Solar Radiation Prediction Research Articles

Machine Learning Dynamic Ensemble Methods for Solar Irradiance and Wind Speed Predictions

This paper proposes to analyze the performance increase in the forecasting of solar irradiance and wind speed by implementing a dynamic ensemble architecture for intra-hour horizon ranging from 10 to 60 min for a 10 min time step data. Global horizontal irradiance (GHI) and wind speed were computed using four standalone forecasting models (random forest, k-nearest neighbors, support vector regression, and elastic net) to compare their performance against two dynamic ensemble methods, windowing and arbitrating. The standalone models and the dynamic ensemble methods were evaluated using the error metrics RMSE, MAE, R2, and MAPE. This work’s findings showcased that the windowing dynamic ensemble method was the best-performing architecture when compared to the other evaluated models. For both cases of wind speed and solar irradiance forecasting, the ensemble windowing model reached the best error values in terms of RMSE for all the assessed forecasting horizons. Using this approach, the wind speed forecasting gain was 0.56% when compared with the second-best forecasting model, whereas the gain for GHI prediction was 1.96%, considering the RMSE metric. The development of an ensemble model able to provide accurate and precise estimations can be implemented in real-time forecasting applications, helping the evaluation of wind and solar farm operation.

Performance Evaluation of LSTM Based Solar Irradiance Prediction Models without Long-term Local Weather Data

Performance Evaluation of LSTM Based Solar Irradiance Prediction Models without Long-term Local Weather Data

A Systematic Review of Neural Network Autoregressive Model with Exogenous Input for Solar Radiation Prediction Modelling Development

Neural Network is one of the Machine Learning methods that has been applied in various Artificial Intelligence system development including solar radiation prediction modelling. Since there are multiple approaches had been developed using the Neural Network method, the study has been focusing on the development of a Multi-layer Neural Network model that can handle non-linearities and highly dynamic data. The integration of the Multi-layer Neural Network and the Non-linear Autoregressive Model with Exogenous Input (NARX) developed a compromising non-linear Neural Network model which can be applied in the modelling of solar radiation. This paper develops a systematic review of the Neural Network Autoregressive Model with Exogenous Input (NNARX) for solar radiation prediction modelling starts from the architecture and the comparative selection for the Training Function. The model is developed and analysed using MATLAB R2019a software. Results showed that the Levenberg-Marquardt Training Function performed better with the R2 value of 0.94 for training and 0.91 for testing, making it the most suitable for the NNARX in the development of solar radiation prediction modelling.

An improved SSA-BiLSTM-based short-term irradiance prediction model via sky images feature extraction

The development and utilization of solar energy has become an important strategic decision for the sustainable development of many countries. Short-term variations in solar irradiation have an impact on the safety and stability of photovoltaic and solar thermal power plants, therefore, the development and accuracy of solar irradiance prediction models have received much attention. This paper proposes a short-term irradiance prediction model based on mixed intelligent optimization algorithm and deep learning algorithm that integrates features of various forms of information. First, the sequence containing the picture attributes as well as the color and texture characteristics are recovered from ground-based cloud images, historical irradiance and meteorological feature information is decomposed and reconstructed by singular spectrum analysis (SSA). Secondly, the bidirectional short term to long term memory (BiLSTM) network is optimized for model training and finally assessment using an enhanced chaotic sparrow search method. The technique exceeds benchmark methods and a number of sophisticated individual algorithms in forecasting ultra-short-term global horizontal irradiance (GHI), according to experimental data, while also offering extremely high accuracy and resilience.

