Solar Energy Prediction Research Articles

An adaptive hybrid model for day-ahead photovoltaic output power prediction

Abstract Accurate and stable photovoltaic (PV) output power prediction is important for the secure, stable and economic operation of power gird. However, due to the indirectness, randomness and volatility of solar energy, accurate and stable PV output power prediction has become a very challenging issue. To obtain a more accurate and stable prediction results, an adaptive hybrid model combined with improved variational mode decomposition (IVMD), autoregressive integrated moving average (ARIMA) and improved deep belief network (IDBN) is developed to predict day-ahead PV output power. First, the original PV output power is decomposed into some regular and irregular components by IVMD. Second, the regular components are predicted by ARIMA, while irregular components are predicted by IDBN. Third, the final forecasting results is obtained by summing the prediction results of each component. The prediction performance is validated by comparing with some other models. Experimental results illustrate that the presented model can improve the prediction performance of PV output power than other models.

Maximization of wireless sensor network lifetime using solar energy harvesting for smart agriculture monitoring

The wireless sensor networks (WSNs) are used for the real-life implementation of the Internet of Things (IoT) in smart agriculture, smart buildings, smart cities, and online industrial monitoring applications. Generally, traditional WSN nodes are powered by limited energy capacity, non-rechargeable batteries. The WSN lifetime (days) depends upon, duty cycle, type of application deployment, and battery state of charge (SoC) level. We propose an innovative solution to the limited energy availability design problem by utilizing the ambient solar energy harvesting for battery charging of WSN nodes. However, there are many challenges in solar energy harvesting like intermittency of available power, solar energy prediction, thermal issues, solar panel conversion efficiency, and other environmental issues. The objective of this research work is to maximize the WSN network lifetime using solar energy harvesting technique. From our simulation results, it is proved that the sensor network lifetime is increased from 5.75 days to 115.75 days @ 25% duty cycle and higher, ideally up to infinite network lifetime. Furthermore, the network throughput is also increased from 100 K bits/s to 160 K bits/s. in SEH-WSNs.

A novel convolutional neural network framework based solar irradiance prediction method

As an important part of solar power system, photovoltaic grid-connected system and solar thermal system, solar irradiance has the inherent characteristics of variability and uncertainty. Existing data analysis methods are difficult to demonstrate better generalization. Hence, resource planners must be adaptable to accommodate these uncertainties while conducting planning. To improve the accuracy of solar energy prediction and efficiently organize the utilization of solar energy, a novel convolutional neural networks framework has been constructed. Firstly, we have established a convolutional neural network framework for solar prediction based on meteorological data from surrounding sites and different sampling times. Secondly, the chaotic GA/PSO1Genetic Algorithm/Particle Swarm Optimization.1 hybrid algorithm is applied to optimize the hyper parameters of the novel framework, which alleviates the imperfect performance caused by improper hyper parameters. At the meantime, the hybrid algorithm can reduce the manpower and resources of manual parameter adjustment. Excitability of the novel framework has been verified by benchmark tests. In the solar irradiance prediction studies, the annual average Mean Absolute Error of the proposed method is reduced by 0.1463 MJ·m-2 compared with single CNN2Convolutional neural network.2 framework. The annual average Mean Absolute Error of the proposed method is reduced by 49.47%, 47.6%, 20.34%, respectively, compared with ANN3Artificial Neural Network.3, K-means-RBF4kmeans-Radical Basis Function.4 and GBRT.5Gradient Boosted Regression Trees.5 The superiority has been fully illustrated through all the simulation test results. Therefore, the proposed method provides a basis for accurate estimation of solar power, which can promote further development of the whole power system.

