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

Turkey has sufficient solar radiation intensities and radiation durations for solar thermal applications since Turkey lies in a sunny belt, between 36° and 42° N latitudes. The yearly average solar-radiation is 3.6 kWh/m²day, and the total yearly radiation period is ∼2610 h. The main focus of this study is to determine the solar-energy potential in Turkey using artificial neural-networks (ANNs). Scaled conjugate gradient (SCG), Pola-Ribiere conjugate gradient (CGP), and Levenberg-Marquardt (LM) learning algorithms and a logistic sigmoid transfer function were used in the network. In order to train the neural network, meteorological data for the last 3 years (2000–2002) from 17 stations (namely cities) spread over Turkey were used as training (11 stations) and testing (6 stations) data. Meteorological and geographical data (latitude, longitude, altitude, month, mean sunshine duration, and mean temperature) are used as inputs to the network. Solar radiation is in the output layer. The maximum mean absolute percentage error was found to be less than 6.7% and R² values to be about 99.8937% for the testing stations. However, the respective values were found to be 2.41 and 99.99658% for the training stations. The trained and tested ANN models show greater accuracies for evaluating solar resource posibilities in regions where a network of monitoring stations has not been established in Turkey. The predicted solar-potential values from the ANN were given in the form of monthly maps. These maps are of prime importance for different working disciplines, like those of scientists, architects, meteorologists, and solar engineers in Turkey. The predictions from ANN models could enable scientists to locate and design solar-energy systems in Turkey and determine the appropriate solar technology.     

Global solar radiation estimation from measurements of visibility and air temperature extremes

Solar radiation is the main source of energy for the survival of life and its associated activities. It is important to know accurate solar radiation value in areas such as agricultural activities, solar energy systems, heating, and meteorology. In this study, we present a model for the estimation of solar radiation value with other meteorological parameters in cases where solar radiation cannot be measured or not available. This model is based on the relationship between solar radiation and measured air temperature and visibility extremes. As is known, the incident global solar radiation is attenuated by clouds, aerosols, ozone layer, water vapor, etc.. In the model, the attenuation of the solar radiation is expressed by dew point temperature, visibility, and the maximum and minimum air temperatures. Dew-point temperature refers to the effect of water vapor on solar radiation, air temperature extremes are used to signify cloudiness. Visibility also gives the effect on the attenuation of solar radiation by air pollutants and aerosols in the model. The model was applied to the data taken from meteorological stations in Turkey. Error analysis was performed and compared with the models in the literature and satisfactory results were obtained.

Abbreviations H: Daily total global solar radiation, units of MJ ? m^?2 ? day^?1; H₀: Extraterrestrial solar radiation, units of MJ ? m^?2 ? day^?1; H_m: Measured daily total global solar radiation, units of MJ ? m^?2 ? day^?1; H_c: Calculated daily total global solar radiation, units of MJ ? m^?2 ? day^?1; T_min: Daily minimum temperature, units of °C; T_max: Daily maximum temperature, units of °C; RH: T_dew: Relative humidity, units of %rh; Dew-point temperature, units of °C     

Forecasting based on neural network approach of solar potential in Turkey

As Turkey lies near the sunny belt between 36 and 42°N latitudes, most of the locations in Turkey receive abundant solar energy. Average annual temperature is 18–20 °C on the south coast, falls down to 14–16 °C on the west coast, and fluctuates 4–18 °C in the central parts. The yearly average solar radiation is 3.6 kW h/m² day, and the total yearly radiation period is 2610 h. The main focus of this study is put forward to solar energy potential in Turkey using artificial neural networks (ANNs). Scaled conjugate gradient (SCG), Pola-Ribiere conjugate gradient (CGP), and Levenberg–Marquardt (LM) learning algorithms and logistic sigmoid transfer function were used in the network. In order to train the neural network, meteorological data for last 4 years (2000–2003) from 12 cities (Çanakkale, Kars, Hakkari, Sakarya, Erzurum, Zonguldak, Balıkesir, Artvin, Çorum, Konya, Siirt, Tekirdağ) spread over Turkey were used as training (nine stations) and testing (three stations) data. Meteorological and geographical data (latitude, longitude, altitude, month, mean sunshine duration, and mean temperature) is used as input to the network. Solar radiation is the output. The maximum mean absolute percentage error was found to be less than 6.78% and R² values to be about 99.7768% for the testing stations. These values were found to be 5.283 and 99.897% for the training stations. The trained and tested ANN models show greater accuracy for evaluating solar resource posibilities in regions where a network of monitoring stations have not been established in Turkey. The predictions from ANN models could enable scientists to locate and design solar energy systems in Turkey and determine the best solar technology.     

