Search

The fast growth of solar photovoltaic (PV) power generation raises the concern of grid stability due to intermittency. The traditional solution based on the broad installation of energy storage systems (ESS) is considered to be costly. Hence, the latest research focuses on the active power ramp rate control (PRRC) that decouples the heavy reliance on ESS. While power-forecasting-based PRRC can be a promising solution, the relevant research is still limited and does not cover the prediction error's negative effect on the power ramp rate control. With the above, this paper investigates the over-regulation issue that is induced by inaccurate prediction and can lead to power losses. More importantly, this paper proposes a novel PRRC scheme incorporating a Gaussian confidence estimator and best-performed deep neural networks (DNN) predictor to implement more accurate control actions while overcoming the over-regulation problem. The experimental tests show that the proposed scheme can simultaneously achieve a more promising PRRC performance and relatively higher energy harvest efficiency compared to the prior arts. • A novel predictive PRRC scheme based on big meteorological data is proposed. • The proposed approach eliminates the reliance on ESS while simultaneously achieving promising PRRC performance and high energy harvest efficiency. • The PRRC's overregulation problem that can bring additional power losses is firstly investigated in this study. • A novel prediction confidence estimator based on Gaussian process is proposed to address the overregulation issue by monitoring the prediction's reliability. • A comprehensive study on the popular DNN-based solar forecasting models is implemented to select the best-performed predictor suitable for sequential meteorological data and the power system's PRRC task. [ABSTRACT FROM AUTHOR]

Highlights • This paper proposes a few novel MPPT techniques, which include Adaptive Cuckoo Search Optimization Algorithm (ACOA), General Regression Neural Network GRNN) with Fruit fly Optimization algorithm (FFOA), and Dragonfly Optimization Algorithm (DFO) to track the MPP under various weather condition. • The proposed techniques enhance the performance of the PV system, save the computational time and greatly reduce the oscillation around the global maximum power point. • For the validation of the proposed techniques, comparative analysis of their results with the Bio-inspired Particle Swarm Optimization (PSO), Cuckoo Search Optimization (CS), Artificial Bee Colony Algorithm (ABC) and PSO Gravitational search Optimization (PSOGS) is presented. • The comparison shows that proposed techniques are better in term of quick power tracking, stability, and high efficiency under various weather conditions, and also demonstrates that the proposed techniques can efficiently locate the GM (global maxima) under the PS and Dynamic Partial Shading (DPS) conditions. Abstract In PV systems, the non-uniform irradiance and diversified unpredictable weather conditions fall into the category of Partial Shading (PS). Under PS, it is challenging for PV systems to obtain the maximum output through Maximum Power Point Tracking (MPPT), i.e., the parameters of the controller are adjusted online to yield the maximum power. In the literature, various techniques have been proposed to track the MPP (Maximum Power Point) under the uniform irradiance. On the contrary, few techniques have been proposed to efficiently track MPP under PS. In this paper, a few novel MPPT techniques have been proposed, which include Adaptive Cuckoo Search Optimization Algorithm (ACOA), General Regression Neural Network GRNN) with Fruit fly Optimization algorithm (FFOA), and Dragonfly Optimization Algorithm (DFO) to track the MPP under various weather condition. The proposed techniques enhance the performance of the PV system, save the computational time and greatly reduce the oscillation around the global maximum power point. For the validation of the proposed techniques, comparative analysis of their results with the Bio-inspired Particle Swarm Optimization (PSO), Cuckoo Search Optimization (CS), Artificial Bee Colony Algorithm (ABC) and PSO Gravitational search Optimization (PSOGS) is presented. The comparison shows that the proposed techniques are better in term of quick power tracking, stability, and high efficiency under various weather conditions. The comparison also demonstrates that the proposed techniques can efficiently locate the GM (global maxima) under the PS and Dynamic Partial Shading (DPS) conditions. Furthermore, statistical analysis is presented to check the stability, sensitivity and robustness of the proposed techniques. [ABSTRACT FROM AUTHOR]

[Display omitted] • A neural network is used to detect faults in photovoltaic modules. • According to the shape of the current vs voltage curve, the faults are classified. • The training uses synthetic curves and the test is performed also on experimental data. • The points are normalized and resampled radially to improve the results. • With two hidden layers of 100 neurons each, the hit rate reaches more than 98.5%. In this paper a model-based procedure for fault detection and diagnosis of photovoltaic modules is presented. A four-layered feedforward artificial neural network learns the correlation between the features of the current vs. voltage curve and the environmental variables, which are the irradiance and the temperature. This correlation describes the behavior of the module at normal conditions. Moreover, the effect of anomalous variation of some parameters is learnt and correlated to the shape of the same curve, thus associated to a specific failure mechanism and to some assigned ranges quantifying the fault severity. The neural network is trained by using synthetic curves simulated by employing the single diode model and some well assessed and validated translation formulae. The obtained results over the simulated set of curves with different failures allow to achieve a classification error lower than 1.5%. The proposed approach has been also validated for detecting anomalous increases of the series resistance in a large experimental set of curves; in this case, a classification error of 2.7% has been achieved. [ABSTRACT FROM AUTHOR]

With the significant increase in photovoltaic (PV) electricity generation, more attention has been given to PV power forecasting. Indeed, accurate forecasting allows power grid operators to better schedule and dispatch their assets, such as energy storage systems and reserve. In this paper, a hybrid deep learning model and a convolutional neural network with memory is proposed, to provide intraday (2 h) solar irradiance forecasts using sequentially-collected images from public webcams. The performance of the proposed model is compared to those of a standard time-series forecast models, a linear regression as well as state-of-the-art neural networks. All models are trained and tested using images collected from two webcams on EPFL's campus for just over a year. The results show that the proposed method outperforms all other models and matches the state-of-the-art methodology while providing simplicity of implementation and efficient computation. • A novel method for estimating global horizontal irradiance is proposed. • The method uses images public webcams instead of all-sky cameras. • The method combines images from different cameras to improve forecasts' accuracy. • The method can be extended to include other types of data (meteorological data). • This simple method matches state-of-the-art methods. [ABSTRACT FROM AUTHOR]

[Display omitted] • Novel physics-based surrogate model for bifacial tandem cell with near instantaneous speed. • Identified key geometric and material parameters for tandem device output at different albedo. • Presented pathway to improve bifacial equivalent efficiency (>2%) from 6 geometric parameters. The perovskite solar cell has attracted considerable interest for tandem device applications due to its bandgap tunability and high efficiency. Although improvements are made to all perovskite tandem cells, their power output is still not close to that of other tandem devices. One possible way to improve the power output is to absorb the rear incident light as a bifacial device. Unlike all previous work, where the focus for optimizing the design of a bifacial tandem device is light absorption, this paper also considers the recombination losses. The six most influential geometric parameters to the tandem device power output were identified through sensitivity analysis. With a physics-based machine learning methodology, we significantly cut down the computational cost to predict the optimal design for a bifacial four-terminal all-perovskite tandem solar cell under nine different albedo conditions. Furthermore, we quantify the change in the recombination losses for the bottom sub-cell under two scenarios: (1) at optimized configuration; (2) at different albedo. It is found that the optimal hole transport layer thickness for the bottom sub-cell increases at higher albedo and the opposite is true for the perovskite layer. A feasible pathway, balancing light absorption and recombination, is presented to improve the tandem device's power output. [ABSTRACT FROM AUTHOR]

