Evaluation of Temperature-Based Empirical Models and Machine Learning Techniques to Estimate Daily Global Solar Radiation at Biratnagar Airport, Nepal

Abstract

Global solar radiation (GSR) is a critical variable for designing photovoltaic cells, solar furnaces, solar collectors, and other passive solar applications. In Nepal, the high initial cost and subsequent maintenance cost required for the instrument to measure GSR have restricted its applicability all over the country. The current study compares six different temperature-based empirical models, artificial neural network (ANN), and other five different machine learning (ML) models for estimating daily GSR utilizing readily available meteorological data at Biratnagar Airport. Amongst the temperature-based models, the model developed by Fan et al. performs better than the rest with an R2 of 0.7498 and RMSE of 2.0162 MJm−2d−1. Feed-forward multilayer perceptron (MLP) is utilized to model daily GSR utilizing extraterrestrial solar radiation, sunshine duration, maximum and minimum ambient temperature, precipitation, and relative humidity as inputs. ANN3 performs better than other ANN models with an R2 of 0.8446 and RMSE of 1.4595 MJm−2d−1. Likewise, stepwise linear regression performs better than other ML models with an R2 of 0.8870 and RMSE of 1.5143 MJm−2d−1. Thus, the model developed by Fan et al. is recommended to estimate daily GSR in the region where only ambient temperature data are available. Similarly, a more robust ANN3 and stepwise linear regression models are recommended to estimate daily GSR in the region where data about sunshine duration, maximum and minimum ambient temperature, precipitation, and relative humidity are available.

Improving Biomass and Grain Yield Prediction of Wheat Genotypes on Sodic Soil Using Integrated High-Resolution Multispectral, Hyperspectral, 3D Point Cloud, and Machine Learning Techniques

“…The SVM is considered a reliable supervised learning technique that is used for classification, pattern recognition, regression, and prediction [74]. The linear kernel is used for a linearly separable dataset [75], whereas the Gaussian SVM is used for complex relationships [76]. Quadratic and cubic functions can provide flexibility in fitting [72].…”

“…Mokarram and Bijanzadeh [96] also reported a better prediction accuracy using the MLP (ANN) network model in comparison to MLR for barley biomass yield (R 2 = 0.89) and grain yield (R 2 = 0.92) on non-constrained soils. Recently, Dhakal, Gautam, and Bhattarai [76] reported that an ANN model with 10 neurons in the hidden layer performed comparatively better in training with R 2 = 0.84 and RMSE = 1.5 MJ•m −2 •day −1 when estimating global solar radiation. Furthermore, their study also found that the stepwise kernel has good potential for estimation with less error (R 2 = 0.88 and RMSE = 1.5 MJ•m −2 •day −1 ).…”

Dew Point Time Series Forecasting at the North Dakota

“…There is still several drawbacks have been approved with the development of classical version of AI models such as ANN, SVR, ANFIS models such as the overfitting, learning process limitations, hyperparameters tuning [31]- [34]. Hence, the exploration of new version of robust soft computing technologies to overcome the mentioned limitations is always the motive for engineers and decision makers [35]- [37]. In addition, based on the researches discussed above, the main aim generally has been paid to forecasting one-step ahead prediction of Tdew.…”

Estimation of Solar Energy Using Different Empirical Models at Mid Hill, Nepal

“…Models can be used to estimate and predict GSR where measurements are not available. There are several models that has been used to calculate solar radiation using easily available meteorological parameters and climatological parameters such as sunshine hour [8,15,16,17], temperature [18,19,20,21], latitude [22,23], relative humidity [24,25], rainfall [26] and cloudiness [27]. In Nepal, the estimation of GSR, using empirical models and RadEst v3.0 software were observed [28].…”