Short-term global horizontal irradiance forecasting based on a hybrid CNN-LSTM model with spatiotemporal correlations

Zang, Haixiang; Liu, Ling; Sun, Li; Cheng, Lilin; Wei, Zhinong; Sun, Guoqiang

doi:10.1016/j.renene.2020.05.150

Cited by 282 publications

(99 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to prevent overfitting, L 2 parameter regularisation is used in the full connection layer. The objective function of adding regular terms is shown in Equation (10) [36]:

e_{M S E} = \frac{1}{n} \sum_{t = 1}^{n} {false(p_{t}^{p r e^{'}} - p_{t}^{'} false)}^{2} + \frac{α}{2} w^{T} w

where p t ’ is the normalised wind power, α is the regular term parameter, w is the connection weight vector of the full connection layer, and n is the number of training samples. In addition, the optimisation method of random deactivation (dropout) is introduced. In the learning process, part of the weight or output of the hidden layer is randomly zeroed to reduce the interdependence between the hidden layer neurons so as to realise the regularisation of the NN. Evaluating index …”

Section: The Stcm Based On Cnn‐lstm For Multi‐step Wind Power Forecasmentioning

confidence: 99%

“…The test set is used to evaluate the accuracy of the model. In this study, the MAE, the mean absolute percentage error (MAPE), root mean square error (RMSE) and the normalised RMSE (NRMSE) are used as the evaluation indexes [36]:

e_{M A E} = \frac{1}{N} \sum_{i = 1}^{N} |p_{t}^{p r e} - p_{_{t}}|

e_{M A P E} = \frac{100 %}{N} \sum_{i = 1}^{N} |\frac{p_{t}^{p r e} - p_{_{t}}}{p_{_{t}}}|

e_{R M S E} = \sqrt{\frac{1}{N} \sum_{i = 1}^{N} {(p_{t}^{p r e} - p_{_{t}})}^{2}}

e_{N R M S E} = \frac{1}{P_{N}} \sqrt{\frac{1}{N} \sum_{i = 1}^{N} {((), p_{t}^{p r e} - p_{t})}^{2}}

…”

Section: The Stcm Based On Cnn‐lstm For Multi‐step Wind Power Forecasmentioning

confidence: 99%

“…In order to prevent overfitting, L 2 parameter regularisation is used in the full connection layer. The objective function of adding regular terms is shown in Equation ( 10) [36]:…”

Section: Model Trainingmentioning

confidence: 99%

See 2 more Smart Citations

Ultra‐short‐term multi‐step wind power forecasting based on CNN‐LSTM

Guan

et al. 2021

IET Renewable Power Gen

126

View full text Add to dashboard Cite

The fluctuation and intermission of large‐scale wind power integration is a serious threat to the stability and security of the power system. Accurate prediction of wind power is of great significance to the safety of wind power grid connection. This study proposes a novel spatio‐temporal correlation model (STCM) for ultra‐short‐term wind power prediction based on convolutional neural networks‐long short‐term memory (CNN‐LSTM). The original meteorological factors at multi‐historical time points of different sites throughout the target wind farm can be reconstructed into the input window of the model, and thus a new data reconstruction method is represented. CNN is used to extract the spatial correlation feature vectors of meteorological factors of different sites and the temporal correlation vectors of the meteorological features in ultra‐short term, which are reconstructed in time series and used as the input data of LSTM. Then, LSTM extracts the temporal feature relationship between the historical time points for multi‐step wind power forecasting. The STCM based on CNN‐LSTM proposed in this study is suitable for wind farms that can collect meteorological factors at different locations. Taking the measured meteorological factors and wind power dataset of a wind farm in China as an example, four evaluation metrics of the CNN‐LSTM model, CNN and LSTM individually used for multi‐step wind power prediction, are obtained. The results show that the proposed STCM based on CNN‐LSTM has better spatial and temporal characteristics extraction ability than the traditional structure model and can forecast the power of wind farm more accurately.

show abstract

“…In order to prevent overfitting, L 2 parameter regularisation is used in the full connection layer. The objective function of adding regular terms is shown in Equation (10) [36]:

e_{M S E} = \frac{1}{n} \sum_{t = 1}^{n} {false(p_{t}^{p r e^{'}} - p_{t}^{'} false)}^{2} + \frac{α}{2} w^{T} w