Solar energy is the feedstock for various applications of renewable energy systems, thus, the necessity of calculating and using global tilted irradiance is acknowledged for the computations of the performance and monitoring of Photovoltaic (PV) Parks and other solar energy applications. Thus, the aim of our research is to develop a model for the correlation of diffuse fraction ( k d ) and the clearness index ( k t ), that can then be used for the evaluation of the diffuse irradiance given the global irradiance. In a companion paper, existing simple empirical models were reviewed and compared based on 10 years of data from Cyprus and then, analytical approaches for the computation of diffuse fraction were employed, where solar altitude was introduced as an additional parameter in the calculations. In the present paper, the same dataset was used, and three additional parameters were introduced to the calculations: global irradiance on the horizontal plane, extraterrestrial irradiance on the horizontal plane and the time of the day. These parameters were chosen due to the strong dependence of the diffuse fraction/clearness index correlation to the season/day of the year and time of day. Due to the non-linear influence of these parameters to the k t – k d correlations and the additional interaction between them, the employment of analytical methods is not applicable. Thus, non-parametric regression analysis was adopted, using supervised machine learning methodologies such as Artificial Neural Networks, which are able to learn the key information patterns from multivariate input. Comparing the non-parametric regression to the analytical models developed in the companion paper, it is shown herewith that the accuracy of the models was slightly improved. The statistical indicators MBE, RMSE and R 2 of the best fit model were −4.69%, 21.54% and 0.90 respectively. [ABSTRACT FROM AUTHOR]

Artificial Intelligence (AI) is increasingly playing a significant role in the design and optimization of renewable energy systems. Many AI approaches and technologies are already widely deployed in the energy sector in applications such as generation forecasting, energy efficiency monitoring, energy storage, and overall design of energy systems. This paper provides a review of the applications of key AI techniques on the analysis, design, optimization, control, operation, and maintenance of Solar Tower systems, one of the most important types of Concentrating Solar Thermal (CST) systems. First, key AI techniques are briefly described and relevant examples of their application to CST systems in general are provided. Subsequently, a detailed review of how these AI techniques are being used to advance the state of the art of solar tower systems is presented. The review is structured around the different subsystems of a solar tower system. [ABSTRACT FROM AUTHOR]

• High-rise design in dense urban areas is addressed using artificial intelligence. • Part 1 presents background, methodology, setup, and machine learning results. • Two design scenarios focus on daylight metrics using parametric modelling. • 31 models reported very high and 9 models reported high prediction accuracies. • The importance of surrogate modelling is discussed for sustainable metropolises. Designing high-rise buildings is one of the complex tasks of architecture because it involves interdisciplinary performance aspects in the conceptual phase. The necessity for sustainable high-rise buildings has increased owing to the demand for metropolises based on population growth and urbanisation trends. Although artificial intelligence (AI) techniques support swift decision-making when addressing multiple performance aspects related to sustainable buildings, previous studies only examined single floors because modelling and optimising the entire building requires extensive computational time. However, different floor levels require various design decisions because of the performance variances between the ground and sky levels of high-rises in dense urban districts. This paper presents a multi-zone optimisation (MUZO) methodology to support decision-making for an entire high-rise building considering multiple floor levels and performance aspects. The proposed methodology includes parametric modelling and simulations of high-rise buildings, as well as machine learning and optimisation as AI methods. The specific setup focuses on the quad-grid and diagrid shading devices using two daylight metrics of LEED: spatial daylight autonomy and annual sunlight exposure. The parametric model generated samples to develop surrogate models using an artificial neural network. The results of 40 surrogate models indicated that the machine learning part of the MUZO methodology can report very high prediction accuracies for 31 models and high accuracies for six quad-grid and three diagrid models. The findings indicate that the MUZO can be an important part of designing high-rises in metropolises while predicting multiple performance aspects related to sustainable buildings during the conceptual design phase. [ABSTRACT FROM AUTHOR]

• Machine Learning is used for energy use prediction of a solar-assisted heat pump. • Attention mechanism and time series decomposition is used to improve model accuracy. • Various combinations of input features are compared. • Results can provide valuable insights for developing Neural Network models. With improved insulation of building envelopes and the use of low-temperature space heating systems, the share of energy use for domestic hot water (DHW) production in buildings has increased significantly, and nearly become the most energy-expensive service in modern buildings. Early prediction of the energy use for DHW is required for many advanced applications such as smart control, demand-side management, and optimal operation of electric or heat storage. However, predicting energy use of the solar-assisted water heating system is more challenging than typical DHW systems, as it is strongly affected by two stochastic phenomena, demand pattern and solar radiation. Given the increasing use of solar-assisted water heating systems, this paper aims to evaluate the potential to predict energy use in such systems using a novel machine learning approach. In this novel model, a Long-Short Term Memory (LSTM) neural network is enhanced by (1) implementing the attention mechanism, a recent development in deep learning inspired by human vision to pay selective attention to the input data, and (2) decomposition of input data into sub-layers. The performance of simple LSTM neural network, Attention-based LSTM neural network (ALSTM) and Attention-based LSTM using decomposed data (ALSTM-D) are compared to a Feed-Forward neural network as a baseline model. Results show that LSTM, ALSTM and ALSTM-D models have a Mean Absolute Error (MAE) of 25%, 28% and 41% lower than Feed-Forward model, respectively. These results indicate the superior performance of the proposed ALSTM-D model over conventional models for solar-assisted DHW systems. [ABSTRACT FROM AUTHOR]

• An asphalt solar collector is simulated by a novel CFD model. • Several experiments are carried out to validate the numerical results. • An artificial neural network model is developed to predict the performance of the system. • Effective parameters in the variation of the outlet water temperature and efficiency are discussed. This paper presents a numerical model, development, and experimental validation of an Asphalt Solar Collector (ASC) which subsequently is modeled by Artificial Neural Network (ANN). In this research, 2 m long galvanized pipes are connected in parallel and embedded into an asphalt slab with 1 m 2 area. The slab is buried in the ground to have thermal heat transfer with adjacent soil and to resemble real conditions. Several experiments are carried out in both warm and cold months of the year with different water flow rates from 9:00 to 17:00. The ASC is modeled with CFD techniques and the desired parameters effect are independently studied based on a reference condition. An ANN model is proposed to expand the parametric study and reduce the high computational cost of numerical modeling. The inputs of the ANN include the design and the operating parameters and the outlet water temperature is considered as the output. These models are capable of evaluating the performance of the ASC at every hour during the day and under different boundary conditions. The parametric study shows that improvement of surface absorptivity and thermal conductivity of asphalt leads to a higher increase of the daily efficiency in August, however, the inlet water temperature has the same effect in November. The maximum water temperature difference and the thermal efficiency of the ASC can reach up to 24 °C and 45% in August and 14 °C and 35% in November respectively. [ABSTRACT FROM AUTHOR]

• The specific heat capacity (SCH) of CuO/water nanofluids was accurately modelled using SVR and ANN models. • The SVR model has a higher prediction accuracy than the ANN model. • Both SVR and ANN models performed much better than the existing empirical relations. This paper presents the modelling of the specific heat capacity (SHC) of CuO/water nanofluids using a support vector regression (SVR) and artificial neural network models (ANN). The models presented were developed from the experimental data of SCH of CuO nanoparticles, the volume fractions of CuO nanoparticles and fluid temperature. The volume fraction of CuO nanoparticles considered ranges from 0.4 to 2% while the temperature range includes 293–338 K. The results obtained revealed that the SVR model exhibits slightly higher accuracy compared to the ANN model. However, both the SVR and ANN models clearly demonstrate better prediction performance for the SHC of CuO/water nanofluids compared to the existing theoretical models. The results obtained in this study proves that machine learning models provide a more accurate estimation of SHC of CuO/water nanofluids and they are recommended for heat transfer calculations due to their superior accuracy. [ABSTRACT FROM AUTHOR]