Solar Radiation Prediction Based on Conformer-GLaplace-SDAR Model

Solar energy, as a clean energy source, has tremendous potential for utilization. The advancement of solar energy utilization technology has led to an increasing demand for solar energy, resulting in a growing need for the accurate prediction of solar radiation. The main objective of this study is to develop a novel model for predicting solar radiation intervals, in order to obtain accurate and high-quality predictions. In this study, the daily sunshine duration (SD), average relative humidity (RHU), and daily average temperature (AT) were selected as the indicators affecting the daily global solar radiation (DGSR). The empirical study conducted in this research utilized daily solar radiation data and daily meteorological data collected at the Hami station in Xinjiang from January 2009 to December 2016. In this study, a novel solar radiation interval prediction model was developed based on the concept of “point prediction + interval prediction”. The Conformer model was employed for the point prediction of solar radiation, while the Generalized Laplace (GLaplace) distribution was chosen as the prior distribution to account for the prediction error. Furthermore, the Solar DeepAR Forecasting (SDAR) model was utilized to estimate parameters of the fitted residual distribution and achieve the interval prediction of solar radiation. The results showed that both models performed well, with the Conformer model achieving a Mean Squared Error (MSE) of 0.8645, a Mean Absolute Error (MAE) of 0.7033 and the fitting coefficient R2 of 0.7751, while the SDAR model demonstrated a Coverage Width-based Criterion (CWC) value of 0.068. Compared to other conventional interval prediction methods, our study’s model exhibited superior accuracy and provided a more reliable solar radiation prediction interval, offering valuable information for ensuring power system safety and stability.

Prediction of the daily global solar Irradiation of the great Maghreb region using the complex-valued neural networks

In this paper, the prediction of the daily global solar irradiation of the great Maghreb’s region using the complex-valued neural networks (CVNN) is presented. This paper is an extension of our recent published work which is the complex-valued forecasting of the global solar irradiation. Both multi-input single-output (MISO) and multi-input multi-output (MIMO) strategies are considered. The data of the capitals of the great Maghreb, which are Tripoli (Libya), Tunis (Tunisia), Algiers (Algeria), Rabat (Morocco), El Aaiun (Western Sahara) and Nouakchott (Mauritania), are used like sample from each country. To test the applicability and the feasibility of the CVNN to predict the daily global irradiation for the great Maghreb case, several models are presented. Results obtained throughout this paper show that the CVNN technique is suitable for prediction of the daily solar irradiation of the region of region of the great Maghreb.

Unveiling the SALSTM-M5T model and its python implementation for precise solar radiation prediction

The world is increasingly embracing cleaner and more sustainable energy sources, with solar energy playing a crucial role in mitigating greenhouse gas emissions and addressing climate change. Accurate solar radiation predictions are vital for optimizing solar energy resource utilization and identifying suitable locations for solar power plants. Therefore, our study introduces a new model to advance solar and renewable energy systems. In this paper, we suggest a novel hybrid model, Self-attention (SA) mechanism-long short-term memory neural network (LSTM)-M5Tree (SALSTM5T) model, for predicting solar radiation. The SALSTM-M5T model combines the advantages of the Self-attention- LSTM (SALSTM) and M5Tree models. The LSTM component captures long-term solar radiation dependencies. SA improves the accuracy of LSTM and M5T models by focusing on relevant input features at different time steps. The study utilizes K-fold cross-validation to overcome the limitations of traditional methods for determining the size of training and testing data. By combining the advantages of the SALSTM and M5Tree models, our research presents a new framework for accurate solar radiation prediction. Existing techniques are developed by using SA and K-fold cross-validation. Furthermore, our paper emphasizes the practical applications of solar radiation prediction, such as identifying suitable areas for solar power plants and optimizing energy production. Our study concluded that the self-attention mechanism and LSTM model improved the efficiency of M5T models for analyzing solar time series data as these models can attend to important features. The centralized root mean square error (CRMSE) of the SALSTM-M5T, LSTM-M5T, LSTM, ANN, and M5T models was 0.04, 0.17, 0.25, 0.49, and 0.70, respectively. The SALSTM-M5T, LSTM-M5T, LSTM, ANN, and M5T models' correlation coefficients were 0.99, 0.98, 0.96, 0.89, and 0.82, respectively. Our SALSTM-M5T model contributes to advancing renewable energy planning and decision-making. The main innovation of the current article is the development of the M5T model using the capabilities of the SA and LSTM models. The SALSTM-M5T hybrid model, which combines self-attention, LSTM and M5Tree techniques, is proposed as an effective approach for solar radiation prediction. The results show its superiority over other models in terms of accuracy and suitability, which can be useful for practical applications in the field of renewable energy, such as site selection for solar power plants and optimization of energy production. Furthermore, this paper aligns with the advancement of digital sensors to enhance the accuracy of solar radiation prediction.