An ensemble learning model based on Bayesian model combination for solar energy prediction

To improve the reliability of solar irradiance prediction methods, an ensemble learning method based on the Bayesian model combination has been developed in this paper for solar utilization systems. First, a novel data sampling method has been proposed, including the advantages of clustering and cross validation, which can effectively ensure that the training subsets are different from each other and can cover a variety of different meteorological samples. Second, an ensemble learning model with multiple base learners has been designed. Each training subset is utilized to train the corresponding base learner. Then, a novel Bayesian model combination strategy expands hypothesis space E on Bayesian model averaging, which is applied to frame the combination strategy based on the accuracy of each base learner on the validation set. The prediction values of multiple learners are framed through the model combination strategy. Thus, a novel ensemble learning model based on Bayesian model combination has been established. Finally, experiments are carried out and the proposed method is compared with the Artificial Neural Network (ANN), K-means (Radial Basis Function) RBF, Support Vector Machine (SVM), and Multikernel SVM. The annual average mean absolute error of the ensemble learning method based on Bayesian model combination is reduced by 0.0374 MJ×m−2 compared with the ensemble learning method. The annual average mean absolute error of the proposed method is reduced by 42.6%, 38.2%, 52%, and 48.7%, respectively, compared with ANN, K-means RBF, SVM, and Multikernel SVM. The effectiveness as well as the reliability of the proposed method in solar energy prediction have been found to perform better and have verified our approach.

Solar Radiation Forecasting for Moderate Climatic Zone

The challenge with solar energy prediction is that the solar radiation is intermittent and uncontrollable. Energy forecasting can be used to mitigate some of the challenges that arise from the uncertainty in the resource. Weather data was sourced from India Meteorological Department for Bangalore and Chennai location. This paper provides statistical approach to predict the solar power in future. Analysis was done for different predictive models; Multiple Regression Model is used as we have multiple inputs. The results indicate the prediction of solar radiation has better accuracy during higher irradiation period rather than lower irradiation period.

Deep Learning Neural Networks Trained with MODIS Satellite-Derived Predictors for Long-Term Global Solar Radiation Prediction

Solar energy predictive models designed to emulate the long-term (e.g., monthly) global solar radiation (GSR) trained with satellite-derived predictors can be employed as decision tenets in the exploration, installation and management of solar energy production systems in remote and inaccessible solar-powered sites. In spite of a plethora of models designed for GSR prediction, deep learning, representing a state-of-the-art intelligent tool, remains an attractive approach for renewable energy exploration, monitoring and forecasting. In this paper, algorithms based on deep belief networks and deep neural networks are designed to predict long-term GSR. Deep learning algorithms trained with publicly-accessible Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data are tested in Australia’s solar cities to predict the monthly GSR: single hidden layer and ensemble models. The monthly-scale MODIS-derived predictors (2003–2018) are adopted, with 15 diverse feature selection approaches including a Gaussian Emulation Machine for sensitivity analysis used to select optimal MODIS-predictor variables to simulate GSR against ground-truth values. Several statistical score metrics are adopted to comprehensively verify surface GSR simulations to ascertain the practicality of deep belief and deep neural networks. In the testing phase, deep learning models generate significantly lower absolute percentage bias (≤3%) and high Kling–Gupta efficiency (≥97.5%) values compared to the single hidden layer and ensemble model. This study ascertains that the optimal MODIS input variables employed in GSR prediction for solar energy applications can be relatively different for diverse sites, advocating a need for feature selection prior to the modelling of GSR. The proposed deep learning approach can be adopted to identify solar energy potential proactively in locations where it is impossible to install an environmental monitoring data acquisition instrument. Hence, MODIS and other related satellite-derived predictors can be incorporated for solar energy prediction as a strategy for long-term renewable energy exploration.

