Elsevier

Measurement

Volume 76, December 2015, Pages 148-155
Measurement

Using measured daily meteorological parameters to predict daily solar radiation

https://doi.org/10.1016/j.measurement.2015.08.004Get rights and content

Abstract

A major factor for an efficient design of solar energy systems is to provide accurate estimations of the solar radiation. Many of the existing studies are focused on the analysis of monthly or annual solar radiation. This is while less attention has been paid to the determination of daily solar radiation. Accordingly, the main goal of this paper is to develop a robust machine learning approach, based on genetic programming (GP), for the estimation of the daily solar radiation. The solar radiation is formulated in terms of daily air temperature, relative humidity, atmospheric pressure, wind speed, and earth temperature. A comprehensive database containing about 7000 records collected for about 20 years (1995–2014) in a nominal city in Iran is used to develop the GP model. The performance of the derived model is verified using different criteria. A multiple linear regression analysis is performed to benchmark the GP model with a classical technique. The influences of the input variables on the solar energy are evaluated through a sensitivity analysis. The proposed model has a very good prediction performance and significantly outperforms the traditional regression model.

Introduction

Renewable energy sources such as solar have emerged as effective alternatives to fossil fuels. Optimal design of solar systems requires exact prediction of the solar energy [1], [2], [3], [4]. To this aim, several empirical methods have been developed to avoid performing costly in-situ solar radiation measurements [5], [6]. Some of the well-known methods in this area are auto-regression, Markov chain, or robust optimization techniques [7], [8], [9]. Among the empirical methods, machine learning has been widely used to solve real world problems [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28]. Artificial neural networks (ANNs) are well-known machine learning systems that have been utilized to predict the solar radiation [2], [3], [4], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40]. As typical examples in this context, Rahimikhoob [41] successfully applied ANN for predicting the global solar radiation. Alam et al. [42] employed ANN to predict beam solar radiation. Mellit et al. [43] conducted a comprehensive study on application multi-layer perceptron of ANN for estimating total solar radiation data. However, a limitation of ANNs is that they cannot always be converted into explicit forms to provide the details of the prediction process [44].

Genetic programming (GP) [45] is a new approach with notable simulation capabilities. It is a division of genetic algorithms (GA) that generates computer programs rather [44]. For the last decade, GP has been used to formulate complicated engineering problems [44], [46], [47], [48], [49], [50], [51], [52]. Gene expression programming (GEP) [53], [54] is a recent branch of GP evolving programs of various sizes and shapes. Compared to other soft computing methods, it has been barely applied to energy related problems [2], [3], [55], [56], [57]. Although ANNs are used to predict the “daily” solar radiation [41], [58], [59], [60], [61], [62], none of the existing studies have focused on GP-based analysis of daily solar radiation. Therefore, this study presents the GEP technique for predicting the daily solar radiation in Iran. A regression analysis was later preformed to benchmark the proposed model.

Section snippets

Methodology

GP is an optimization technique to create computer programs. The process is inspired by the biological evolution of living organisms [45]. Generally, in GP, the main goal is to find a program that connects inputs to outputs (see Fig. 1) [55]. The solutions derived by conventional GP are shown as tree structures [63]. However, there are other branches of GP that produce models in different shapes (see Fig. 2) [44], [55]. Among different variants, linear variants have a higher speed permitting

The methodology for predicting daily solar radiation

The soft computing tools commonly follow similar steps to develop a prediction model for the daily solar radiation [3], [55], [67]. A similar approach was also considered herein.

Performance verification

It is known that |R| > 0.8 is a good indicator of accuracy of a regression model [69]. Moreover, the values of RMSE and MAE should be minimum. It can be observed from Fig. 5 that the GEP model has acceptable performance both for the training (RTraining = 0.866, RMSETraining = 1.032 kW h/m2/d, MAETraining = 0.817 kW h/m2/d) and testing (Rtesting = 0.881, RMSEtesting = 1.202 kW h/m2/d, MAEtesting = 0.940 kW h/m2/d) data. Besides, new measures were checked for more verification of the models on the testing [70], [71].

Sensitivity analysis

As discussed before, the effect of all of the considered parameters (i.e., Tave, Tmin, Tmax, H, P, W, and E) on DS is well understood. Ignoring any of these three parameters for the model development resulted in a model with poor performance. Herein, a sensitivity analysis was conducted to provide a more in depth understanding of the contribution of these important parameters to the prediction of DS. A common approach for the sensitivity analysis in the GP-based modeling is to obtain the input

Conclusion

In this study, a new GEP approach is presented for the prediction of the daily solar radiation. A comprehensive database containing data collected for about 20 years was used to develop the GEP model. The model predicts the daily solar radiation with an acceptable accuracy and outperforms the developed regression-based model. GEP does not need a predefined function for the modeling of the solar radiation. As expected, the solar radiation is more affected by average of air temperature, wind speed

References (71)

  • N. Lertpalangsunti et al.

    An architectural framework for the construction of hybrid intelligent forecasting systems: application for electricity demand prediction

    Eng. Appl. Artif. Intell.

    (1998)
  • V. Miranda et al.

