Figure 5 - available via license: CC BY
Content may be subject to copyright.
Comparison of actual and forecast values for both LSTM and GRU models on days with irregular solar irradiance patterns.
Source publication
![Figure 1. Structure of a Gated Recurrent Units (GRU) cell [24].](publication/336801638/figure/fig1/AS:817875866619905@1572008210468/Structure-of-a-Gated-Recurrent-Units-GRU-cell-24_Q320.jpg)
Variation in solar irradiance causes power generation fluctuations in solar power plants. Power grid operators need accurate irradiance forecasts to manage this variability. Many factors affect irradiance, including the time of year, weather and time of day. Cloud cover is one of the most important variables that affects solar power generation, but...
Context in source publication
Context 1
... the standard error bars are wider during these months, indicating high variation in accuracy in that time period. Figure 5 shows the actual solar irradiance and forecasting results for each of the three experiments for GRU and LSTM on four different days throughout the testing year. On these given days, cloud cover was present throughout the day and the amount of solar irradiance during daylight hours was not consistent. ...
Similar publications
With the surge in energy demand worldwide, renewable energy is becoming increasingly important. Solar power, in particular, is positioning itself as a sustainable and environmentally friendly alternative, and is increasingly playing a role not only in large-scale power plants but also in small-scale home power generation systems. However, small-sca...
The Korea Institute of Energy Research builds Korean solar irradiance datasets, using gridded solar insolation estimates derived using the University of Arizona solar irradiance based on Satellite-Korea Institute of Energy Research (UASIBS-KIER) model, with the incorporation of geostationary satellites over the Korean Peninsula, from 1996 to 2019....
![Figure 1. Gated Recurrent Unit [19]. The activation functions shown...](profile/Majid-Hosseini-7/publication/343349866/figure/fig1/AS:919413737922560@1596216725391/Gated-Recurrent-Unit-19-The-activation-functions-shown-have-associated-sets-of_Q320.jpg)
Power grid operators rely on solar irradiance forecasts to manage uncertainty and variability associated with solar power. Meteorological factors such as cloud cover, wind direction, and wind speed affect irradiance and are associated with a high degree of variability and uncertainty. Statistical models fail to accurately capture the dependence bet...
In response to the problem of low forecasting accuracy in wind and solar power outputs, this study proposes a joint forecasting method for wind and solar power outputs by using their spatiotemporal correlation. First, autocorrelation analysis and causal testing are used to screen the forecasting factors. Then, a convolutional neural network–long sh...
Solar photovoltaic (PV) power generation is prone to drastic changes due to cloud cover. The power is easily affected within a very short period of time. Thus, the accuracy of grasping cloud distribution is important for PV power forecasting. This study proposes a novel sky image method to obtain the cloud coverage rate used for short-term PV power...
Citations
... The LSTM model predicts with less error. In addition, GRU and LSTM models were found to give better results than univariate statistical models [100]. In another GRU-LSTM comparison study, the GRU model showed higher accuracy in hourly and daily solar radiation forecasting for Buson, Korea. ...
... After a short period, Wojtkiewicz et al. [26] evaluated the use of gated recurrent units (GRUs) to predict solar irradiance and reported the results of utilizing multivariate GRUs to estimate the hourly solar irradiance in Phoenix, Arizona. Using purely historical solar irradiance data as well as the inclusion of external meteorological factors and cloudiness data, the authors compared and assessed the performance of GRUs and LSTM. ...
... It can be noticed that the proposed method performed better than all of the other listed state-of-the-art methods. Several methodologies in the literature focus on forecasting solar irradiance for a one-hour-ahead horizon [25][26][27]. The proposed method was similarly configured to predict solar irradiance in one-hour intervals, ensuring that the input features and training duration were aligned with those of the established state-of-the-art methods. ...
In the context of global warming, renewable energy sources, particularly wind and solar power, have garnered increasing attention in recent decades. Accurate forecasting of the energy output in microgrids (MGs) is essential for optimizing energy management, reducing maintenance costs, and prolonging the lifespan of energy storage systems. This study proposes an innovative approach to solar irradiance forecasting based on the theory of belief functions, introducing a novel and flexible evidential method for short-to-medium-term predictions. The proposed machine learning model is designed to effectively handle missing data and make optimal use of available information. By integrating multiple predictive models, each focusing on different meteorological factors, the approach enhances forecasting accuracy. The Yager combination method and pignistic transformation are utilized to aggregate the individual models. Applied to a publicly available dataset, the method achieved promising results, with an average root mean square error (RMS) of 27.83 W/m2 calculated from eight distinct forecast days. This performance surpasses the best reported results of 30.21 W/m2 from recent comparable studies for one-day-ahead solar irradiance forecasting. Comparisons with deep learning-based methods, such as long short-term memory (LSTM) networks and recurrent neural networks (RNNs), demonstrate that the proposed approach is competitive with state-of-the-art techniques, delivering reliable predictions with significantly less training data. The full potential and limitations of the proposed approach are also discussed.