Solar panel monitoring and energy prediction for smart solar system

<p>Solar Energy is established as an alternative source of energy known as renewable energy. In a developing country like India, the perspective of Solar Energy is important, as it supports a limitless source of energy. Monitoring and prediction of photo-voltaic energy generation help to reduce the energy loss and empower to utilize more energy. Solar energy prediction is challenging as it depends on the fluctuating solar radiations and climate conditions. The problem statement is to monitor solar panels and predict energy generation for energy management procedure. In this paper, the Internet of Things and Machine Learning algorithms are used as a powerful tool for developing a smart solar system. The metro-logical data such as humidity, temperature and photovoltaic panel data is used as input to forecast solar power generation. For prediction, we examine time-series of solar energy data with Hidden Markov Model. This model considers the probabilistic correlation between previous values to next value in time-series. Experimental results shows that individual panel dead state is located successfully and time-series based solar energy prediction emulate the actual power generation.</p>

Evaluation and forecasting of solar radiation using time series adaptive neuro‐fuzzy inference system: Seoul city as a case study

Given that solar radiation is unpredictable, an accurate solar energy prediction model must be developed. This study aimed to evaluate the changes in solar radiation over the past 37 years in Seoul city. The capability of the adaptive neuro-fuzzy inference system (ANFIS) to forecast solar radiation using chaotic time series inputs was analysed. Results demonstrate the capability of ANFIS to provide a relatively good monthly solar prediction model with a normalised root mean square error of 0.22%, a root mean square error of 55.4, and a coefficient of determination of 0.8. The Jarque–Bera test was implemented as well to test the null hypothesis for the normal distribution of standardised residual. Results support the null hypothesis with P-value = 0.222, which indicates the normal distribution of the standardised residual and its goodness. The standardised residual shows that the model can effectively predict solar radiation on a monthly basis.

Simulation and applications of cumulative anisotropic sky radiance patterns

The prediction of the incident solar energy on a surface under a partially obstructed sky is an important task for most solar energy-related architectural and engineering applications. In previous contributions, Ivanova described an approach that solves this problem using cumulative sky radiance patterns. The method considers the numerical integration of fisheye orthographic projections of these patterns on a considered surface, taking into account the shading caused by all existing obstructing objects in the environment. Further developments of the approach are proposed here to improve the method’s accuracy. The resolution system called SOLARES includes two separate units: a pre-processor and a main program. For each new site, the sky radiance patterns have to be computed once with the pre-processor. The latter prepares daily cumulative sky radiance patterns based on the PVGIS TMY solar database. The direct and diffuse components of the sky radiance are modeled separately. The values of diffuse radiance are generated using an anisotropic model for the luminance and radiance sky distributions. The results are saved in binary files for ulterior use in the main program. The fisheye orthographic projections of the cumulative radiance patterns of the sky hemisphere onto a receiving surface are discretized into matrices of pixels and corresponding direct and diffuse radiance values. The shading objects are projected onto that surface, and the results are saved in a separate shading matrix generated for a specific viewpoint or averaged over the whole surface with the help of the finite-element method. The present numerical method can be used to generate accurate long-term time series of irradiance on any tilted exterior or interior surface in complex urban environments with many neighboring shading surfaces of any tilt and orientation. Additionally, the method allows the visualization of the different system components to enhance architectural or solar applications.

Bottom-up estimation and top-down prediction: Solar energy prediction combining information from multiple sources

Accurately forecasting solar power using the data from multiple sources is an important but challenging problem. Our goal is to combine two different physics model forecasting outputs with real measurements from an automated monitoring network so as to better predict solar power in a timely manner. To this end, we consider a new approach of analyzing large-scale multilevel models for computational efficiency. This approach features a division of the large-scale data set into smaller ones with manageable sizes, based on their physical locations, and fit a local model in each area. The local model estimates are then combined sequentially from the specified multilevel models using our novel bottom-up approach for parameter estimation. The prediction, on the other hand, is implemented in a top-down matter. The proposed method is applied to the solar energy prediction problem for the US Department of Energy’s SunShot Initiative.