Abstract: Photovoltaic (PV) generation systems (PGSs) have become an attractive option among renewable energy sources because they are clean, maintenance-free and environmental friendly. For PGSs, a simple and fast maximum power point tracking (MPPT) algorithm is essential. Although the static tracking efficiency of conventional MPPT method is usually high, it drops noticeably in case of rapidly changing irradiance conditions. In this paper, two fast and accurate digital MPPT methods for fast changing environments are proposed. By using piecewise line segments or cubic equation to approximate the maximum power point (MPP) locus, two high-speed, low-complexity MPPT techniques can be developed. To make the developed system more convenient for common PGS users, neural network (NN)-based program which can be used to calculate the parameters of the emulated MPP locus is also developed and embedded into the proposed digital MPPT system. Theoretical derivation and detailed design procedure will be provided in this paper. The advantages of the proposed system include low computation requirement, fast tracking speed and high static/dynamic tracking efficiencies. To validate the effectiveness and correctness of the proposed methods, simulation and experimental results of a 230W PV system will also be provided. [Copyright &y& Elsevier]

Abstract: Recently, various time series models have been proposed to predict solar radiation, for instance the ARIMA (autoregressive integrated moving average) model and neural networks. Before building a model for the data, however, it is advisable to check whether the data suggest this type of modeling. More specifically, a nonlinearity test is suggested before further analysis with the linear or nonlinear tools are to be applied. In this paper, we test the presence of nonlinearity in the solar radiation time series by the method of surrogate data. The surrogate test method used in this paper is based on evaluation of the differences between the original time series and the linear model that best approximates it. Nonlinearity tests are carried out for four data sets including 5-min, hourly, daily and monthly global solar radiation time series from the UO (University of Oregon) Solar Radiation Monitoring Laboratory. The test statistics show that the 5-min, hourly, daily global solar radiation time series exhibit apparently nonlinearity while the monthly time series does not. [Copyright &y& Elsevier]

Abstract: Solar irradiation is one of the major renewable energy sources and technologies related with this source have reached to high level applications. Prediction of solar irradiation shows some uncertainties depending on atmospheric parameters such as temperature, cloud amount, dust and relative humidity. These conditions add new uncertainties to the prediction of this astronomical parameter. In this case, prediction of generated electricity by photovoltaic or other solar technologies could be better than directly solar irradiation. In this paper, firstly, Artificial Neural Networks (ANNs) methodology is applied to data obtained from a 750W power capacity of solar PV panel. The main objective of this paper is to determine time horizon having the highest representative for generated electricity prediction of small scale solar power system applications. It is seen that 5min time horizon gives the best solar power prediction for short term and 35min could be used for medium terms in April. In addition, these time horizons have increased to 3 and 40min for very short time and medium time prediction respectively during August. During April and August Root Mean Square Errors (RMSEs) between measured and testing values changed between 33–55W and 37–63W ranges respectively. Especially, during August for solar irradiation, stationary conditions are observed and these situations let ANN predict easily generated electricity from 30 to 300min ahead. [Copyright &y& Elsevier]

• Optimum design and economic analysis of HRES for rural electrification in Korkadu. • RES are solar PV, wind turbine and bio-diesel generators. • Estimates the load forecasting for selected district in the Pre- HOMER analysis. • Desired HRES has to meet forecasted load demand for reliable electrification. This paper validates the optimal design and techno-economic feasibility of hybrid renewable energy system (HRES) for rural area electrifying applications. Plan to a design of improved performance electrification system through village owned renewable resources, such as solar irradiations, wind speed and bio mass etc. The selected HRE system has to meet out electrical needs in optimum performance manner. Hear conducted a case study on remote village Korkadu is located in Union Territory of Pondicherry, India. The expected village demand of 179.32 Kwh/day and peak of 19.56 Kw was met with proposed HRE structure, which is consists of solar PV array, wind turbine, Bio mass power generator and Battery backup system in effectively. Load growth of the village was predicted through artificial neural network (ANN-BP) feed-back propagation and Levenberg-Marguardt (LM) data training optimum technique. Encounter the optimum performance of different HRE configuration was evaluated over by HOMER software. System's economic dispatch was analysed through various dispatch strategy and come across the proposing companied dispatch strategy has more economical and performance benefits as total NPC of system as INR 1.21 million, one unit energy generation cost as INR 13.71 and annual battery throughput as 36.648 KWh/yr. This study also expresses the comparison analysis between proposed HRES structure performance with basic utility grid extension. The consequence of the proposed work shows the HRES in remote location can be a cost effective solution for sustainable development of rural regions. [ABSTRACT FROM AUTHOR]

• GHI forecast bias is related to the Sun-Earth geometry and atmospheric variables. • ANN proven to be useful in correcting GHI predictions of NWP models. • Optimised ANN configurations according to clearness index result in higher accuracy. • Lower performance of IFS model for cloudy conditions is improved through the ANN. This paper proposes a corrective algorithm for improving the accuracy of global horizontal irradiation (GHI) forecasts obtained from the numerical weather prediction (NWP) model of the European Centre for Medium-range Weather Forecast (ECMWF). Firstly, GHI forecasts were compared with experimental values from two ground-based stations located at the south of Portugal (Évora and Sines), and the influence of Sun-Earth geometry and atmospheric variables on the differences between predictions and measurements was analysed in order to identify the most relevant parameters. These differences are shown to be correlated mainly with the clearness index, solar zenith angle, mean air temperature, relative air humidity and total water column. Since the ECMWF model directly or indirectly provides all these variables, it is possible to estimate the bias of the predicted GHI as a function of forecast data taking as reference the measurements, which means that an algorithm based on correlations of such parameters can be used to correct forecasts in an operational time horizon. With that goal, an artificial neural network (ANN) based algorithm was developed in this work to improve GHI predictions, including also as input the global solar irradiation predicted by a reference clear sky model. The internal structure of the ANN was optimised, and a spatial and temporal downscaling procedure was also developed to obtain half-hour irradiation values. This algorithm was tested against the original ECMWF forecasts and a persistence model for four locations with different orography and climate as well as for various sky conditions (overcast, partly cloudy and clear sky), showing that it successfully improves the model predictions. Higher values of a Global Performance Index (GPI) based on seven statistical indicators and Forecast Score (FS) were found for the algorithm simulations, e.g. respectively 1.066 and 0.348 when considering all the locations and cloud cover conditions, while for the original ECMWF predictions the GPI is −1.874 and the FS is 0.282. This algorithm is useful if integrated into energy management tools of solar energy systems, namely low/medium temperature solar thermal and photovoltaic systems. [ABSTRACT FROM AUTHOR]