The solar radiation climate of Thailand

Geographical, seasonal, and diurnal variations of global solar radiation in Thailand are surveyed. Seasonal effects are shown by separate studies for eight 1.5 month periods of the year defined by standard solar declination values. Detailed maps are given of the geographical distribution of solar radiation prepared from data on cloudiness at 44 stations, duration of sunshine at 18 stations, and linear regressions relating radiation to sunshine at Chiang Mai and Bangkok. The highest mean values are above 19.5 MJ m⁻² d⁻¹ and are widespread in spring. The lowest values are below 15.0 MJ m⁻² d⁻¹ in restricted localities with heavy rainfall in autumn.Rough estimates of diffuse solar radiation and atmospheric turbidity are made from the radiation-sunshine regression parameters. Diffuse radiation averages 8.4 MJ m⁻² d⁻¹. Turbidity at Chiang Mai is high in spring and low in summer and autumn; at Bangkok it is high throughout the year.The diurnal variation of global solar radiation determined from hourly measurements at Chiang Mai and Bangkok is analysed. The mean midday radiation fluxes range from 0.80 kW m⁻² in spring to 0.60 kW m⁻² in autumn. On the average the radiation received in the afternoon is slightly less than that received in the morning.     

Solar-energy potential in Turkey

In this study, a new formula based on meteorological and geographical data was developed to determine the solar-energy potential in Turkey using artificial neural-networks (ANNs). Scaled conjugate gradient (SCG) and Levenberg–Marquardt (LM) learning algorithms and a logistic sigmoid transfer function were used in the network. Meteorological data for the last four years (2000 → 2003) from 18 cities (Bilecik, Kırşehir, Akhisar, Bingöl, Batman, Bodrum, Uzunköprü, Şile, Bartın, Yalova, Horasan, Polatlı, Malazgirt, Köyceğiz, Manavgat, Dörtyol, Karataş and Birecik) spread over Turkey were used as data in order to train the neural network. Meteorological and geographical data (latitude, longitude, altitude, month, mean sunshine duration, and mean temperature) were used in the input layer of the network. Solar radiation is the output layer. One-month test data for each city was used, and these months data were not used for training. The results show that the maximum mean absolute percentage error (MAPE) was found to be 3.448% and the R² value 0.9987 for Polatlı. The best approach was found for Kırşehir (MAPE=1.2257, R²=0.9998). The MAPE and R² for the testing data were 3.3477 and 0.998534, respectively. The ANN models show greater accuracy for evaluating solar-resource possibilities in regions where a network of monitoring stations has not been established in Turkey. This study confirms the ability of the ANN to predict solar-radiation values precisely.     

Predicting total solar irradiation values using artificial neural networks

This study explores the possibility of developing an artificial neural networks model that could be used to predict monthly average daily total solar irradiation on a horizontal surface for locations in Uganda based on geographical and meteorological data: latitude, longitude, altitude, sunshine duration, relative humidity and maximum temperature. Results have shown good agreement between the predicted and measured values of total solar irradiation. A correlation coefficient of 0.997 was obtained with mean bias error of 0.018 MJ/m² and root mean square error of 0.131 MJ/m². Overall, the artificial neural networks model predicted with an accuracy of 0.1% of the mean absolute percentage error.     

Estimating monthly mean valuesof daily total solar radiation for Australia

Monthly mean values of daily total solar radiation were obtained for the widest possible network acrossAustralia. Bureau of Meteorology sources yielded 11 stations with long term records of both measured daily total solar radiation and sunshine hour values. Monthly modified Angstrom equations were developed from these data and used to estimate radiation values for a further 90 stations in the Bureau of Meteorology network that had sunshine hour data. Measured daily total solar radiation data were obtained from a variety of sources mostly outside the Bureau of Meteorology network for an additional 33 stations. Finally, estimates of solar radiation from detailed cloud cover data were used for a further 12 stations, selected because they filled in significant gaps in coverage. These various sources yielded a total of 146 sets of monthly mean values of daily total solar radiation. For each month optimal surfaces, which were functions of position only, were fitted to this network of values using Laplacian smoothing splines with generalized cross validation. Residuals from the fitted surfaces at the data points were acceptably low. Fitted surfaces which included, in addition to position variables, a cloudiness index based on a transform of mean monthly precipitation further reduced these residuals. The latter fitted surfaces permit estimation of monthly mean values of total daily solar radiation at any point on the continent with a root mean square predictive error of no more than 1.25 MJ m⁻² day⁻¹ (5.2 per cent of the network mean) in summer and 0.74 MJ m⁻² day⁻¹ (5.5 per cent of the network mean) in winter.     

