case, one-step-ahead values for unscaled root mean square error (RMSE), mean relative error (MRE), normalized mean average error (NMAE), mean absolute percentage error (MAPE) and R2 of 0.16, 1.27%, 1.22%, 8% and 0.94 were obtained, respectively. These results compare very favorably with existing alternatives found in the literature.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Literature Reviewmentioning

confidence: 99%

See 1 more Smart Citation

Short-Term Forecasting Photovoltaic Solar Power for Home Energy Management Systems

Bot

Ruano

2021

Inventions

View full text Add to dashboard Cite

show abstract

“…Long short term memory (LSTM) introduces the gate structure based on RNN structure, which realizes the function of selective memory of historical information and solves the problem of long-term dependence of RNN. Li et al [23] proposed a hybrid model based on CNN and LSTM, which used CNN to deeply mine the nonlinear characteristics and invariant structures of data, and then combined LSTM for prediction. Mei et al [24] used quantile regression averaging (QRA) and LSTM integrated model to obtain the probability prediction of PV output.…”

Section: Introductionmentioning

confidence: 99%

Short-Term Photovoltaic Power Forecasting Based on VMD and ISSA-GRU

et al. 2021

View full text Add to dashboard Cite

Photovoltaic (PV) power generation is affected by many meteorological factors and environmental factors, which has obvious intermittent, random, and volatile characteristics. To improve the accuracy of short-term PV power prediction, a hybrid model (VMD-ISSA-GRU) based on variational mode decomposition (VMD), improved sparrow search algorithm (ISSA) and gated recurrent unit (GRU) is proposed. First of all, the PV time series is decomposed into a series of different subsequences by VMD to reduce the non-stationarity of the original data. Then, the main factors affecting PV power generation are obtained by using the correlation coefficients of Spearman and Pearson, which reduces the computational complexity of the model. Finally, the GRU network optimized by ISSA is used to predict all the subsequences and residual error of VMD, and the prediction results are reconstructed. The results show that the hybrid VMD-ISSA-GRU model has stronger adaptability and higher accuracy than other traditional models. The mean absolute error (MAE) in the whole test set is 1.0128 KW, the root mean square error (RMSE) is 1.5511 KW.INDEX TERMS short-term PV power forecast, variational mode decomposition, improved sparrow search algorithm, GRU neural network.

show abstract

“…Research reveals that various ML and DL algorithms have been used for the purpose of RE forecasting [10]. Different assembled AI-based models have been developed to enhance the RE forecast accuracy [17]. To predict RE generation, several time horizons have been investigated such as minutely, hourly, daily, weekly, and monthly depending on the objective of the forecast [18].…”

Section: Introductionmentioning

confidence: 99%

AB-Net: A Novel Deep Learning Assisted Framework for Renewable Energy Generation Forecasting

et al. 2021

View full text Add to dashboard Cite

Renewable energy (RE) power plants are deployed globally because the renewable energy sources (RESs) are sustainable, clean, and environmentally friendly. However, the demand for power increases on a daily basis due to population growth, technology, marketing, and the number of installed industries. This challenge has raised a critical issue of how to intelligently match the power generation with the consumption for efficient energy management. To handle this issue, we propose a novel architecture called ‘AB-Net’: a one-step forecast of RE generation for short-term horizons by incorporating an autoencoder (AE) with bidirectional long short-term memory (BiLSTM). Firstly, the data acquisition step is applied, where the data are acquired from various RESs such as wind and solar. The second step performs deep preprocessing of the acquired data via several de-noising and cleansing filters to clean the data and normalize them prior to actual processing. Thirdly, an AE is employed to extract the discriminative features from the cleaned data sequence through its encoder part. BiLSTM is used to learn these features to provide a final forecast of power generation. The proposed AB-Net was evaluated using two publicly available benchmark datasets where the proposed method obtains state-of-the-art results in terms of the error metrics.

show abstract

Photovoltaic Power Forecasting With a Hybrid Deep Learning Approach

Cited by 124 publications

References 51 publications

Short-Term Forecasting Photovoltaic Solar Power for Home Energy Management Systems

Short-Term Forecasting Photovoltaic Solar Power for Home Energy Management Systems

Short-Term Photovoltaic Power Forecasting Based on VMD and ISSA-GRU

AB-Net: A Novel Deep Learning Assisted Framework for Renewable Energy Generation Forecasting

Contact Info

Product

Resources

About