• A problem of global solar regression from satellite data and physical models is tackled. • Machine Learning regression techniques are applied. • Comparison of Neural networks, SVRs and Gaussian Processes is carried out. • Machine Learning techniques improves the performance of physical models in solar radiation estimation. In this paper we evaluate the performance of several Machine Learning regression techniques in a problem of global solar radiation estimation from geostationary satellite data. Different types of neural networks, Support Vector Regression and Gaussian Processes have been selected as regression techniques to be evaluated, due to their good performance in similar problems in the past. The study area is located in the surroundings of the radiometric station of Toledo, Spain. In order to train the regression techniques considered, one complete year of hourly global solar radiation data is used as the target of the experiments, and different input variables are considered: a cloud index, a clear-sky solar radiation model and several reflectivity values from Meteosat visible images. To assess the results obtained by the Machine Learning algorithms, we have selected as a reference three different physical-based methods, a model based on the Heliosat-2 method (Heliosat-2), the Copernicus Atmosphere Monitoring Service (CAMS) and the SolarGIS model (Soevaluate the performance of Machine Learning regressors when the physical models are included as input variables, in a class of post-processing of these physical approaches. The results obtained show the capacity of Machine Learning regressors to obtain reliable global solar radiation estimation by using satellite measurements. [ABSTRACT FROM AUTHOR]

• A modified structure of MPPT regulator of photovoltaic generators is presented in this paper. • A new adaptive PSO-RBFNN and Fuzzy Logic Controller based on the tracking of maximum power point is also presented. • The adaptive PSO- RBFNN with fuzzy controller for MPPT of PV is validated by stimulation and experimentation using MyRio card-1900. • A comparative study between the results obtained with the proposed MPPT with PSO-RBF Neural Network algorithm and adaptive neuro-fuzzy inference system technique. Given the nonlinear feature of a photovoltaic generator, a maximum power point tracking algorithm (MPPT) is required in a photovoltaic system leading to maximum power point (MPP) operation and maximizing the power generated. The tracking MPP techniques are based on an actual or estimated research mechanism using experimental data. Conventional MPPT techniques like perturbe and observe (P&O), incremental conductance, etc., are good enough to track the maximum power for the PV systems, but they are less stable, more oscillating around the MPP. Generally, techniques based on the estimated research mechanisms, such as the Artificial Neural Network (ANN), the Adaptive Neuro-Fuzzy Inference System (ANFIS), and the Radial Basis Function Neural Network (RBFNN), etc., are supervised automatic learning techniques, which aims to create a model for an unknown function in order to find a relationship between input data and output data. In the case of RBF Neural Network, the center of the radial base function, the variance of the radial function base and the weight must be chosen. If these variables are not chosen appropriately, the RBF neural network can degrade the validity and accuracy of the modeling. On the other hand the RBF network suffers from a growth in the size of the hidden layer comparable to that of a set of learning data which also implies more computational time. The solution of these two problems is the motivation of this research. The PSO algorithm is used to optimize the parameters of the RBFNN by introducing a new adaptive strategy of particle swarm optimizer to dynamically adjust the inertia weight factor ω and the new velocity v id t + 1 with a new μ coefficient. The obtained results based on RBFNN hybrid approach with PSO (PSO-RBFNN approach) were compared with the results obtained with the adaptive Neuro-Fuzzy Inference System (ANFIS). The experimental test bench of the PSO-RBFNN approach has been implemented using a MyRio card, which prove the good performances of the new proposed technique in terms of the average relative errors of the learning, test and control data, for the model PSO-RBFNN which converge approximately to 0.26%, 0.294% and 0.8% respectively, and energy efficiency MPPT in the case of atmospheric parameters varying over time can reach 99.04%. [ABSTRACT FROM AUTHOR]

Highlights • 3D-CNN is used to extract features of ground-based cloud images for DNI prediction. • The automatically extracted features include texture and temporal information of clouds. • Multiple inputs including image features and DNI data are used to build forecasting model. • The proposed model achieves the best result and outperforms obviously the basic model. Abstract Cloud cover and cloud motion have a large impact on solar irradiance. One of the effective ways for direct normal irradiance (DNI) prediction is to use cloud features, which has been extensively studied. A Convolutional Neural Network (CNN) has the advantage of automatic features extraction by using strong computing capabilities. In this paper, a novel 3D-CNN method is proposed by processing multiple consecutive ground-based cloud (GBC) images in order to extract cloud features including texture and temporal information. The resulting features and the DNI data are then used to establish a DNI forecasting model. The experiments are carried out to evaluate the performance of the proposed forecasting method by using the data from January 1, 2013 to December 31, 2014. The experimental results show that the proposed method coupled with the multilayer perceptron (MLP) model achieves forecast skill of 17.06% for 10-minute ahead DNI prediction. [ABSTRACT FROM AUTHOR]

Highlights • Performance prediction of SPV modules. • Predictive models include electrical equivalent models and neural networks. • Cascade implementation of 7 parameter and feed forward neural network. • MATLAB GUI developed to predict the SPV module performance. • Five I-V curve parameters determined as a function of temperature and irradiance. • Simultaneous comparison of various predictive models. • Analysis for five SPV module technologies. Abstract Before a photovoltaic (PV) system is installed, a prerequisite modelling and performance analysis are carried out which estimates the performance parameters and reliability in operation of PV system. This paper proposes a neural network approach to performance prediction of PV Modules. Here the feed forward neural networks are used to predict I-V curve parameters as a function on input irradiance and temperature. K Fold cross-validation is used to validate model accuracy for determination of I-V curve parameters for five different technology modules, i.e., CdTe, CIGS, MICROMORPH, MUTICRYSTALLINE, MAXEON. A comparison is also drawn between the neural network predictor and the existing modeling procedures available. Model parameters have been determined following iterative, analytical and regression of known data points for seven and five parameter model of PV Modules. Among the electrical equivalent models, the seven parameter model is the most efficient model for performance prediction however commutation of model parameters in complex and tedious. Neural network model simplifies the computational process at the expense of higher error variance in comparison to electrical equivalent models. Further, a cascade implementation of the above two is designed and tested on Multicrystalline and Maxeon technology modules for higher model accuracy. The results obtained, verify the proposed cascaded model to be the most efficient model in comparison to independent models where mean bias error deviations are less than ±1% and error variance is reduced significantly. Also, a MATLAB based graphical user interface (GUI) is developed that can be used to predict the performance based on the analysis carried out. The proposed model is tested against a set of operating conditions and compared to the actual experimental values obtained using outdoor tests. [ABSTRACT FROM AUTHOR]

Highlights • We proposed a new method for I-V curve prediction of photovoltaic modules. • The proposed method is based on two artificial neural networks. • The proposed method is general for types of photovoltaic modules. • The accuracy of the proposed method is almost 99%. Abstract This paper presents a new I-V curve prediction method using artificial neural networks. The proposed method is based on two artificial neural networks namely generalized regression artificial neural network and cascaded forward neural network. An experiment is set up so as to extract a dataset that includes records of solar radiation, ambient temperature, current and voltage for different photovoltaic modules. The developed model is a general model for all photovoltaic modules whereas the inputs of the model are solar radiation, ambient temperature and datasheet specifications of photovoltaic module (open circuit voltage and short circuit current). Matlab is used to train, test and validate the proposed model. Moreover, the proposed model is validated experimentally. The results show that the proposed model has a high accuracy in predicting I-V curves with average mean absolute percentage error, mean bias error and root mean square error of 1.09%, 0.0229 A and 0.0336 A respectively. Such a model is very helpful in generating I-V curves for different photovoltaic modules. [ABSTRACT FROM AUTHOR]