Decomposition spectral graph convolutional network based on multi-channel adaptive adjacency matrix for renewable energy prediction

The intermittence and randomicity of renewable energy can seriously disrupt its productivity and reliability, particularly in unforeseen circumstances. Yet, most existing renewable energy prediction (REP) methods fail to model the inter-temporal correlations among input variables and suffer from impractical decomposition process, resulting in less good prediction accuracy. Aiming to improve these issues and power REP technique to better serve modern society, this work proposes a novel decomposition spectral graph convolutional network, named DASGCN, as a general solution for REP. Compared with existing REP models, DASGCN has two main characteristics: (1) It incorporates a carefully designed decomposition block which can decompose input multivariate time series efficiently, (2) A multi-channel adaptive adjacent matrix is proposed to model the complex correlation among entangled input variables automatically and generate the interpretable graph structure without prior domain knowledge. The comprehensive experiments on four real-world datasets demonstrate that the proposed DASGCN can significantly outperform benchmark models and achieve high performance on wind speed, wind power, solar radiation and photovoltaic power prediction tasks.

Assessment of the high-resolution estimations of global and diffuse solar radiation using WRF-Solar

Compared with physical models, WRF-Solar, as an excellent numerical forecasting model, includes abundant novel cloud physical and dynamical processes, which enablesenable the high-frequency output of radiation components which are urgently needed by the solar energy industry. However, the popularisation of WRF-Solar in a wide range of applications, such as the estimation of diffuse radiation, suffers from unpredictable influences of cloud and aerosol optical property parameters. This study assessed the accuracy of the improved numerical weather prediction (WRF-Solar) model in simulating global and diffuse radiation. Aerosol optical properties at 550 nm, which were provided by a moderate resolution imaging spectroradiometer, were used as input to analyse the differences in accuracies obtained by the model with/without aerosol input. The sensitivity of WRF-Solar to aerosol and cloud optical properties and solar zenith angle (SZA) was analysed. The results show the superiority of WRF-Solar to WRF-Dudhia in terms of their root mean square error (RMSE) and mean absolute error (MAE). The coefficients of determination between WRF-Solar and WRF-Dudhia revealed no statistically significant difference, with values greater than 0.9 for the parent and nested domains. In addition, the relative RMSE (RRMSE%) reached 46.60%. The experiment on WRF-Solar and WRF-Dudhia revealed a negative bias for global radiation, but WRF-Solar attained a slightly lower RMSE and higher correlation coefficient than WRF-Dudhia. The WRF-Solar-simulated results on diffuse radiation under clear sky conditions were slightly poorer, with RMSE, RRMSE, mean percentage error and MAE of 181.93 W m−2, 170.52%, 93.04% and 138 W m−2, respectively. Based on Himawari-8 cloud data, statistical results on cloud optical thickness (COT) for cloudy days revealed that WRF-Solar overestimated diffuse radiation at COTs greater than 20. Moreover, when the aerosol optical depth was greater than or equal to 0.8, WRF-Solar also overestimated the diffuse radiation, with a mean difference of 58.57 W m−2. The errors of WRF-Solar simulations in global and diffuse radiation exhibited a significant dependence on the SZA. The dispersion degree of deviation increased gradually with the decrease in the SZA. Thus, WRF-Solar serves as an improved numerical tool that can provide high temporal and high-spatial-resolution solar radiation data for the prediction of photovoltaic power. Studies should explore the improvement of cumulus parameterisation schemes to enhance the accuracy of solar radiation component estimation and prediction under cloudy conditions.