    Evolutionary computation in power systems

    Int. J. Power Energy Syst.

    (1998)
  • J.M. Molina et al.

    Hydroelectric power plant management relying on neural networks and expert system integration

    Eng. Appl. Artif. Intel.

    (2000)
  • G.N. Nguene et al.

    A fuzzy-based approach for strategic choices in electric energy supply The case of a Swiss power provider on the eve of electricity market opening

    Eng. Appl. Artif. Intel.

    (2007)
  • R. Pino et al.

    Forecasting next-day price of electricity in the Spanish energy market using artificial neural networks

    Eng. Appl. Artif. Intel.

    (2008)
  • L.M. Romeo et al.

    Hybrid System for fouling control in biomass boilers

    Eng. Appl. Artif. Intel.

    (2006)
  • D. Srinivasan

    Energy demand prediction using GMDH networks

    Neurocomputing

    (2008)
  • A. Azadeh et al.

    Annual electricity consumption forecasting by neural network in high energy consuming industrial sectors

    Energy Convers. Manage.

    (2008)
  • A.H. Alavi et al.

    Prediction of principal ground-motion parameters using a hybrid method coupling artificial neural networks and simulated annealing

    Comput. Struct.

    (2011)
  • A. Mellit et al.

    A simplified model for generating sequences of global solar radiation data for isolated sites: using artificial neural network and a library of Markov transition matrices approach

    Sol. Energy

    (2005)
  • A. Sfetsos et al.

    Univariate and multivariate forecasting of hourly solar radiation with artificial intelligence techniques

    Sol. Energy

    (2000)
  • A. Sözen et al.

    Forecasting based on neural network approach of solar potential in Turkey

    Renewable Energy

    (2005)
  • A. Sözen et al.

    Use of artificial neural networks for mapping of solar potential in Turkey

    Appl. Energy

    (2004)
  • C. Voyant et al.

    Time series modeling and large scale global solar radiation forecasting from geostationary satellites data

    Sol. Energy

    (2014)
  • O. Şenkal et al.

    Estimation of solar radiation over Turkey using artificial neural network and satellite data

    Appl. Energy

    (2009)
  • C. Paoli et al.

    Forecasting of preprocessed daily solar radiation time series using neural networks

    Sol. Energy

    (2010)
  • M.A. Behrang et al.

    The potential of different artificial neural network (ANN) techniques in daily global solar radiation modeling based on meteorological data

    Sol. Energy

    (2010)
  • A. Rahimikhoob

    Estimating global solar radiation using artificial neural network and air temperature data in a semi-arid environment

    Renewable Energy

    (2010)
  • S. Alam et al.

    Computation of beam solar radiation at normal incidence using artificial neural network

    Renewable Energy

    (2006)
  • A. Mellit et al.

    Methodology for predicting sequences of mean monthly clearness index and daily solar radiation data in remote areas: application for sizing a stand-alone PV system

    Renewable Energy

    (2008)
  • A. Garg et al.

    Stepwise approach for the evolution of generalized genetic programming model in prediction of surface finish of the turning process

    Adv. Eng. Soft.

    (2014)
  • O. Kisi et al.

    A machine code-based genetic programming for suspended sediment concentration estimation

    Adv. Eng. Soft.

    (2010)
  • D. Howard et al.

    Target detection in SAR imagery by genetic programming

    Adv. Eng. Soft.

    (1999)
  • I.F. Kara

    Prediction of shear strength of FRP-reinforced concrete beams without stirrups based on genetic programming

    Adv. Eng. Soft.

    (2011)
  • M.M. Kose et al.

    Modeling of transfer length of prestressing strands using genetic programming and neuro-fuzzy

    Adv. Eng. Soft.

    (2010)
  • Cited by (33)

    • A comprehensive review of hybrid models for solar radiation forecasting

      2020, Journal of Cleaner Production
      Citation Excerpt :

      The authors also justify the choice of CS algorithm by performing a comparison with other optimization algorithm DE and ACO and showed that CS provides an accurate performance compared to the DE and ACO. Mousavi et al. (2015) proposed a new artificial approach for daily global solar radiation prediction on horizontal surface in Iran. The proposed hybridization method integrates the ANN model with Simulated annulling (SA) to form ANN-SA ensemble prediction model.

    • New daily global solar irradiation estimation model based on automatic selection of input parameters using evolutionary artificial neural networks

      2019, Journal of Cleaner Production
      Citation Excerpt :

      On the other hand, recent studies made an effort to use parameters that are routinely recorded at a large number of meteorological stations such as E. F. Alsina et al. in (Alsina et al., 2016) where the used parameters were: top of atmosphere irradiation, day length, number of rainy days, average rainfall, latitude and altitude. In another important work (Mousavi et al., 2015), the authors proposed a prediction model using daily air temperature, relative humidity, atmospheric pressure, wind speed and earth temperature providing good accuracy in the prediction of solar irradiation. Furthermore, we note that all works used one or two hidden layers, which is sufficient for modeling and estimating solar irradiation (Al-Alawi and Al-Hinai, 1998; Benghanem et al., 2009; Premalatha and Valan Arasu, 2016).

    View all citing articles on Scopus
    View full text