An Accurate forecast of PV output power is essential to optimize the relationship between energy supply and demand. However, it is a challenging task due to the intermittent nature of solar PV output and the effect of number of meteorological variables on it. In this paper, a multivariate neural network (NN) ensemble forecast framework is proposed. First multiple neural predictors are trained with input data from meteorological variables and then accurate predictors are combined with a Bayesian model averaging (BMA) technique. To identify the best performing framework, three different NN ensemble networks are created, namely feedforward neural network (FNN) , Elman backpropagation network (ELM) and cascade-forward backpropagation (NewCF) network and trained with three different training techniques. The real time recorded solar PV data along with meteorological variables of the University of Queensland’s solar facility from 2014 to 2015 is used. To validate the forecast framework, one day ahead (24 h) forecasts are selected for different seasons. The results show that the proposed ensemble framework substantially improves the forecast accuracy of PV power output as compared benchmark methods, particularly for short term forecasting horizons. [ABSTRACT FROM AUTHOR]

Integrating photovoltaic (PV) produced electricity into the electric power system is proving to be a growing challenge due to its fluctuating nature. The combination of more rigid regulation for feed-in of PV production and steadily rising electricity prices means that battery systems are becoming more attractive to private households as a way of upping their self-consumption. At the same time, batteries make the household’s electricity purchasing strategy more complex. For these reasons, control concepts are required for PV + battery systems that ensure grid-friendly operation as well as considering the household’s primary objectives. This paper presents a forecast-based modelling approach for the operation of a battery in combination with a grid-connected PV system. PV production and electricity demand are forecasted on an hourly time-resolution using artificial neural networks (ANN). The battery charging and discharging is optimized to maximize self-consumption, and additionally a variable feed-in tariff is considered to incentivize a grid-friendly operation. The developed model was applied for a household with 3300 kWh electricity consumption equipped with a 5 kWp PV system and a 5 kWh battery. For this case, we show that the model enables a grid-friendly operation of the battery as well as an intensified usage. However, the inevitable forecasting errors lead to overall lower economic benefits for the consumer in comparison with a simple strategy that only maximizes self-consumption. Considering the inaccuracy of forecasting, we conclude that if a grid-friendly integration of PV + battery systems is to be promoted in the future, households should be provided with better forecasting data or offered other incentives to compensate their lost benefits. [ABSTRACT FROM AUTHOR]

In the solar absorption cooling systems, concentrating collector can be adopted to achieve higher driven temperature and improve the system performance. The Compound Parabolic Concentrator (CPC) is a good option with both concentrating and stationary features. With higher working temperature, solar absorption cooling system with concentrating collector has more choices of absorption chillers than the system with non-concentrating collector. The aim of this paper is to find the better chiller choice for solar absorption cooling systems with CPC. The analyzed LiBr-water absorption chillers include the single effect chiller, the double effect absorption chiller and a novel variable effect chiller built by the author. Simulation is carried out in TRNSYS. Model of the variable effect absorption chiller is first derived from MATLAB artificial neural network (ANN) toolbox based on experimental data, and then built in TRNSYS. Performance of the three systems is calculated under the same condition. Performance of the three systems is compared to get the better choices. The impacts of the solar collector area, cutoff driven temperature and storage tank volume on the solar cooling fraction, auxiliary heat input, average chiller COP and average solar efficiency are analyzed. The variable effect system has high solar cooling fraction, low auxiliary heat input and high solar efficiency. The single effect system has medium solar fraction, high auxiliary heat input and medium solar efficiency. The double effect system has low solar cooling fraction, low auxiliary heat input and low solar efficiency. [ABSTRACT FROM AUTHOR]

In this paper an optimal performance of three phase induction motor drives a centrifugal water pump and fed from photovoltaic (PV) system without storage elements during starting and running is presented. A three level three phase inverter is used to convert the DC voltage from the PV array to a variable voltage and frequency to supply the three phase induction motor. The output voltage and frequency of the inverter are controlled to extract the maximum power from solar panel during running at different levels of irradiance and temperatures using a Teaching Learning Based Optimization (TLBO) algorithm with minimum motor losses. The ratio of voltage magnitude and frequency is held within rated values to avoid saturation and motor overheating. The rating of PV array is chosen to develop the rated power of the pump at normal irradiance and temperature. The output voltage of the inverter is controlled during starting to prevent an excessive current from PV and to develop a torque larger than pump torque. An artificial neural network (ANN) is developed to give an optimal inverter voltage and frequency to extract maximum power from the PV array. The complete model is simulated using MATLAB/Simulink. The simulated results emphasize the significance of the proposed method to attain the maximum power from PV with minimum motor losses. [ABSTRACT FROM AUTHOR]

The performance of a neural network model depends on the data used to develop the model and nature of data preprocessing used in its training algorithm. This paper proposes a hybrid clustering algorithm to preprocess the data that is used for neural network training as well as data division technique to predict the power output of a soiled solar panel. Deterministic characteristics of soil on the panel namely particle size composition, X-ray diffraction (XRD) analysis and Fourier transform infrared spectroscopy (FTIR) analysis are used to model the soil. Artificial soiling experiment is conducted with each soil sample on the panel at irradiance level in the range of 200–1200 W/m 2 for a set of 18 tilt angles to collect data of short circuit current ( I sc ) and open circuit voltage ( V oc ) leading to power output. NNR (Neural Network random) with random data division, NNF (Neural network fuzzy) with Fuzzy C Means clustering before training, NNK (Neural network k -means) with K -means data clustering and NNH (Neural network hybrid) with hybrid data clustering and division techniques are developed using these data. The performance of these networks for a known and unknown soil samples is compared. For a known soil sample, NNH performed better (Maximum percentage error of 2%) as compared to NNR (−6.3%), NNF (−6.7%) and NNK (13%). In case of unknown soil sample NNH outperforms other models with a maximum error margin as low as (−10%) as compared to NNR (50%), NNF (−81%) and NNK (124%). [ABSTRACT FROM AUTHOR]

Nowadays, the estimation of the energy yield of a stand-alone or grid-connected photovoltaic (PV) systems is crucial for ensuring their economic feasibility and the proper sizing of system components. In fact, the energy yield estimation allows to avoid outages and it ensures quality and continuity of supply. In this context, this paper analyzes and compares two different approaches to estimate energy yield of a 1.05 kW p experimental PV plant located at ENEA Portici Research Centre: the first one is based on the physical modelization of the plant; the other one is related to various topologies of Artificial Neural Networks (ANN). In particular, in the second case, a new hybrid method, called Hybrid Physical Artificial Neural Network (HPANN), based on an ANN and clear sky solar radiation curves is proposed and compared with a Multi-Layer Perceptron (MLP) ANN method widely used in the scientific literature. Moreover, using the same structure of the HPANN, a nonlinear AutoRegressive eXogenous (ARX) model, which uses a wavelet network as its nonlinearity estimator, and an approach founded on Adaptive Network based Fuzzy Inference System (FIS) have also been developed. In order to verify the effectiveness of the implemented approaches, measured and estimated data have been compared and errors have been calculated by means of different statistical coefficients. Results demonstrate that the HPANN approach allows a more precise estimation of the ac energy yield, obtaining, in the worst case, values of Relative Root Mean Square Error less than 10%. [ABSTRACT FROM AUTHOR]