... The GRU model was chosen in this analysis because of its superiority in weather prediction. GRU has been widely used in various domains, for example landslide displacement prediction [27], traffic prediction [28], electricity load forecasting [29], solar radiation forecasting [30], precision agriculture [31], wind speed and temperature estimation [32] carbon dioxide concentration prediction [33], and solar radiation forecasting [34]. The GRU model has shown promising results in terms of accuracy, prediction performance, and efficiency in various weather prediction tasks, making it a good choice for this analysis. ...
Floods are a major global problem affect communities and businesses. For these effects to be mitigated and emergency measures to be improved, accurate prediction is essential. Conventional flood prediction models frequently fail because the models ignore important hydrological elements like water discharge and instead solely use rainfall data. This limitation was addressed by the combination of rainfall and water discharge data on internet of things (IoT)-based technologies. It focuses on analyzing historical records from flood-prone areas in Semarang using gated recurrent unit (GRU) models. The findings demonstrate how effectively the GRU model performs when rainfall and water discharge factors are taken into account, resulting in very accurate and dependable predictions of flood events. Precision, Recall, and F1-score are evaluation metrics that demonstrate the accuracy on which the model determines flood emergency statuses. This study advances flood prediction methods and highlights the value of integrating internet of things data to improve preparedness and resilience against flood disasters.
... The effectiveness of the gating mechanism and memory cells in the LSTM architecture allowed it to learn long-term data dependencies, proving the suitability of LSTM and its variations for accurate predicting and timeseries data on solar irradiance. Gated Recurrent Units (GRU) were widely used in the literature before the widespread use of LSTM Wojtkiewicz et al. (2019). With fewer parameters and less memory needed, GRU's computational efficiency results in faster execution than LSTM. ...
Global power grid management depends on accurate solar energy estimation, yet present prediction techniques frequently suffer from unreliability as a result of abnormalities in solar energy data. Solar radiation projections are affected by variables such as anticipated horizon length, meteorological classification, and power measuring techniques. Therefore, a Solar Wind Energy Prediction System (SWEPS) is proposed as a solution to these problems. It improves renewable energy projections by taking sun trajectories and atmospheric characteristics into account. In addition to using a variety of optimization methods and pre-processing techniques, such as Principal Component Analysis (PCA), Recursive Feature Elimination (RFE), Least Absolute Shrinkage Selection Operator (LASSO), and recursive feature addition processes (RFA), complemented by a genetic algorithm for feature selection (GAFS). The SWEPS also makes use of sophisticated machine learning algorithms and Statistical Correlation Analysis (SCA) to find important connections. Neural Network Algorithms (NNA) and other metaheuristic techniques like Cuckoo Search Optimization (CSO), Social Spider Optimization (SSO), and Particle Swarm Optimization (PSO) are adopted in this work to increase the predictability and accuracy of models. Utilizing the strengths of machine learning and deep learning techniques (Artificial Neural Networks (ANN), Decision Trees, Support Vector Machine (SVM), Recurrent Neural Networks (RNN), and Long Short Term Memory (LSTM)) for robust forecasting, as well as meta-heuristic optimization techniques to fine-tune hyper-parameters and achieve near-optimal values and significantly improve model performance, are some of this work contributions to the development of a comprehensive prediction system.
... The MAE and MSE values for the LSTM model were slightly higher than those for the GRU, at 0.03659 and 0.00648, respectively, while the RNN model yielded 0.0476 and 0.00802. In contrast, some studies have shown that LSTM has greater accuracy than its alternative, GRU, although it comes with an increased time trade-off [122], [123]. A BiLSTM, another RNN model, may forecast PV power more accurately than LSTM and GRU, as shown in Table 2. Based on data from a meteorological station in Amherst, USA, it was determined that the BiLSTM model predicts solar energy slightly better than the GRU model [124]. ...
... Some studies have shown that LSTM tends to have greater accuracy than its alternative, GRU, although it does come with a trade-off of increased time. In [123], the LSTM model has a slightly lower error rate than the GRU model. However, the GRU model requires less time for training (2.9 hours) and prediction (11.02 seconds) than the LSTM model, which takes 3.27 hours for training and 11.84 seconds for prediction. ...