Solar energy prediction and task scheduling for wireless sensor nodes based on long short term memory

Traditional wireless sensor networks (WSNs) with limited energy capacity suffer from bounded lifetimes, which have become a major restriction on the development of WSNs. As a common renewable energy, solar energy instead of ordinary batteries can solve energy supply problem in wireless sensor nodes by transferring ambient energy to usable electrical power. In the process of solar energy harvesting, energy prediction is an essential precondition to ensure the reasonable task scheduling of wireless sensor nodes. As the uncertainty of solar energy, we present a long short term memory recurrent neural network (LSTM-RNN) solar energy prediction method, to predict solar energy in the next three days based on historical solar energy collection data and environmental data. Based on energy prediction results, the predictive task scheduling strategy is put forward to improve the performance of wireless sensor nodes. Experimental results show that the proposed LSTM-RNN solar energy prediction method has higher precision and better convergence than other conventional methods. The predictive strategy significantly increases the task completion rate of the wireless sensor node by using the prediction energy information while maintaining an approximate survival time.

Survey of different data-intelligent modeling strategies for forecasting air temperature using geographic information as model predictors

Air temperature modelling is a paramount task for practical applications such as agricultural production, designing energy-efficient buildings, harnessing of solar energy, health-risk assessments, and weather prediction. This paper entails the design and application of data-intelligent models for air temperature estimation without climate-based inputs, where only the geographic factors (i.e., latitude, longitude, altitude, & periodicity or the monthly cycle) are used in the model design procedure performed for a large spatial study region of Madhya Pradesh, central India. The evaluated data-intelligent models considered are: generalized regression neural network (GRNN), multivariate adaptive regression splines (MARS), random forest (RF), and extreme learning machines (ELM), where the forecasted results are cross-validated independently at 11 sparsely distributed sites. Observed and forecasted temperature is benchmarked with the coefficient of determination (R2), root mean square error (RMSE), mean absolute error (MAE), Nash-Sutcliffe’s coefficient (E), Legates & McCabe’s Index (LMI), and the spatially-represented temperature maps. In accordance with statistical metrics, the temperature forecasting accuracy of the GRNN model exceeds that of the MARS, RF and ELM models, as did the overall areal-averaged results for all tested sites. In terms of the global performance indicator (GPI; as a universal metric combining the expanded uncertainty, U95 and t-statistic at 95% confidence interval with conventional metrics, bias error, R2, RMSE) providing a complete assessment of the site-averaged results, the GRNN model yielded a GPI = 0.0181 vs. 0.0451, 0.1461 and 0.6736 for the MARS, RF and ELM models, respectively, which concurred with deductions made using U95 and t-statistic. Spatial maps for the cool winter, hot summer and monsoon seasons also confirmed the preciseness of the GRNN model, as did the 12-monthly average annual maps, and the inter-model evaluation of the most accurate and the least accurate sites using Taylor diagrams comparing the RMSE-centered difference and the correlations with observed data. In accordance with the results, the study ascertains that the GRNN model was a qualified data-intelligent tool for temperature estimation without a need for climate-based inputs, at least in the present investigation, and this model can be explored for its utility in energy management, building and construction, agriculture, heatwave studies, health and other socio-economic areas, particularly in data-sparse regions where only geographic and topographic factors are utilized for temperature forecasting.

Research on Solar Resource Evaluation Method Based on Mathematical Statistics

According to the satellite data provided by the US Renewable Energy Laboratory, the datum of solar resources 25 typical cities in China for 29 years were analyzed by method of principal component analysis and the whole evaluation of this kind of energy was summarized. The indicators for evaluation of solar energy resources, including richness, stability, utilization value and mean of effective sunshine duration, are incorporated into three main parameters. And the score of principal component is considered as the main factor of the resource evaluation. The evaluation and prediction of solar energy in a region are efficient and accurate by this method, which provide guidance for utilization and development of solar energy.

Performance modeling and valuation of snow-covered PV systems: examination of a simplified approach to decrease forecasting error.