Global,direct and diffuse solar radiation on horizontal and tilted surfaces in Jeddah,Saudi Arabia

The measured data of global and diffuse solar radiation on a horizontal surface, the number of bright sunshine hours, mean daily ambient temperature, maximum and minimum ambient temperatures, relative humidity and amount of cloud cover for Jeddah (lat. 21°42′37′′N, long. 39°11′12′′E), Saudi Arabia, during the period (1996–2007) are analyzed. The monthly averages of daily values for these meteorological variables have been calculated. The data are then divided into two sets. The sub-data set I (1996–2004) are employed to develop empirical correlations between the monthly average of daily global solar radiation fraction (H/H₀) and the various weather parameters. The sub-data set II (2005–2007) are then used to evaluate the derived correlations. Furthermore, the total solar radiation on horizontal surfaces is separated into the beam and diffuses components. Empirical correlations for estimating the diffuse solar radiation incident on horizontal surfaces have been proposed. The total solar radiation incident on a tilted surface facing south H_t with different tilt angles is then calculated using both Liu and Jordan isotropic model and Klucher’s anisotropic model. It is inferred that the isotropic model is able to estimate H_t more accurate than the anisotropic one. At the optimum tilt angle, the maximum value of H_t is obtained as ∼36 (MJ/m² day) during January. Comparisons with 22 years average data of NASA SSE Model showed that the proposed correlations are able to predict the total annual energy on horizontal and tilted surfaces in Jeddah with a reasonable accuracy. It is also found that at Jeddah, the solar energy devices have to be tilted to face south with a tilt angle equals the latitude of the place in order to achieve the best performance all year round.     

Comparison of artificial intelligence and empirical models for estimation of daily diffuse solar radiation in North China Plain

Accurate diffuse solar radiation (H_d) data is highly crucial for the development and utilization of solar energy technologies. However, due to expensive cost and technology requirements, measurements of H_d are not available in many regions of North China Plain (NCP), where the diffuse and direct solar radiation are affected by severe particulate pollution. Thus, development of models for precisely estimating H_d is indeed essential in NCP. On this account, the present studies proposed four artificial intelligence models, including the extreme learning machine (ELM), backpropagation neural networks optimized by genetic algorithm (GANN), random forests (RF), and generalized regression neural networks (GRNN), for estimating daily H_d at two meteorological stations of NCP. Daily global solar radiation and sunshine duration along with the estimated extraterrestrial radiation and maximum possible sunshine duration were selected as model inputs to train the models. Meanwhile, the proposed AI models were compared with the empirical Iqbal model to test their performance using measured H_d data. The results indicated that the ELM, GANN, RF, and GRNN models all performed much better than the empirical Iqbal model for estimating daily H_d. All the models underestimated H_d for both stations, with average relative error ranging from ?5.8% to ?5.4% for AI models and 19.1% for Iqbal model in Beijing, ?5.9% to ?4.3% and ?26.9% in Zhengzhou, respectively. Generally, GANN model had the best accuracy, and ELM ranked next, followed by RF and GRNN models. The ELM model had a slightly poorer performance but the highest computation speed, and both the GANN and ELM models could be highly recommended to estimate daily H_d in NCP of China.     