Solar energy is largely dependent on weather conditions, resulting in unpredictable, fluctuating, and unstable photovoltaic (PV) power outputs. Thus, accurate PV power forecasts are increasingly crucial for managing and controlling integrated energy systems. Over the years, advanced artificial neural network (ANN) models have been proposed to increase the accuracy of PV power forecasts for various geographical regions. Hence, this paper provides a state-of-the-art review of the five most popular and advanced ANN models for PV power forecasting. These include multilayer perceptron (MLP), recurrent neural network (RNN), long short-term memory (LSTM), gated recurrent unit (GRU), and convolutional neural network (CNN). First, the internal structure and operation of these models are studied. It is then followed by a brief discussion of the main factors affecting their forecasting accuracy, including forecasting horizons, meteorological conditions, and evaluation metrics. Next, an in-depth and separate analysis of standalone and hybrid models is provided. It has been determined that bidirectional GRU and LSTM, whether used as a standalone model or in a hybrid configuration, offer greater forecasting accuracy. Furthermore, hybrid and upgraded metaheuristic algorithms have demonstrated exceptional performance when applied to standalone and hybrid ANN models. Finally, this study discusses various limitations and shortcomings that may influence the practical implementation of PV power forecasting.
... Our expanded suite of exogenous variables builds on this work. Distinctions between 'cloudy' and 'sunny' have been made by Lim et al. [18] for solar power generation, and whilst cloud cover is certainly a useful exogenous variable in this instance, research by Wojtkiewicz et al. [20] has suggested that a greater range of variables could have improved accuracies further. One-hour-ahead forecast errors were significantly reduced for solar irradiance when variables such as temperature and humidity were introduced in this study, even independently of cloud cover. ...
Accurately forecasting energy metrics is essential for efficiently managing renewable energy generation. Given the high variability in load and renewable energy power output, this represents a crucial area of research in order to pave the way for increased adoption of low-carbon energy solutions. Whilst the impact of different neural network architectures and algorithmic approaches has been researched extensively, the impact of utilising additional weather variables in forecasts have received far less attention. This article demonstrates that weather variables can have a significant influence on energy forecasting and presents methodologies for using these variables within a long short-term memory (LSTM) architecture to achieve improvements in forecasting accuracy. Moreover, we introduce the use of the seasonal components of the target time series, as exogenous variables, that are also observed to increase accuracy. Load, solar and wind generation time series were forecast one hour ahead using an LSTM architecture. Time series data were collected in five Spanish cities and aggregated for analysis, alongside five exogenous weather variables, also recorded in Spain. A variety of LSTM architectures and hyperparameters were investigated. By tuning exogenous weather variables, a 33% decrease in mean squared error was observed for solar generation forecasting. A 22% decrease in mean absolute squared error (MASE), compared to 24-h ahead forecasts made by the Transmission Service Operator (TSO) in Spain, was also observed for solar generation. Compared to using the target variable in isolation, utilising exogenous weather variables decreased MASE by approximately 10%, 15% and 12% for load, solar and wind generation, respectively. By using the seasonal component of the target variables as an exogenous variable itself, we demonstrated decreases in MASE of 19%, 12% and 8% for load, solar and wind generation, respectively. These results emphasise the significant benefits of incorporating weather and seasonal components into energy-related time series forecasts.
... To consider temporal factors, research studies on solar irradiance forecasting using deep learning have used mainly LSTM [21] and GRU [22], which are deep learning models specialized for time-series prediction. However, to consider spatial factors, they have used either a convolution layer in combination or a convolution operation method [23,24]. ...
This paper proposes a method for forecasting surface solar irradiance (SSI), the most critical factor in solar photovoltaic (PV) power generation. The proposed method uses 16-channel data obtained by the GEO-KOMPSAT-2A (GK2A) satellite of South Korea as the main data for SSI forecasting. To determine feature variables related to SSI from the 16-channel data, the differences and ratios between the channels were utilized. Additionally, to consider the fundamental characteristics of SSI originating from the sun, solar geometry parameters, such as solar declination (SD), solar elevation angle (SEA), and extraterrestrial solar radiation (ESR), were used. Deep learning-based feature selection (Deep-FS) was employed to select appropriate feature variables that affect SSI from various feature variables extracted from the 16-channel data. Lastly, spatio-temporal deep learning models, such as convolutional neural network–long short-term memory (CNN-LSTM) and CNN–gated recurrent unit (CNN-GRU), which can simultaneously reflect temporal and spatial characteristics, were used to forecast SSI. Experiments conducted to verify the proposed method against conventional methods confirmed that the proposed method delivers superior SSI forecasting performance.
... G. Guariso et al. [40] validated the accuracy of FF and LSTM networks for predicting environmental variable time series, emphasizing the effect of null values and midnight samples on performance metrics. J. Wojtkiewicz et al. [41] employ univariate and multivariate GRU and LSTM models to predict Phoenix, Arizona's solar irradiance based on historical data, weather variables, and cloud cover data. ...