Prediction of solar irradiance based on Python

The rapid development of modern industrial society has relied heavily on cheap and abundant fossil fuel energy. However, to achieve sustainable development, there is an increasing focus on developing new energy sources such as photovoltaics (PV) and wind energy. In the context of using solar irradiance to generate electricity, predicting the solarpower in advance is crucial for efficient utilization. This paper utilizes the pvlib-python model to predict three types of irradiance in clear sky conditions: POA_DNI, POA_GHI, and POA_DHI. Furthermore, we incorporate aerosol data from pvlib to improve the prediction accuracy.Three sites from BSRN are selected and the predicted data are compared with the observed data to evaluate the model's prediction effectiveness. The result reveals that the model performs best for POA_GHI and the actual cloud cover distribution has a significant impact on the prediction accuracy.

Solar forecasting for a PV-battery powered DC system

The photovoltaic (PV) power generation sector has been growing rapidly as a result of the rising need for solar power and the advancement of PV technology. PV Power generation is affected by weather factors such as cloud cover, solar irradiation, temperature, breeze direction and speed, and the amount of rain or snow. As a result, a highly precise PV power predictor is essential to improve security and reliability in the face of financial penalties and ambiguity. Hence, this paper suggests a novel approach to improve the efficiency of PV-battery-powered DC systems by combining solar irradiance prediction using the Long Short-Term Memory (LSTM) algorithm with a power electronic converter design that incorporates a bidirectional port. The LSTM algorithm was employed to predict one week of solar data with a remarkable R2 score of 0.96. A steady-state analysis of the proposed Three-Port Converter (TPC) is performed for five different operating modes to guarantee optimal performance. The suggested system's prediction performance was tested using several error metrics such as Mean Absolute Error (MAE), Mean Square Error (MSE), and Root Mean Squared Error (RMSE), which were computed as 0.0318, 0.0027, and 0.0526, respectively. Results from the above error measures show that the suggested approach performs more effectively in estimating solar irradiance. The Adaptive Neural-Fuzzy Interface System (ANFIS) and Incremental Conductance (IC) algorithms are employed for Maximum Power Point Tracking (MPPT) and assessed against various atmospheric conditions. From the MATLAB simulation results, the tracking efficiency of the ANFIS-based MPPT technique is 99.97 %, which is superior to the IC-based MPPT technique. Furthermore, it proved that the suggested approach improves the efficiency of PV-battery-powered DC systems, which is more appropriate for real-world applications such as DC microgrids and Electric Vehicles.

Interpretable deep learning for hourly solar radiation prediction: A real measured data case study in Tokyo

With their superior performance in capturing complex relationships between input and output variables, deep learning models have gained popularity in the field of solar radiation prediction. However, the use of deep learning models as a predictive tool is often constrained by their inherent complexity, which can limit the interpretability and usability of the model in practical scenarios, such as model predictive control (MPC) in energy systems. In this study, time series are transformed into an image, and the hourly solar radiation prediction is realized through a convolutional neural network (CNN). The experiments demonstrate that the proposed CNN model enhances interpretability at the cost of a minor 6 % reduction in prediction accuracy, compared to the baseline model. Analogous to the deconvolution operation for image classification, feature maps of CNN networks can thus be used for model interpretability, whereby the closer the feature maps are to the input (In the context of forward propagation, this means that the layers of a CNN closer to the input data), the higher is the interpretability. In extracting the feature map using CNN, solar radiation and radiation time played the most important roles in the entire prediction process.

