Sky Image Prediction Model Based on Convolutional Auto-Encoder for Minutely Solar PV Power Forecasting

Fu, Yuwei; Chai, Hua; Zhen, Zhao; Wang, Fei; Xu, Xunjian; Li, Kangping; Shafie‐khah, Miadreza; Dehghanian, Payman; Catalão, João P.S.

doi:10.1109/tia.2021.3072025

Cited by 47 publications

(8 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(2) Cloud clusters are a key factor affecting the accuracy of DPV power forecasting at the ultra-short time scale. The cloud motion displacement and cloud feature can be extracted from sky or satellite images which have been achieved in previous studies [34,52,53]. These tricks will be combined with the proposed model to further improve the accuracy of DPV power forecasting.…”

Section: Discussionmentioning

confidence: 99%

“…As shown in Figure 1b, the correlation between DPV sites is in dynamic change, especially when the output fluctuates sharply. Solar irradiance has a direct impact on DPV output [34]. For the ultra‐short‐term scales, the moving cloud is the main factor that affects the amount of irradiance reaching the ground surface, thus having a significant influence on power volatility [35].…”

Section: Spatiotemporal Correlation Analysismentioning

confidence: 99%

See 1 more Smart Citation

Dynamic directed graph convolution network based ultra‐short‐term forecasting method of distributed photovoltaic power to enhance the resilience and flexibility of distribution network

Wang,

Fu,

Zhang

et al. 2023

IET Generation Trans & Dist

Self Cite

View full text Add to dashboard Cite

Accurately forecasting regional distributed photovoltaic (DPV) power is crucial in mitigating the negative impact of high DPV penetration on the reliability and resilience of the distribution network. However, most of the current photovoltaic power forecasting methods suffer from two key problems: (1) ignoring the asymmetric influence relationship among DPV sites; (2) lack of consideration of dynamic spatiotemporal correlation among DPV sites. As a result, these methods are unable to fully adapt to the characteristics of DPV, making it challenging to directly apply the existing forecasting methods to improve the accuracy of DPV power forecasting. To conquer this limitation, a dynamic directed Graph Convolution Neural Network (DDGCN) is applied to regional DPV ultra‐short‐term power forecasting. Unlike the conventional Graph Convolution Neural Network (GCN) based forecasting methods, the proposed method improves GCN to process the directed graph. On this basis, to capture the dynamic and directed adjacency relationship among graph nodes, a temporal attention mechanism is proposed and combined with the directed GCN model. In this way, the dynamic and asymmetric/directed relationships among DPV sites can be taken into account. It is worth noting that the DPVs’ adjacency relationship can be constructed without any prior knowledge by end‐to‐end model training. The simulation experiment proves that the prediction accuracy can be further improved by taking into account the dynamic directed relationship among the sites via a real DPV power dataset.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Spatiotemporal Correlation Analysismentioning

confidence: 99%

Dynamic directed graph convolution network based ultra‐short‐term forecasting method of distributed photovoltaic power to enhance the resilience and flexibility of distribution network

Wang,

Fu,

Zhang

et al. 2023

IET Generation Trans & Dist

Self Cite

View full text Add to dashboard Cite

show abstract

“…Decision tree [12] Support vector regression [13] Artificial neural network [14,17,18] RNN [15] CNN [20] LSTM [19] Deep learning [16,21] DBN(deep belief networks) [22] DCNN(deep convolutional neural network) [23] AE, CAE [24,25] GAN [26] Hybrid methods…”

Section: Statistical Time-series Methodsmentioning

confidence: 99%

“…The deep belief network (DBN) method has high‐level abstraction extraction ability [22], the deep convolutional neural network (DCNN) approach can fit complex non‐linear mappings and processing different data sets [23], and the autoencoder (AE) approach have the characteristic of dimensionality reduction and denoising [24]. Furthermore, the encoding and decoding structure of CAE also has already demonstrated strong sky image prediction ability for spatiotemporal information in PV power forecasting [25].…”

Section: Literature Reviewmentioning

confidence: 99%

A hybrid method for day‐ahead photovoltaic power forecasting based on generative adversarial network combined with convolutional autoencoder

Pan

Zhou

Sun

et al. 2022

IET Renewable Power Gen

View full text Add to dashboard Cite

Photovoltaic (PV) generation has high impact on the decarbonization pathways of power systems. Accuracy of day‐ahead PV power forecasting has become crucial in the operation and control of power system with high PV penetration. This paper develops a hybrid approach based on generative adversarial network (GAN) combined with convolutional autoencoder (CAE) to improve PV power forecasting accuracy. Self‐organizing map method is first utilized as data pre‐processing to classify target days into different weather types based on solar irradiance. With the ability of GAN to reduce the burden of loss and the advantages of CAE to extract multi‐scale effective features from the weather and PV power, PV power forecasting model consisting of GAN and CAE is proposed. The developed method has been tested on a real dataset in a Chinese PV station and compared with base reference PV forecasting methods. Numerical testing results demonstrate the effectiveness of our method with high accuracy.

show abstract

“…There are common prediction methods based on mathematical methods such as grey theory, time series analysis (Yan et al, 2021), machine learning methods such as support vector regression and BP neural network (Feng et al, 2015). In addition, environmental information like the weather forecast, the satellite image (Wang et al, 2020) and cloud distribution (Fu et al, 2021) are used to support photovoltaic output prediction. The prediction method based on environmental information can achieve high prediction accuracy (Manokar, 2020;Sasikumar and calorimetry, 2020), but these methods needs the using of satellite cloud map and large climate database which will increase the cost of the prediction (Chai et al, 2019), multi-layer perceptron (MLP) , convolution neural network (CNN) and long short term memory (LSTM) belongs to the deep learning networks.…”

Section: Introductionmentioning

confidence: 99%

Photovoltaic power prediction based on sliced bidirectional long short term memory and attention mechanism

Chen

Liu

et al. 2023

Front. Energy Res.

View full text Add to dashboard Cite

Solar photovoltaic power generation has the characteristics of intermittence and randomness, which makes it a challenge to accurately predict solar power generation power, and it is difficult to achieve the desired effect. Therefore, by fully considering the relationship between power generation data and climate factors, a new prediction method is proposed based on sliced bidirectional long short term memory and the attention mechanism. The prediction results show that the presented model has higher accuracy than the common prediction models multi-layer perceptron, convolution neural network, long short term memory and bidirectional long short term memory. The presented sliced bidirectional cyclic network has high prediction accuracy by low root mean square error and mean absolute error of 1.999 and 1.159 respectively. The time cost is only 24.32% of that of long short term memory network and 13.76% of that of bidirectional long short term memory network.

show abstract

Sky Image Prediction Model Based on Convolutional Auto-Encoder for Minutely Solar PV Power Forecasting

Cited by 47 publications

References 43 publications

Dynamic directed graph convolution network based ultra‐short‐term forecasting method of distributed photovoltaic power to enhance the resilience and flexibility of distribution network

Dynamic directed graph convolution network based ultra‐short‐term forecasting method of distributed photovoltaic power to enhance the resilience and flexibility of distribution network

A hybrid method for day‐ahead photovoltaic power forecasting based on generative adversarial network combined with convolutional autoencoder

Photovoltaic power prediction based on sliced bidirectional long short term memory and attention mechanism

Contact Info

Product

Resources

About