Accurate prediction of the power generated from solar energy is required for the successful integration of solar energy into the power grid. In this paper we consider forecasting the electricity power produced by a photovoltaic solar system. While previous research has been concerned with point forecasts, we focus on interval forecasts which are more suitable for the highly variable nature of the solar data. We consider a special type of interval forecasts, called 2D-interval forecasts, where the goal is to predict a range of expected values for the solar power output, for a future time interval. We present a new approach called SVR2D which directly computes the 2D-interval forecasts from previous historical solar power and meteorological data, using support vector regression as a prediction algorithm. We evaluate its performance using Australian photovoltaic data for two years sampled every 1, 5 and 30 min, for various interval lengths. The results show that SVR2D provides accurate predictions, outperforming a number of baselines and other methods used for comparison. [ABSTRACT FROM AUTHOR]

Solar forecasting has become an important issue for power systems planning and operating, especially in islands grids . Power generation and grid utilities need day ahead, intra-day and intra-hour Global Horizontal solar Irradiance (GHI) forecasts for operations. In this paper, we focus on intra-day solar forecasting with forecast horizons ranging from 1 h to 6 h ahead. An Artificial Neural Networks (ANN) model is proposed to forecast GHI using ground measurement data and satellite data (from Helioclim-3) as inputs. In order to compare the forecasting results obtained by the proposed ANN model, we also include in this work a simple naïve model, based on the persistence of the clear sky index (smart persistence model), as well as another reference model, the climatological mean model. The models were trained and tested for two ground measurements stations in Gran Canaria Island, Pozo (south) and Las Palmas (in the north). Firstly, ANN was trained and tested only with past ground measurement irradiance and compared by means of relative metrics with naïve models. While this first step led to better performances, forecasting skills were improved by including exogenous inputs to the model by using GHI satellite data from surrounding area. [ABSTRACT FROM AUTHOR]

This paper proposes a hybrid solar radiation forecasting method based on a novel game theoretic self-organizing map (GTSOM). New strategies are proposed to resolve the limitations of the original SOM for non-winning neurons and increase their competition with winning neurons to obtain more input patterns. Neural gas (NG) and competitive Hebbian Learning (CHL) are used to enhance the learning and quality of the map. Solar radiation data are decomposed by the discrete wavelet transform (DWT). A time series analysis is then used to develop the structure of the training and testing for Bayesian neural networks (BNNs). The proposed GTSOM groups the time-series analyzed datasets into clusters with similar data. The elbow method is used to determine the number of clusters. A cluster selection method is developed to determine the appropriate cluster whose solar radiation data provide the input to the NN. Temperature, wind speed and wind direction data are also included in the inputs to the BNN whose outputs provide the solar radiation forecasts. The historical solar radiation data are used to evaluate the accuracy of the hybrid forecasting with the proposed clustering and its comparison with that of the K-means, the original SOM and NG clustering algorithms. The comparison demonstrates the superior performance of the proposed clustering method. [ABSTRACT FROM AUTHOR]

In this paper, we propose a benchmarking of supervised machine learning techniques (neural networks, Gaussian processes and support vector machines) in order to forecast the Global Horizontal solar Irradiance (GHI). We also include in this benchmark a simple linear autoregressive (AR) model as well as two naive models based on persistence of the GHI and persistence of the clear sky index (denoted herein scaled persistence model). The models are calibrated and validated with data from three French islands: Corsica (41.91°N; 8.73°E), Guadeloupe (16.26°N; 61.51°W) and Reunion (21.34°S; 55.49°E). The main findings of this work are, that for hour ahead solar forecasting, the machine learning techniques slightly improve the performances exhibited by the linear AR and the scaled persistence model. However, the improvement appears to be more pronounced in case of unstable sky conditions. These nonlinear techniques start to outperform their simple counterparts for forecasting horizons greater than 1 h. [ABSTRACT FROM AUTHOR]

In the paper two models implemented to forecast the hourly solar irradiance with a day in advance are described. The models, based on Artificial Neural Networks (ANN), are generated by a master optimization process that defines the best number of neurons and selects a suitable ensemble of ANN. The two models consist of a Statistical (ST) model that uses only local measured data and a Model Output Statistics (MOS) that corrects Numerical Weather Prediction (NWP) data. ST and MOS are tested for the University of Rome “Tor Vergata” site. The models are trained and validated using one year data. Through a cross training procedure, the dependence of the models on the training year is also analyzed. The performance of ST, NWP and MOS models, together with the benchmark Persistence Model (PM), are compared. The ST model and the NWP model exhibit similar results. Nevertheless different sources of forecast errors between ST and NWP models are identified. The MOS model gives the best performance, improving the forecast of approximately 29% with respect to the PM. [ABSTRACT FROM AUTHOR]

Solar cooling is still a young and small but growing market with a large potential. An increasing market development of solar cooling and so-called SolarCombiPlus systems (solar thermal systems providing domestic hot water, space heating and space cooling) can help to reduce primary energy demand and hence emissions of greenhouse gases. To support the market entry and to enhance the market penetration it is important to strengthen consumers’ confidence in these systems. An important aspect for achieving this goal is a standardised method to predict the performance of the complete solar cooling system under real operating conditions. Nonetheless, objective performance test methods are not yet common standard. In this context a performance test method for solar cooling and SolarCombiPlus systems based on the CTSS method (Component Testing – System Simulation) has been developed by the Research and Testing Centre for Thermal Solar Systems (TZS) of the Institute for Thermodynamics and Thermal Engineering (ITW) at the University of Stuttgart within the project “SolTrans”. For the proposed extended CTSS method numerical models are required in order to describe the thermal behaviour of sorption chillers. The main target of the work presented in this paper is dedicated to the development of appropriate models for sorption chillers which can be used for the extended CTSS method. The approach used is the experimental system identification 1 In the field of the experimental system identification a mathematical model of a dynamical system (e.g. sorption chiller) is derived from measurements. 1 based on Artificial Neural Networks (ANN). In this approach experimentally measured data are used to derive an ANN model which is able to predict the outlet temperatures of a sorption chiller. In the work presented, measured data of an adsorption chiller were used to develop such an ANN model which is suitable to predict the outlet temperatures of the three hydraulic loops of adsorption chillers. The model was validated with measured data obtained under real operating conditions. The simulated output temperatures show a very good agreement with the measured temperatures. [ABSTRACT FROM AUTHOR]

Highlights: [•] This paper presents a hybrid CRO–ELM algorithm for solar radiation prediction. [•] Novel predictive meteorological variables are considered in this problem. [•] The CRO is a novel meta-heuristic search algorithm based on simulation of Coral Reefs. [•] The Extreme Learning Machine (ELM) is a state of the art method for training neural networks. [•] The proposed hybrid approach has shown excellent results in real data in Spain. [Copyright &y& Elsevier]

Abstract: The trends of solar radiation are not easy to capture and become especially hard to predict when weather conditions change dramatically, such as with clouds blocking the sun. At present, the better performing methods to forecast solar radiation are time series methods, artificial neural networks and stochastic models. This paper will describe a new and efficient method capable of forecasting 1-h ahead solar radiation during cloudy days. The method combines an autoregressive (AR) model with a dynamical system model. In addition, the difference of solar radiation values at present and lag one time step is used as a correction to a predicted value, improving the forecasting accuracy by 30% compared to models without this correction. [Copyright &y& Elsevier]

