عنوان مقاله [English]
The solar energy is the source of energy on the planet, one of the most important sources of clean energy, especially in Iran. The correct estimation of solar radiation is considered one of the important parameters in simulating plant growth and the estimation of evapotranspiration is very important. Measuring the intensity of solar radiation, although it has a relatively long history in Iran, due to the high cost, most of the stations in the country are not equipped with a radiation meter, and this defect is even seen in developing countries. Since the prediction of solar radiation intensity in across Iran daily range with GeneXproTools software has not been done so far, due to the wide range of different climates in different regions of the country, this research has been carried out across Iran. Therefore, the ability of the simulation model of GeneXproTools software to predict solar radiation based on altitude, longitude, latitude, precipitation, relative humidity, maximum temperature, minimum temperature, average temperature and solar radiation on a daily basis in 31 stations over Iran. The period from 2007 to 2016 has been taken.
Materials and Methods
Iran with an average annual rainfall of 241 mm is located in dry and semi-arid latitudes of the planet, between the two meridians of 44⁰ and 64⁰ in the east, and two circuits 25⁰ and 40⁰ in the north. About 94.84% of its surface is located in arid and semi-arid regions with low atmospheric rainfall and high evapotranspiration. Meteorological stations were selected based on climatic variation in this research. In this way, the study stations were divided into six submergence classification systems, hyper-arid, arid, semi-arid, Mediterranean, humid and very humid (A). Gene expression planning (GEP) is one of the newest methods of artificial intelligence. This is the generalized method of genetic algorithm (GA), which was based on Darwin's theory and invented by Ferreira in 1999 (Roshgar and Mirhidaryan, 2014). In order to compare the results of the gene expression algorithm were defined in the prediction of solar radiation, 8 scenarios are considered based on the parameters affecting solar radiation such as altitude (m), longitude (degree), latitude (degree), precipitation (mm) relative humidity (%), maximum temperature (°C), minimum temperature (°C) and average temperature (°C).
The main objective of this research is to select the best model for predicting daily average solar radiation in Iran using meteorological parameters. The number of training and testing data for the GEP model is presented in Table 1. Among the data collected, 80% of the data were used for training (total 8000 data, 6400 data from every parameter for every 31 stations) for the model. The experiment was performed for 20% of the data (a total of 1600 data from every parameter for each 31 stations) for the model.
Table 1. Training and testing data for study stations for the GEP model
The performance of the GEP model by evaluating the best fitness fittings, namely RMSE, MAE, NSE and R2, which at best (Best Mode) of the values of best fitness RMSE, MAE, NSE and R2were 1000, 0, 0, 1, and 1, respectively. Comparing the results of model assessment statistics in different scenarios, it was found that in all scenarios, c scenario due to consideration of latitude (degree) parameters, precipitation (mm), average relative humidity (%), maximum temperature (°C), minimum temperature (°C), average temperature (°C), and average solar radiation (MJ.m-2.d-1), the desired prediction has presented. The best model for this study is R2 for training and data testing of 0.72 and 0.73, respectively. The GEP model shows the optimal output model as a tree, and also the equation derived from this structure. Since the four genes in this study are composed, each gene has a subtree and its own equation, which ultimately results from the final equation bond function. The final equation is a relation 1.
Rs= SUB ET1+ SUB ET2 + SUB ET3 + SUB ET4 = sin((((((2RH)-cos(Tmax))-cos(Tmean))-p)/altitude)) + Tmean + sin(altitude) + ((((altitude + 7.30)-(2 Tmean))-p)/((sin(-8.41)+ 5.76)+atan(Tmin))))
From the results of the analysis, it was found that the predicted values are well adapted to the values measured in the Gene Expression Programming Model (GEP). In the GEP model, the e scenario is due to the fact that it does not consider altitude (degree), longitude (degree), latitude (degree), precipitation (mm), relative humidity (%), maximum temperature (°C) and minimum temperature (°C), with a lower correlation coefficient and more error and less efficiency. In this model, in the c scenario, with the addition of average temperature, and solar radiation parameters, a more favorable estimation of solar radiation is obtained. The results of the c and e scenarios are very close together, but the e scenario is weaker in the prediction of solar radiation, and the best model in this study is c scenario. More precise results will be obtained when latitude (degree), precipitation (mm), relative humidity (%), maximum temperature (°C), minimum temperature (°C), average temperature (°C) and solar radiation (MJ.m-2.d-1) are predicted for solar radiation. Also, the criteria for assessment under c scenario are higher with R2= 0.72, RMSE = 3.59, MAE = 2.82, and NSE = 0.72 in the education section of other scenarios. In general, the GEP method has the most accurate results in estimating daily solar radiation in Iran.
10.Muzzammil, M., J. Alam, M. Danish, 2015, Application of gene expression programming in flood frequency analysis. Journal of Indian Water Resources Society. No. 2, pp. 1-6.75.
11.Neelamegam, P., V. Arasu Amirtham, 2016, Prediction of solar radiation for solar systems by using ANN models with different back propagation algorithms. Journal of Applied Research Technology. pp. 206- 214.
12.Nia, M., M. Chegaar, M.F. Benatallah, M. Aillerie, 2013, Contribution to the quantification of solar radiation in Algeria. Journal of Energy Procedia, pp. 730- 737.
13.Roshangar, K. Sh. Mirahidaryan, 2014, Application of evolutionary method of gene expression planning in estimating the scour depth of bridge bases in non-adhesive substrates based on laboratory and field data. The 8th National Civil Engineering Congress of Babylon. (In Farsi)
14.Roshangar, K., H. Bonakdari, S. Akhgar, F. Harsami, 2015, Investigating the Functioning of Gene Expression Programming (GEP) in predicting energy dissipation on stepper overflow in a skimming regime. 10th International Congress on Civil Engineering.
15.Seydou, T., G. Aytac, 2012, Regional-Specific numerical models of evapotranspiration using Gene-Expression Programming interface in Sahel. Journal of Water Resources Management. pp. 4367-4380.
16.Traore, S., Y. Luo, G. Fipps, 2017, Gene-Expression Programming for Short-Term Forecasting of Daily Reference Evapotranspiration Using Public Weather Forecast Information. Journal of Water Resources Management.
17. Yazdanpanah, H., R. Mirmaghibian, H. Beraki, H, 2009, Estimation of total solar radiation in the horizontal plane of the Earth in Isfahan. Journal of Geography and Environmental Planning, No. 1.