The advent of modern solar energy technologies can improve the costs of energy consumption on a global, national, and regional level, ultimately spanning stakeholders from governmental entities to utility companies, corporations, and residential homeowners. For those stakeholders experiencing the four seasons, accurately accounting for snow-related energy losses is important for effectively predicting photovoltaic performance energy generation and valuation. This paper provides an examination of a new, simplified approach to decrease snow-related forecasting error, in comparison to current solar energy performance models. A new method is proposed to allow model designers, and ultimately users, the opportunity to better understand the return on investment for solar energy systems located in snowy environments. The new method is validated using two different sets of solar energy systems located near Green Bay, WI, USA: a 3.0-kW micro inverter system and a 13.2-kW central inverter system. Both systems were unobstructed, facing south, and set at a tilt of 26.56°. Data were collected beginning in May 2014 (micro inverter system) and October 2014 (central inverter system), through January 2018. In comparison to reference industry standard solar energy prediction applications (PVWatts and PVsyst), the new method results in lower mean absolute percent errors per kilowatt hour of 0.039 and 0.055%, respectively, for the micro inverter system and central inverter system. The statistical analysis provides support for incorporating this new method into freely available, online, up-to-date prediction applications, such as PVWatts and PVsyst.

Solar energy harvesting wireless sensor network nodes: A survey

Solar energy harvesting that provides an alternative power source for an energy-constrained wireless sensor network (WSN) node is completely a new idea. Several developed countries like Finland, Mexico, China, and the USA are making research efforts to provide design solutions for challenges in renewable energy harvesting applications. The small size solar panels suitably connected to low-power energy harvester circuits and rechargeable batteries provide a loom to make the WSN nodes completely self-powered with an infinite network lifetime. Recent advancements in renewable energy harvesting technologies have led the researchers and companies to design and innovate novel energy harvesting circuits for traditional battery powered WSNs, such as Texas Instruments Ultra Low Energy Harvester and Power Management IC bq25505 [see https://store.ti.com/BQ25505 for Texas Instruments (TI) Ultra Low Power Boost Charger IC bq25505 with Battery Management and Autonomous Power Multiplexor for Primary Battery in Energy Harvester Applications datasheets (2015).]. In modern days, the increasing demand of smart autonomous sensor nodes in the Internet of Things applications (like temperature monitoring of an industrial plant over the internet, smart home automation, and smart cities) requires a detailed literature survey of state of the art in solar energy harvesting WSN (SEH-WSN) for researchers and design engineers. Therefore, we present an in-depth literature review of Solar cell efficiency, DC-DC power converters, Maximum Power Point Tracking algorithms, solar energy prediction algorithms, microcontrollers, energy storage (battery/supercapacitor), and various design costs for SEH-WSNs. As per our knowledge, this is the first comprehensive literature survey of SEH-WSNs.

Intraday Solar Energy Prediction Using ANN and Validation with Real-Time Roof-Top PV System

Intraday Solar Energy Prediction Using ANN and Validation with Real-Time Roof-Top PV System

Short-term prediction of solar energy in Saudi Arabia using automated-design fuzzy logic systems.

Solar energy is considered as one of the main sources for renewable energy in the near future. However, solar energy and other renewable energy sources have a drawback related to the difficulty in predicting their availability in the near future. This problem affects optimal exploitation of solar energy, especially in connection with other resources. Therefore, reliable solar energy prediction models are essential to solar energy management and economics. This paper presents work aimed at designing reliable models to predict the global horizontal irradiance (GHI) for the next day in 8 stations in Saudi Arabia. The designed models are based on computational intelligence methods of automated-design fuzzy logic systems. The fuzzy logic systems are designed and optimized with two models using fuzzy c-means clustering (FCM) and simulated annealing (SA) algorithms. The first model uses FCM based on the subtractive clustering algorithm to automatically design the predictor fuzzy rules from data. The second model is using FCM followed by simulated annealing algorithm to enhance the prediction accuracy of the fuzzy logic system. The objective of the predictor is to accurately predict next-day global horizontal irradiance (GHI) using previous-day meteorological and solar radiation observations. The proposed models use observations of 10 variables of measured meteorological and solar radiation data to build the model. The experimentation and results of the prediction are detailed where the root mean square error of the prediction was approximately 88% for the second model tuned by simulated annealing compared to 79.75% accuracy using the first model. This results demonstrate a good modeling accuracy of the second model despite that the training and testing of the proposed models were carried out using spatially and temporally independent data.