Solar potential in Turkey

Most of the locations in Turkey receive abundant solar-energy, because Turkey lies in a sunny belt between 36° and 42°N latitudes. Average annual temperature is 18 to 20 °C on the south coast, falls to 14–16 °C on the west coat, and fluctuates between 4 and 18 °C in the central parts. The yearly average solar-radiation is 3.6 kW h/m² day, and the total yearly radiation period is ∼2610 h. In this study, a new formulation based on meteorological and geographical data was developed to determine the solar-energy potential in Turkey using artificial neural-networks (ANNs). Scaled conjugate gradient (SCG), Pola-Ribiere conjugate gradient (CGP), and Levenberg–Marquardt (LM) learning algorithms and logistic sigmoid (logsig) transfer function were used in the networks. Meteorological data for last four years (2000–2003) from 12 cities (Çanakkale, Kars, Hakkari, Sakarya, Erzurum, Zonguldak, Balikesir, Artvin, Çorum, Konya, Siirt, and Tekirdaǧ) spread over Turkey were used in order to train the neural-network. Meteorological and geographical data (latitude, longitude, altitude, month, mean sunshine-duration, and mean temperature) are used in the input layer of the network. Solar-radiation is in the output layer. The maximum mean absolute percentage error was found to be less than 3.832% and R² values to be about 99.9738% for the selected stations. The ANN models show greater accuracy for evaluating solar-resource posibilities in regions where a network of monitoring stations has not been established in Turkey. This study confirms the ability of the ANN to predict solar-radiation values accurately.     

A synthesis method for cobalt doped carbon aerogels with high surface area and their hydrogen storage properties

Carbon aerogels doped with nanoscaled Co particles were prepared by first coating activated carbon aerogels using a wet-thin layer coating process. The resulting metal-doped carbon aerogels had a higher surface area (∼1667 m² g⁻¹) and larger micropore volume (∼0.6 cm³ g⁻¹) than metal-doped carbon aerogels synthesised using other methods suggesting their usefulness in catalytic applications. The hydrogen adsorption behaviour of cobalt doped carbon aerogel was evaluated, displaying a high ∼4.38 wt.% H₂ uptake under 4.6 MPa at −196 °C. The hydrogen uptake capacity with respect to unit surface area was greater than for pure carbon aerogel and resulted in ∼1.3 H₂ (wt. %) per 500 m² g⁻¹. However, the total hydrogen uptake was slightly reduced as compared to pure carbon aerogel due to a small reduction in surface area associated with cobalt doping. The improved adsorption per unit surface area suggests that there is a stronger interaction between the hydrogen molecules and the cobalt doped carbon aerogel than for pure carbon aerogel.     

Synthesis and characterization of anhydrous proton conducting inorganic–organic composite membranes for medium temperature proton exchange membrane fuel cells (PEMFCs)

We report on anhydrous proton conducting inorganic–organic composite membranes synthesized from tetraethoxysilane/poly(-dimethylsiloxane)/trimethylphosphate and 1-ethyl-3 methylimidazolium-bis (trifluoromethanesulfonyl) imide ionic liquid as sol–gel precursors. The Fourier transform infrared spectroscopy, ³¹P, ¹H, and ¹³C Nuclear magnetic resonance, thermo gravimetric and differential thermal analysis measurements confirmed that the prepared hybrid membranes possess good chemical stability and are thermally stable up to 350 °C. Conductivity of all the fabricated hybrid membranes was measured under anhydrous conditions within the temperature range −20–150 °C, and a value of 4.87 × 10⁻³ S/cm at 150 °C was achieved for 40 wt% [EMI][TFSI] ionic liquid doped 72TEOS–18PDMS–10PO (OCH₃)₃ (mol %) hybrid membrane. For 40 wt% ionic liquid doped composite membrane, the measured hydrogen permeability value at 150 °C was 4 × 10⁻¹² mol/cm s Pa.     

Investigation of solar radiation in Botswana and some anomalous phenomena observed

Daily global insolation on a horizontal surface in Botswana is recorded continuously at several synoptic stations and at the University of Botswana's Physics Department. Over a number of years, daily total insolation on a tilted surface (β = −30°) was recorded at the Botswana Technology Centre. Hourly, and instantaneous direct normal, global, diffuse and UV-components are continuously recorded at the University of Botswana. All these measurements are done with standard EPLAB equipment.It is found out that the instantaneous direct normal radiation at Solar noon can be as high as 1150 W·m⁻²; and that at 30 min before sunset it can be above 600 W·m⁻²; and it can also be as high as 100 W·m⁻² at sunset or sunrise moments (i.e. with half of the solar disk under the horizon).Daily direct normal solar radiation can exceed 45 MJ·m⁻². Mean daily global radiation varies from 31 MJ·m⁻² in December to 16 MJ·m⁻² in June. Such big values of daily direct normal and global radiation are explained by low humidity and low turbidity.Cases of an anomalous phenomenon which lead to an abnormally big phase shift when direct normal radiation is increasing greatly after Solar noon are observed, and discussed. It is also found that when humidity is low and visibility is high, hourly I_g values recorded with a pyranometer can be less than I_bn (cosθ_z) + I_d-values. This discrepancy could be quite common for regions where humidity and turbidity are low. The trend in the behaviour of the UV-component during the last five years is also analyzed and discussed. The conclusion is made that the ozone layer over Botswana is continuously being depleted.     