Section: The Stcm Based On Cnn‐lstm For Multi‐step Wind Power Forecasmentioning

confidence: 99%

e_{M A E} = \frac{1}{N} \sum_{i = 1}^{N} |p_{t}^{p r e} - p_{_{t}}|

e_{M A P E} = \frac{100 %}{N} \sum_{i = 1}^{N} |\frac{p_{t}^{p r e} - p_{_{t}}}{p_{_{t}}}|

e_{R M S E} = \sqrt{\frac{1}{N} \sum_{i = 1}^{N} {(p_{t}^{p r e} - p_{_{t}})}^{2}}

e_{N R M S E} = \frac{1}{P_{N}} \sqrt{\frac{1}{N} \sum_{i = 1}^{N} {((), p_{t}^{p r e} - p_{t})}^{2}}

…”

Section: The Stcm Based On Cnn‐lstm For Multi‐step Wind Power Forecasmentioning

confidence: 99%

See 1 more Smart Citation

Ultra‐short‐term multi‐step wind power forecasting based on CNN‐LSTM

Guan

et al. 2021

IET Renewable Power Gen

126

View full text Add to dashboard Cite

show abstract

“…LSTMs belong to the RNN architecture and are capable to learn long-term dependencies by avoiding the vanishing gradient problem of standard RNNs [16]. A hybrid structure combining those two has been widely applied in the short-term prediction especially when involving spatial-temporal features, such as the load profile prediction [17] and weather prediction [18] in multiple locations, as it takes advantages of respective merits from both CNN and LSTM models.…”

Section: Layersmentioning

confidence: 99%

Additive Gaussian process prediction for electrical loads compared with deep learning models

Ding

Mcculloch

2021

Proceedings of the Twelfth ACM International Conference on Future Energy Systems

View full text Add to dashboard Cite

Probabilistic prediction for electrical loads receives more attention in recent years for leveraging big data and assessing diverse scenarios. Since the classical machine learning (ML) model as a 'blackbox' predictor cannot produce the probabilistic prediction directly, Gaussian process (GP) model appears to be an effective solution. This paper proposes an additive GP model for the shortterm electrical load prediction, in which the characteristics of each feature can be encoded in the kernel. For comparison, we survey the literature and construct two deep learning models, quantile and ensemble deep neural networks (NNs), which can produce a probabilistic prediction. The results show the proposed additive GP model can outperform deep learning models, with the optimal kernel selection for features such as weather-related variables. CCS CONCEPTS• Computing methodologies → Supervised learning; Gaussian processes; Neural networks; • Applied computing → Forecasting.

show abstract

“…Comparison with the SVM and BPNN, the LSTM-RNN has the more contribution for PV power prediction. Zhang et.al [12] proposed the CNN-LSTM to predict PV power. By the evaluation of the prediction performance and generation ability, the proposed model has some advantages of PV power prediction.…”

Section: State Of the Artmentioning

confidence: 99%

Short-Term PV Power Prediction Based on Optimized VMD and LSTM

Wang

Liu

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Because of intermittence and fluctuation of photovoltaic (PV) power, it is difficult to enhance prediction accuracy. To sustain high-efficient operation of power system, this paper proposes a hybrid method to predict the short-term PV power. It consists of components separation of PV power, parameters optimization and reconstruction of prediction result. Firstly, the methods based on the identifying of feature frequency and mutual information maximum are proposed to optimize the mode number and penalty factor of VMD, respectively. The optimized VMD (OVMD) is used to decompose the complicated fluctuation components of PV power into single component. Then, the improved PSO (IPSO) based on non-linear inertia weight of anti-sine function is proposed to optimize the number of hidden layer nodes, learning rate and iteration number of LSTM network. The optimized LSTM is used to predict each single component of OVMD decomposition. Thirdly, the prediction result of each single component is reconstructed to obtain the final PV prediction power. The experiment result indicates that the prediction accuracy of the proposed method (OVMD-IPSO-LSTM) outperformances the other typical methods. By the improvement of the traditional method (VMD and PSO) and the parameter optimization of LSTM, this hybrid method makes a contribution to the prediction of short-term PV power.

show abstract

Short-term global horizontal irradiance forecasting based on a hybrid CNN-LSTM model with spatiotemporal correlations

Cited by 282 publications

References 46 publications

Ultra‐short‐term multi‐step wind power forecasting based on CNN‐LSTM

Ultra‐short‐term multi‐step wind power forecasting based on CNN‐LSTM

Additive Gaussian process prediction for electrical loads compared with deep learning models

Short-Term PV Power Prediction Based on Optimized VMD and LSTM

Contact Info

Product

Resources

About