Bangladesh's subtropical climate with an abundance of sunlight throughout the greater portion of the year results in increased effectiveness of solar panels. Solar irradiance forecasting is an essential aspect of grid-connected photovoltaic systems to efficiently manage solar power's variation and uncertainty and to assist in balancing power supply and demand. This is why it is essential to forecast solar irradiation accurately. Many meteorological factors influence solar irradiation, which has a high degree of fluctuation and uncertainty. Predicting solar irradiance multiple steps ahead makes it difficult for forecasting models to capture long-term sequential relationships. Attention-based models are widely used in the field of Natural Language Processing for their ability to learn long-term dependencies within sequential data. In this paper, our aim is to present an attention-based model framework for multivariate time series forecasting. Using data from two different locations in Bangladesh with a resolution of 30 min, the Attention-based encoder-decoder, Transformer, and Temporal Fusion Transformer (TFT) models are trained and tested to predict over 24 steps ahead and compared with other forecasting models. According to our findings, adding the attention mechanism significantly increased prediction accuracy and TFT has shown to be more precise than the rest of the algorithms in terms of accuracy and robustness. The obtained mean square error (MSE), the mean absolute error (MAE), and the coefficient of determination (R²) values for TFT are 0.151, 0.212, and 0.815, respectively. In comparison to the benchmark and sequential models (including the Naive, MLP, and Encoder-Decoder models), TFT has a reduction in the MSE and MAE of 8.4–47.9% and 6.1–22.3%, respectively, while R² is raised by 2.13–26.16%. The ability to incorporate long-distance dependency increases the predictive power of attention models.
... In [21], a GA and a BiLSTM model are combined for very short-term PV energy forecasting, demonstrating the strength of this type of model in forecasting complex and nonlinear time series. In [22], a gated recurrent unit (GRU) model is used to predict solar irradiance, with tests showing that the inclusion of cloud cover data and other exogenous variables enhances forecast quality. ...
Photovoltaic (PV) power prediction plays a critical role amid the accelerating adoption of renewable energy sources. This paper introduces a bidirectional long short-term memory (BiLSTM) deep learning (DL) model designed for forecasting photovoltaic power one hour ahead. The dataset under examination originates from a small PV installation located at the Polytechnic School of the University of Alcala. To improve the quality of historical data and optimize model performance, a robust data preprocessing algorithm is implemented. The BiLSTM model is synergistically combined with a Bayesian optimization algorithm (BOA) to fine-tune its primary hyperparameters, thereby enhancing its predictive efficacy. The performance of the proposed model is evaluated across diverse meteorological and seasonal conditions. In deterministic forecasting, the findings indicate its superiority over alternative models employed in this research domain, specifically a multilayer perceptron (MLP) neural network model and a random forest (RF) ensemble model. Compared with the MLP and RF reference models, the proposed model achieves reductions in the normalized mean absolute error (nMAE) of 75.03% and 77.01%, respectively, demonstrating its effectiveness in this type of prediction. Moreover, interval prediction utilizing the bootstrap resampling method is conducted, with the acquired prediction intervals carefully adjusted to meet the desired confidence levels, thereby enhancing the robustness and flexibility of the predictions.
... Over the years, a variety of research studies have utilized DL models, such as long short-term memory (LSTM) [10], [11], [12] and gated recurrent unit (GRU) [13], [14], to forecast solar PV power generation. LSTM models excel in capturing complex temporal patterns in solar PV power generation data, enabling accurate predictions. ...
Solar photovoltaic (PV) power generation is gradually increasing, but its intermittent nature poses challenges to grid stability. To address this, advanced forecasting methods, such as deep learning (DL) algorithms, can be employed to ensure a more stable and reliable energy supply. Accurate short-term forecasts are essential for electricity grids to effectively mitigate the impact of solar intermittency and enhance grid performance. This research contributes by developing a hybrid DL model that combines a 1-dimensional convolutional neural network (1D CNN) with a gated recurrent unit (GRU), referred to as “1D CNN-GRU”. The 1D CNN module extracts essential features from time series data, such as solar PV power generation, while the GRU component provides high-precision short-term forecasts. Additionally, data preparation techniques, including feature selection using SHapley Additive exPlanations (SHAP), data smoothing with an exponential moving average (EMA), and data augmentation with Gaussian noise, are employed to enhance the performance of the proposed 1D CNN-GRU model. To evaluate the effectiveness of the proposed model, it was compared with other DL models, including CNN, GRU, long short-term memory (LSTM), and CNN-GRU. The forecasting was performed using the Hydro-Floating Solar Plant dataset, obtained from the 45 MW hydro-floating solar installation located at Sirindhorn Dam in Ubon Ratchathani province, Thailand. The proposed 1D CNN-GRU model was tested using data from three different seasons: winter, summer, and the rainy season. The model achieved the lowest root mean square error (
RMSE
) across all seasons, with values of 0.025 (winter), 0.050 (summer), and 0.094 (rainy), and demonstrated the shortest training time. The forecasting results indicated that the proposed model outperformed all other models in terms of both accuracy and training time.