A newly isolated green alga, Tetraspora sp. CU2551, from Thailand with efficient hydrogen production

A novel unicellular hydrogen-producing green alga was isolated from fresh water pond in Pathumthani province, Thailand. Under light microscope, this alga was identified as belonging to the genus Tetraspora. Phylogenetic analysis of 18S rRNA sequence revealed that the green alga, identified as Tetraspora sp. CU2551, is closely related to other unicellular green algal species. Tetraspora sp. CU2551 had the shortest doubling time when grown in Tris-acetate-phosphate (TAP) medium under a light intensity of 48–92 μE/m²/s and a temperature of 36 °C. Hydrogen production increased with increasing pH from 5.75 to 9.30; however, almost no production was observed at a pH of 5.25. Addition of 0.5 mM β-mercaptoethanol to the TAP medium stimulated hydrogen production about two-fold. During the hydrogen production phase, the use of TAP medium lacking both nitrogen and sulfur resulted in about 50% increase in the hydrogen production. This was in contrast to only a small increase in the production when either nitrogen or sulfur was omitted in TAP medium. The stimulation of hydrogen production by 0.5 mM β-mercaptoethanol under nitrogen- and sulfur-deprived conditions occurred only when the cells were grown at a light intensity lower than 5 μE/m²/s with no effects at higher intensities. Maximal calculated hydrogen production, 17.3–61.7 μmol/mg Chl a/h, is a very high production rate compared to other green algae and makes Tetraspora sp. CU2551 an interesting model strain for photobiological hydrogen production.     

Photocatalytic conversion of CO2 into methanol using graphitic carbon nitride under solar,UV laser and broadband radiations

Photocatalytic conversion of carbon dioxide into methanol using highly efficient g‐C₃N₄, in conjunction with three different radiations (solar radiation, broad‐band ultraviolet (UV)–visible lamp, and laser beam) is presented. The optical, structural, and morphological properties of the synthesized g‐C₃N₄ were studied using advanced analytical techniques like Fourier transform infrared spectroscopy, UV–visible spectrometer, X‐ray diffraction, high‐resolution transmission electron microscopy, high‐angle annular dark field, and X‐ray photoelectron spectroscopy. The relative merits of the three sources of radiation in the presence of g‐C₃N₄ were studied in terms of key figures of merit of the photocatalytic process, namely, methanol production yield and quantum yield. As expected, after 40 min of irradiation, 355‐nm laser (40 mJ/pulse, 10 Hz) with g‐C₃N₄ rendered the best methanol production yield (510 μmol g ^{? 1} h ^{? 1}), followed by solar radiation (130 μmol g ^{? 1} h ^{? 1}), and UV broadband lamp. This indicates that the photon flux and the spectral properties of incident light are the key factors for the enhancement of methanol production yield. Although the methanol production yield with 355‐nm laser radiation is quite impressive because of the inherent high photon flux and the monochromatic nature of laser, the methanol yield of 130 μmol g ^{? 1} h ^{? 1} with natural sunlight is quite an important result, as it can be used for the development of large‐scale solar fuel generation facilities by harnessing the naturally abundant solar radiation. Copyright © 2017 John Wiley & Sons, Ltd.     

SOLAR GLOBAL AND ULTRAVIOLET RADIATION MEASUREMENTS OVER KUWAIT

Data presented here for continuous measurements of global (285-2800?nm) and ultraviolet solar radiation (295-385?nm) are on horizontal level, which have been carried out for three years in Kuwait from January 96 to December 98. The ratio of monthly daily ultraviolet to global solar radiation was found to lie between 4.07% and 5.4%. The highest and lowest intensity monthly-daily recorded values for global radiation were 9.29 and 0.45?Kwh/m² but for ultraviolet 445 and 31?Wh/m² respectively. Dependence of the global and ultraviolet solar radiation on the atmospheric humidity was investigated. Linear regression functions were obtained with coefficient of determination R²> 0.7. The decrease in ultraviolet solar radiation from 270 to 240?Wh/m² justifiably as a result of increasing the chemical pollutants SO₂ and NO₂.     