Application of Deep Learning to the Prediction of Solar Irradiance through Missing Data

The task of predicting solar irradiance is critical in the development of renewable energy sources. This research is aimed at predicting the photovoltaic plant’s irradiance or power and serving as a standard for grid stability. In practical situations, missing data can drastically diminish prediction precision. Meanwhile, it is tough to pick an appropriate imputation approach before modeling because of not knowing the distribution of datasets. Furthermore, not all datasets benefit equally from using the same imputation technique. This research suggests utilizing a recurrent neural network (RNN) equipped with an adaptive neural imputation module (ANIM) to estimate direct solar irradiance when some data is missing. Without imputed information, the typical projects’ imminent 4-hour irradiance depends on gaps in antique climatic and irradiation records. The projected model is evaluated on the widely available information by simulating missing data in each input series. The performance model is assessed alternative imputation techniques under a range of missing rates and input parameters. The outcomes prove that the suggested methods perform better than competing strategies when measured by various criteria. Moreover, combine the methodology with the attentive mechanism and invent that it excels in low-light conditions.

Solar irradiance prediction using reinforcement learning pre-trained with limited historical data

Accurate day-ahead forecasting of solar irradiance is crucial for maintaining a steady power supply and minimizing energy losses. To date, various solar irradiance prediction models have been established, but these typically require extensive weather data collected over long periods within the area of prediction or consistent updates using field measurements. This research introduces a reinforcement learning-based model capable of long-term solar irradiance prediction, even in areas with limited accumulated data. Our proposed model can forecast solar radiation for more than a year using just two weeks of solar radiation learning and readily available weather forecasts. It demonstrated a promising performance, with an annual average CVRMSE error of 7.0%, which is a more optimized predictive performance than the 12.8% CVRMSE yielded by the existing LSTM-based Reference model constructed by adding out-of-atmosphere solar radiation input values.

Long-term prediction of daily solar irradiance using Bayesian deep learning and climate simulation data

Long-term prediction of daily solar irradiance using Bayesian deep learning and climate simulation data

Explainable forecasting of global horizontal irradiance over multiple time steps using temporal fusion transformer

Accurate prediction of solar irradiance is essential for the successful integration of solar power plants into electrical systems. Despite recent advancements in deep learning technology yielding impressive results in solar forecasting, their lack of interpretability has hindered their widespread adoption. In this paper, we propose a novel approach that integrates a Temporal Fusion Transformer (TFT) with a McClear model to achieve accurate and interpretable forecasting performance. The TFT is a deep learning model that provides transparency in its predictions through the use of interpretable self-attention layers for long-term dependencies, recurrent layers for local processing, specialized components for feature selection, and gating layers to suppress extraneous components. The model is capable of learning temporal associations between continuous time-series variables, namely, historical global horizontal irradiance (GHI) and clear sky GHI, accounting for cloud cover variability and clear sky conditions that are often ignored by most machine learning solar forecasters. Additionally, it minimizes a quantile loss during training to produce accurate probabilistic forecasts. In this study, we evaluate the performance of hourly GHI forecasts on eight diverse datasets with varying climates: temperate, cold, arid, and equatorial, for multiple temporal horizons of 2, 3, 6, 12, and 24 h. The model is benchmarked against both climatological persistence for deterministic forecasting and Complete History Persistence Ensemble for probabilistic forecasting. To prove that our model is not location locked, it has been blind tested on four completely different datasets. The results demonstrate that the proposed model outperforms its counterparts across all forecast horizons.

Hourly global solar radiation prediction based on seasonal and stochastic feature

Accurate and detailed solar radiation data play a crucial role in the simulation of building thermal and photovoltaic systems. However, developing a highly precise and dependable solar radiation model using cost-effective data has proven challenging. This work proposes a new attenuation solar radiation model formed by conducting a comprehensive analysis of existing models and gaining new insights into solar radiation's seasonal and stochastic properties. Meanwhile, the model is constructed using easily obtainable surface meteorological parameters. The results demonstrate that the proposed model exhibits good performance in terms of prediction accuracy. Moreover, the majority of existing hourly solar radiation models have been primarily developed for clear-sky conditions. However, there is a growing demand for solar radiation hourly estimations that can uphold a high level of accuracy and reliability even in different weather state. Conversely, the proposed model is developed and validated by more than twenty year's meteorological data encompassing various weather conditions in Japan. It effectively captures the stochastic nature of solar radiation by utilizing turbidity parameters, even on cloudy and rainy days. Additionally, the inclusion of interaction variables significantly enhances its interpretability.