Abstract: Earlier studies have shown that the thermal efficiency of a solar thermal collector is influenced by the quality of coolant flow distribution within the array. The quality of flow distribution within a header/riser type fluid network is influenced by a number of parameters including the coolant mass flow rate, the direction of flow in the manifolds, and most importantly the geometry of the array. As a building integrated photovoltaic thermal (BIPVT) collector will be a made to measure product, meaning its dimensions will be dictated by the specific roofing and energy requirements of the customer, the issue of flow distribution and its effect on both thermal and photovoltaic performance raises some concern. In order to quantify the detrimental effect flow distribution may have on the photovoltaic output of a BIPVT array, a numerical approach was presented by authors in earlier work where factors known to influence flow distribution were varied. The authors demonstrated that photovoltaic output may be severely hindered if the issue of flow distribution is not adequately addressed. In this new study we use this numerical method to additionally show that photovoltaic output is not always improved by reducing the fin width W (by increasing the number of risers for each module) despite the theoretical rise in fin efficiency F as a consequence of flow distribution. Under certain scenarios it was shown that a single riser will in fact improve PV output and also reduce system installation cost and complexity. The combination of these new finding and results previously obtained by the authors highlights the need in selecting the optimal design parameters to meet the specific requirements of each installation. The numerical approach carried out to calculate PV output however was time consuming and computationally intensive and therefore not feasible to perform for each and every customer. To address this issue, the authors have proposed in this paper to train an artificial neural network which can be used to approximate the photovoltaic yield of an array of specified shape operating under parallelreverse flow in the manifolds and also with one or two fluid channels cooling each string of cells. By approximating the yield for each scenario, the optimal configuration can then be chosen. It was found that the neural network can be successfully trained for this specific application offering a fast alternative to the original numerical approach. [Copyright &y& Elsevier]

Abstract: The quantity of solar radiation received by the earth’s surface is very important to numerous renewable energy applications. However, direct measurement of solar data is not widely available, especially in developing countries. This paper uses Particle Swarm Optimization (PSO) to train an artificial neural network (PSO–ANN) using data from available measurement stations to estimate monthly mean daily Global Solar Radiation (GSR) at locations where no measurement stations are available. The inputs to the networks are: month of the year, latitude, longitude, altitude, and sunshine duration, and the output is the monthly mean daily GSR at the specified location. Using training data from 31 stations and testing data from 10 locations, the PSO–ANN outperforms a neural network trained using the standard backpropagation (BP) algorithm (BP–ANN) with an average Mean Absolute Percentage Error (MAPE) of 8.85% for the PSO–ANN and 12.61% for the BP–ANN. The performance is improved significantly, when we use the leave-one-out method, where data from 40 locations is used for training and data from the 41st station is used for assessing the performance. In this case the average of MAPE on data from the 10 testing stations is about 7%. We used the same method to assess the performance of the PSO–ANN on testing data from each of the 41 stations with an overall average MAPE of about 10.3%. Comparison with BP–ANN and an empirical model showed the superiority of the PSO–ANN. [Copyright &y& Elsevier]

Abstract: Artificial neural network (ANN) based maximum power point tracking (MPPT) algorithm makes use of the advantages of ANNs such as noise rejection capability and not requiring any prior knowledge of the physical parameters relating to PV system. This paper proposes a genetic algorithm (GA) optimized ANN-based MPPT algorithm implemented in a stand-alone PV system with direct-coupled induction motor drive. The major objective of this design is to eliminate dc–dc converter and its accompanying losses. Implementing off-line ANN in DSP needs optimization of ANN structure to obtain an ideal size. GA optimization was used in this study to determine neuron numbers in multi-layer perceptron neural network. Another objective of this work is to prevent the necessity of the trade-off between the tracking speed and the oscillations around the maximum power point. Hence, varying step size is used in MPPT algorithm and PI-controller is adopted for simple implementation. Simulation and experimental results have been used to demonstrate effectiveness of the proposed method. [Copyright &y& Elsevier]

Abstract: Solar CombiSystems (SCSs) are very efficient systems for reducing conventional energy consumption of building but their thermal performances are strongly dependent on the environment where they are installed (type of climate and thermal quality of the building). Currently it is impossible to predict the energy savings generated by a SCS as there is no standard test to characterise SCS performances. Currently, the Short Cycle System Performance Test (SCSPT), based on a 12days test of the complete SCS on a semi-virtual test bench, is able to predict annual energy savings with a good accuracy, but the performance prediction is limited to only one environment (the building and the climate corresponding with the test). Based on the SCSPT procedure, this paper proposes an improvement of the method by identifying a global SCS model from the test data. Then, the identified model would be able to simulate the tested SCS in any environment and thus to characterise its performances. The proposed model to identify is a “grey box” model, mixing a “White Box” model composed of known physical equations and a “Black Box” model, which is an Artificial Neural Network (ANN). A complete process is developed to train and select a relevant global SCS model from such a test. This approach has been validated through numerical simulations of three detailed SCS models. Compared to those annual results, “Grey Box” SCS models trained from a twelve days sequence are able to predict energy consumption with a good accuracy for 27 different environments. An experimental application of this procedure has been used to characterise a real system. [Copyright &y& Elsevier]

Abstract: Power forecasting is an important factor for planning the operations of photovoltaic (PV) system. This paper presents an advanced statistical method for solar power forecasting based on artificial intelligence techniques. The method requires as input past power measurements and meteorological forecasts of solar irradiance, relative humidity and temperature at the site of the photovoltaic power system. A self-organized map (SOM) is trained to classify the local weather type of 24h ahead provided by the online meteorological services. A unique feature of the method is that following a preliminary weather type classification, the neural networks can be well trained to improve the forecast accuracy. The proposed method is suitable for operational planning of transmission system operator, i.e. forecasting horizon of 24h ahead and for PV power system operators trading in electricity markets. Application of the forecasting method on the power production of an actual PV power system shows the validity of the method. [Copyright &y& Elsevier]

Abstract: This paper presents a novel methodology for Maximum Power Point Tracking (MPPT) of a grid-connected 20kW photovoltaic (PV) system using neuro-fuzzy network. The proposed method predicts the reference PV voltage guarantying optimal power transfer between the PV generator and the main utility grid. The neuro-fuzzy network is composed of a fuzzy rule-based classifier and three multi-layered feed forwarded Artificial Neural Networks (ANN). Inputs of the network (irradiance and temperature) are classified before they are fed into the appropriated ANN for either training or estimation process while the output is the reference voltage. The main advantage of the proposed methodology, comparing to a conventional single neural network-based approach, is the distinct generalization ability regarding to the nonlinear and dynamic behavior of a PV generator. In fact, the neuro-fuzzy network is a neural network based multi-model machine learning that defines a set of local models emulating the complex and nonlinear behavior of a PV generator under a wide range of operating conditions. Simulation results under several rapid irradiance variations proved that the proposed MPPT method fulfilled the highest efficiency comparing to a conventional single neural network and the Perturb and Observe (P&O) algorithm dispositive. [ABSTRACT FROM AUTHOR]