Experimental study on dynamic performance analysis of a flat-plate solar solid-adsorption refrigeration for ice maker

A flat-plate solid-adsorption refrigeration ice maker has been built for demonstration purposes. The working pair consists of methanol used as the refrigerant and activated carbon as the adsorption medium. The adsorbent bed is constructed of two flat-plate collectors, with a total surface area of 1.5 m². Solar radiation can be simulated with quartz lamps and some important parameters such as temperature and pressure of each subsystem can be handled by a computer. The experimental results show that this machine can produce 4–5 kg of ice after receiving 14–16 MJ of radiation energy with a surface area of 0.75 m², while producing 7–10 kg of ice after receiving 28–30 MJ of radiation energy with 1.5 m². These are the most advanced results for a solar ice maker so far. All these successful achievements will speed up the commercial processing of a solar ice maker.     

The fluctuation of solar radiation in Thailand

Fluctuations in the daily solar radiation are examined in an unbroken 5-yr sequence of measurements at Bangkok, and are also estimated from daily sunshine measurements at Bangkok and 3 other stations in Thailand. Seasonal effects are shown by separate studies for eight $\text{1}\text{1}\text{2}$ month periods of the year defined by standard solar declination values.During the dry season in winter and spring the frequency distribution of daily totals of global solar radiation at Bangkok has a peak near 20 MJ m^?2 d^?1 and is skewed towards low values. During the wet season in summer and autumn the distribution is more dispersed. Elsewhere the distributions are similar to those at Bangkok.The time series of daily totals of global solar radiation at Bangkok is analysed as a second order random process. The observed annual frequencies of runs of consecutive days with low radiation at Bangkok are given. There are 32.0 isolated days, 9.4 pairs of days, and 3.4 runs of 3 days per year with radiation less than 12.57 MJ m^?2 d^?1. These results are adequately described by the second order theory. The runs are most likely to occur in summer and autumn. Elsewhere in Thailand the annual frequencies of the runs and their seasonal distributions are almost the same as at Bangkok, except that in the south the runs are more likely to occur later in the year.     

Hourly simulation and performance of solar electric-vapor compression refrigeration system

A solar electric-vapor compression refrigeration (SE-VCR) system has been proposed in this study. The SE-VCR system was investigated for different evaporating temperatures and months in Adana city located in the southern region of Turkey. First, the hourly cooling load capacities (heat gain) of a sample building during the 23rd days of May, June, July, August and September months were determined by using meteorological data such as hourly average solar radiations and atmospheric temperatures. The hourly total heat gain of the sample building comprised of wall, window, humans, illumination and devices were determined by using the Cooling Load Hourly Analysis Program (HAP) 4.4. Then, the hourly variations of various parameters such as coefficient of the performance, condenser capacity and compressor power consumption were calculated. In addition, the minimum photovoltaic panel surface area was determined to meet the compressor power demand according to the hourly average solar radiation data. For evaporating temperature T_e = 0 °C, the maximum compressor power consumption was obtained as 2.53 kW at 15:00 PM on August 23. The required photovoltaic panel surface area was found to be around 31.26 m². It was determined that the SE-VCR system could be used for home/office-cooling purposes during the day in the southern region of Turkey.     

Assessment of diffuse solar energy under general sky condition using artificial neural network

In this paper, artificial neural network (ANN) models are developed for estimating monthly mean hourly and daily diffuse solar radiation. Solar radiation data from 10 Indian stations, having different climatic conditions, all over India have been used for training and testing the ANN model. The coefficient of determination (R²) for all the stations are higher than 0.85, indicating strong correlation between diffuse solar radiation and selected input parameters. The feedforward back-propagation algorithm is used in this analysis. Results of ANN models have been compared with the measured data on the basis of percentage root-mean-square error (RMSE) and mean bias error (MBE). It is found that maximum value of RMSE in ANN model is 8.8% (Vishakhapatnam, September) in the prediction of hourly diffuse solar radiation. However, for other stations same error is less than 5.1%. The computation of monthly mean daily diffuse solar radiation is also carried out and the results so obtained have been compared with those of other empirical models. The ANN model shows the maximum RMSE of 4.5% for daily diffuse radiation, while for other empirical models the same error is 37.4%. This shows that ANN model is more accurate and versatile as compared to other models to predict hourly and daily diffuse solar radiation.