NEW MODEL FOR SOLAR RADIATION ESTIMATION FROM MEASURED AIR TEMPERATURE AND RELATIVE HUMIDITY IN NIGERIA

Solar radiation prediction is essential for effective and reliable solar power project, predicted solar radiation can be used for accurate solar energy prediction. Solar radiation measurement is not sufficient in Nigeria for various reasons such as maintenance and repair cost, calibration of instrument, and expansive of measuring device. In this paper, adaptive neuro-fuzzy inference system (ANFIS) model was developed to predict the monthly average solar radiation in Nigeria. Air temperature of monthly mean minimum temperature, maximum temperature and relative humidity obtained from Nigerian Meteorological Agency (NIMET) were used as inputs to the ANFIS model and monthly mean global solar radiation was used as out of the model. Statistical evaluation of the model was done based on root mean square error (RMSE) and correlation coefficient R to examine the accuracy of the developed model. The values of RMSE and R for the training data are 0.91315MJ/m 2 and 0.91264MJ/m 2 respectively. The obtained result showed a good correlation between the predicted and measured solar radiation which proves ANFIS to be a good model for solar radiation prediction. http://dx.doi.org/10.4314/njt.v36i3.35

NEW MODEL FOR SOLAR RADIATION ESTIMATION FROM MEASURED AIR TEMPERATURE AND RELATIVE HUMIDITY IN NIGERIA

Solar radiation prediction is essential for effective and reliable solar power project, predicted solar radiation can be used for accurate solar energy prediction. Solar radiation measurement is not sufficient in Nigeria for various reasons such as maintenance and repair cost, calibration of instrument, and expansive of measuring device. In this paper, adaptive neuro-fuzzy inference system (ANFIS) model was developed to predict the monthly average solar radiation in Nigeria. Air temperature of monthly mean minimum temperature, maximum temperature and relative humidity obtained from Nigerian Meteorological Agency (NIMET) were used as inputs to the ANFIS model and monthly mean global solar radiation was used as out of the model. Statistical evaluation of the model was done based on root mean square error (RMSE) and correlation coefficient R to examine the accuracy of the developed model. The values of RMSE and R for the training data are 0.91315MJ/m2 and 0.91264MJ/m2 respectively. The obtained result showed a good correlation between the predicted and measured solar radiation which proves ANFIS to be a good model for solar radiation prediction.  http://dx.doi.org/10.4314/njt.v36i3.35

The global solar radiation estimation and analysis of solar energy: Case study for Osmaniye, Turkey

ABSTRACTThis paper investigates the prediction of solar radiation model and actual solar energy in Osmaniye, Turkey. Four models were used to estimate using the parameters of sunshine duration and average temperature. In order to obtain the statistical performance analysis of models, the coefficient of determination (R2), mean absolute percentage error (MAPE), mean absolute bias error (MABE), and root mean square error (RMSE) were used. Results obtained from the linear regression using the parameters of sunshine duration and average temperature showed a good prediction of the monthly average daily global solar radiation on a horizontal surface. In order to obtain solar energy, daily and monthly average solar radiation values were calculated from the five minute average recorded values by using meteorological measuring device. As a result of this measurement, the highest monthly and yearly mean solar radiation values were 698 (April in 2013) and 549 (2014 year) W/m2 respectively. On an annual scale the maximum global solar radiation changes from 26.38 MJ/m2/day by June to 19.19 MJ/m2/day by September in 2013. Minimum global solar radiation changes from 14.05 MJ/m2/day by October to 7.20 MJ/m2/day by January in 2013. Yearly average energy potential during the measurement period was 16.53 MJ/m2/day (in 2013). The results show that Osmaniye has a considerable solar energy potential to produce electricity.