Abstract: In this paper, we present an application of Artificial Neural Networks (ANNs) in the renewable energy domain. We particularly look at the Multi-Layer Perceptron (MLP) network which has been the most used of ANNs architectures both in the renewable energy domain and in the time series forecasting. We have used a MLP and an ad hoc time series pre-processing to develop a methodology for the daily prediction of global solar radiation on a horizontal surface. First results are promising with nRMSE∼21% and RMSE∼3.59MJ/m2. The optimized MLP presents predictions similar to or even better than conventional and reference methods such as ARIMA techniques, Bayesian inference, Markov chains and k-Nearest-Neighbors. Moreover we found that the data pre-processing approach proposed can reduce significantly forecasting errors of about 6% compared to conventional prediction methods such as Markov chains or Bayesian inference. The simulator proposed has been obtained using 19years of available data from the meteorological station of Ajaccio (Corsica Island, France, 41°55′N, 8°44′E, 4m above mean sea level). The predicted whole methodology has been validated on a 1.175kWc mono-Si PV power grid. Six prediction methods (ANN, clear sky model, combination…) allow to predict the best daily DC PV power production at horizon d +1. The cumulated DC PV energy on a 6-months period shows a great agreement between simulated and measured data (R 2 >0.99 and nRMSE<2%). [ABSTRACT FROM AUTHOR]

Abstract: This paper proposes a hybrid approach based on soft computing techniques in order to estimate monthly and daily ambient temperature. Indeed, we combine the back-propagation (BP) algorithm and the simple Genetic Algorithm (GA) in order to effectively train artificial neural networks (ANN) in such a way that the BP algorithm initialises a few individuals of the GA’s population. Experiments concerned monthly temperature estimation of unknown places and daily temperature estimation for thermal load computation. Results have shown remarkable improvements in accuracy compared to traditional methods. [Copyright &y& Elsevier]

Abstract: This paper deals with the development of methods for non steady state test of solar thermal collectors. Our goal is to infer performances in steady-state conditions in terms of the efficiency curve when measures in transient conditions are the only ones available. We take into consideration the method of identification of a system in dynamic conditions by applying a Graybox Identification Model and a Dynamic Adaptative Linear Neural Network (ALNN) model. The study targets the solar collector with evacuated pipes, such as Dewar pipes. The mathematical description that supervises the functioning of the solar collector in transient conditions is developed using the equation of the energy balance, with the aim of determining the order and architecture of the two models. The input and output vectors of the two models are constructed, considering the measures of 4 days of solar radiation, flow mass, environment and heat-transfer fluid temperature in the inlet and outlet from the thermal solar collector. The efficiency curves derived from the two models are detected in correspondence to the test and validation points. The two synthetic simulated efficiency curves are compared with the actual efficiency curve certified by the Swiss Institute Solartechnik Pufung Forschung which tested the solar collector performance in steady-state conditions according to the UNI-EN 12975 standard. An acquisition set of measurements of only 4 days in the transient condition was enough to trace through a Graybox State Space Model the efficiency curve of the tested solar thermal collector, with a relative error of synthetic values with respect to efficiency certified by SPF, lower than 0.5%, while with the ALNN model the error is lower than 2.2% with respect to certified one. [Copyright &y& Elsevier]

Abstract: Forecasting of solar irradiance is in general significant for planning the operations of power plants which convert renewable energies into electricity. In particular, the possibility to predict the solar irradiance (up to 24h or even more) can became – with reference to the Grid Connected Photovoltaic Plants (GCPV) – fundamental in making power dispatching plans and – with reference to stand alone and hybrid systems – also a useful reference for improving the control algorithms of charge controllers. In this paper, a practical method for solar irradiance forecast using artificial neural network (ANN) is presented. The proposed Multilayer Perceptron MLP-model makes it possible to forecast the solar irradiance on a base of 24h using the present values of the mean daily solar irradiance and air temperature. An experimental database of solar irradiance and air temperature data (from July 1st 2008 to May 23rd 2009 and from November 23rd 2009 to January 24th 2010) has been used. The database has been collected in Trieste (latitude 45°40′N, longitude 13°46′E), Italy. In order to check the generalization capability of the MLP-forecaster, a K-fold cross-validation was carried out. The results indicate that the proposed model performs well, while the correlation coefficient is in the range 98–99% for sunny days and 94–96% for cloudy days. As an application, the comparison between the forecasted one and the energy produced by the GCPV plant installed on the rooftop of the municipality of Trieste shows the goodness of the proposed model. [Copyright &y& Elsevier]

Abstract: In this work an application of a methodology to obtain solar radiation maps is presented. This methodology is based on a neural network system [Lippmann, R.P., 1987. An introduction to computing with neural nets. IEEE ASSP Magazine, 4–22] called Multi-Layer Perceptron (MLP) [Haykin, S., 1994. Neural Networks. A Comprehensive Foundation. Macmillan Publishing Company; Hornik, K., Stinchcombe, M., White, H., 1989. Multilayer feedforward networks are universal approximators. Neural Networks, 2(5), 359–366]. To obtain a solar radiation map it is necessary to know the solar radiation of many points spread wide across the zone of the map where it is going to be drawn. For most of the locations all over the world the records of these data (solar radiation in whatever scale, daily or hourly values) are non-existent. Only very few locations have the privilege of having good meteorological stations where records of solar radiation have being registered. But even in those locations with historical records of solar data, the quality of these solar series is not as good as it should be for most purposes. In addition, to draw solar radiation maps the number of points on the maps (real sites) that it is necessary to work with makes this problem difficult to solve. Nevertheless, with the application of the methodology proposed in this paper, this problem has been solved and solar radiation maps have been obtained for a small region of Spain: Jaén province, a southern province of Spain between parallels 38°25′ N and 37°25′ N, and meridians 4°10′ W and 2°10′ W, and for a larger region: Andalucía, the most southern region of Spain situated between parallels 38°40′ N and 36°00′ N, and meridians 7°30′ W and 1°40′ W. [Copyright &y& Elsevier]

Abstract: Field testing carried out for solar energy applications is costly, time consuming and depends heavily on prevailing weather conditions. Adequate security and weather protection must be provided at the test site. Measurements may also suffer from delays that can be caused by system failures and bad weather. To overcome these problems the need for accurate model becomes evermore important. To achieve such prediction task, an artificial neural network, ANN, model is regarded as a cost-effective technique superior to traditional statistical methods. In this paper, Levenberg optimization function is adopted to predict insolation data in different spectral bands for Helwan (Egypt) monitoring station. The predicted values were then compared with the actual values and presented in terms of usual statistics. The results hint that, the ANN model predicted infrared, ultraviolet, and global insolation with a good accuracy of approximately 95%, 93% and 96%, respectively. In addition, ANN model was tested to predict the same components for Aswan over an 11 month period. The predicted values of the ANN model compared to the actual values for Aswan produced an accuracy of 95%, 91% and 92%, respectively. Data for Aswan were not included as a part of ANN training set. Hence, these results demonstrate the generalization capability of this approach over unseen data and its ability to produce accurate estimates. [Copyright &y& Elsevier]

Abstract: Several methods for sizing stand alone photovoltaic (pv) systems has been developed. The more simplistic are called intuitive methods. They are a useful tool for a first approach in sizing stand alone photovoltaic systems. Nevertheless they are very inaccurate. Analytical methods use equations to describe the pv system size as a function of reliability. These ones are more accurate than the previous ones but they are also not accurate enough for sizing of high reliability. In a third group there are methods which use system simulations. These ones are called numerical methods. Many of the analytical methods employ the concept of reliability of the system or the complementary term: loss of load probability (LOLP). In this paper an improvement for obtaining LOLP curves based on the neural network called Multilayer Perceptron (MLP) is presented. A unique MLP for many locations of Spain has been trained and after the training, the MLP is able to generate LOLP curves for any value and location. [Copyright &y& Elsevier]