Dual-Path Information Fusion and Twin Attention-Driven Global Modeling for Solar Irradiance Prediction.

Accurate prediction of solar irradiance holds significant value for renewable energy usage and power grid management. However, traditional forecasting methods often overlook the time dependence of solar irradiance sequences and the varying importance of different influencing factors. To address this issue, this study proposes a dual-path information fusion and twin attention-driven solar irradiance forecasting model. The proposed framework comprises three components: a residual attention temporal convolution block (RACB), a dual-path information fusion module (DIFM), and a twin self-attention module (TSAM). These components collectively enhance the performance of multi-step solar irradiance forecasting. First, the RACB is designed to enable the network to adaptively learn important features while suppressing irrelevant ones. Second, the DIFM is implemented to reinforce the model's robustness against input data variations and integrate multi-scale features. Lastly, the TSAM is introduced to extract long-term temporal dependencies from the sequence and facilitate multi-step prediction. In the solar irradiance forecasting experiments, the proposed model is compared with six benchmark models across four datasets. In the one-step predictions, the average performance metrics RMSE, MAE, and MAPE of the four datasets decreased within the ranges of 0.463-2.390 W/m2, 0.439-2.005 W/m2, and 1.3-9.2%, respectively. Additionally, the average R2 value across the four datasets increased by 0.008 to 0.059. The experimental results indicate that the model proposed in this study exhibits enhanced accuracy and robustness in predictive performance, making it a reliable alternative for solar irradiance forecasting.

Forecasting solar irradiance with hybrid classical–quantum models: A comprehensive evaluation of deep learning and quantum-enhanced techniques

Predicting solar irradiance has proven to be a challenging task due to its inherently unpredictable and chaotic characteristics. Although machine learning and deep learning models have demonstrated considerable success in this field, the emergence of quantum computing brings a new layer of potential to this complex problem. This paper delves into these emerging methodologies, focusing specifically on their application in the domain of solar irradiance forecasting. This study investigates the integration of quantum layers into a conventional deep feedforward neural network (FFN) and the development of a fully connected quantum neural network (QNN). A bidirectional long short-term memory (BiLSTM) model is employed as a performance benchmark. Among the various models developed within this study, a few have demonstrated notable performance. For example, an FFN composed of five regular layers and two quantum layers (FFN5L2Q), having a total parameter count of 339, generated an error rate of 6.737%. Conversely, another FFN model, featuring eight regular layers and one quantum layer (FFN8L1Q) and having a total parameter count of 5551, exhibited superior performance with an error rate of 4.254%. When compared to the BiLSTM model, which has 49,666 parameters and an error rate of 3.875%, these quantum-enhanced FFNs display robust and competitive performance. This study, therefore, emphasizes the promising prospects of quantum-integrated techniques in augmenting the precision of solar irradiance prediction models.

Prediction of Solar Radiation using Empirical Models over Lowland Region Nepal

Renewable energy sources are becoming more and more popular as a result of the world's rapidly rising energy needs and the depletion of fossil fuel stocks. Solar energy significantly increases the need for electricity. Due to high energy demands, solar energy most viable out among all of energy sources. Before choosing a location for a solar-powered power plant, it is necessary to predict or anticipate solar energy. For the estimation of the solar energy we use meteorological parameters such as sunshine hour, temperature and relative humidity on selected empirical models to find the empirical coefficients at the study site Biratnagar (Lat. 26° 27' 15" E, Long. 87° 16' 47" N, and alt. 72 m) Nepal. Regression technique is utilized on meteorological parameters including global solar radiation at different models viz. Angstrom-Prescott (A-P) model, Ogunlade model, Gopinathan model and Hargreaves model. Among the four models, the A-P model is better than other models by analyzing the statistical tools. Finally, the empirical constants of A-P model and are found. The finding coefficients can be used to estimate the solar radiation and solar energy at similar geographical locations of Nepal.