Comprehensive review of solar radiation modeling based on artificial intelligence and optimization techniques: future concerns and considerations
DOI: https://doi.org/10.1007/s10098-022-02434-7
IF: 4.3
2022-12-06
Clean Technologies and Environmental Policy
Abstract:An alternative energy source such as solar is one of the most important renewable resources. A reliable solar radiation prediction is essential for various applications in agriculture, industry, transport, and the environment because they reduce greenhouse gases and are environmentally friendly. Solar radiation data series have embedded fluctuations and noise signals due to complexity, stochasticity, non-stationarity, and nonlinearity with uncertain and time-varying nature. Aside from being highly nonlinear, solar radiation is highly influenced by the environment and environmental parameters such as air temperature, cloud cover, surface reflectivity, and aerosols. In addition, the spatial measurements of these variables are not readily available. To tackle these challenges, it is necessary to consider data preprocessing techniques and to develop and test precise solar radiation predicting models at different forecast horizons. There is, however, controversy regarding the performance of such models in various studies. Comparisons are not conducted systematically among the different studies. Using a critical literature review, the authors hope to answer these questions and believe that further investigation of solar radiation can benefit researchers and practitioners alike. This study presents a comprehensive evaluation of solar radiation modeling using artificial intelligence in the last 15 years and provides a novel detailed analysis of the available models. The studies conducted in different climates of the world that were published in distinguished journals were considered (i.e., 90 papers in total) for this purpose. Newly discovered procedures for optimizing forecasts, data cleaning, feature selection, classification methods, and stand-alone or hybrid data-driven models for solar radiation prediction and modeling were evaluated. The results strikingly showed that the most used artificial intelligence methods were artificial neural network, adaptive neuro-fuzzy inference system, and decision tree family of models. In addition, the extreme learning machine, support vector machine, and particle swarm optimization were the most used optimization techniques in solar radiation modeling. In terms of forecast horizons, the most common forecast horizon found in papers was on the daily scale (51% of studies), followed by the hourly scale (26%), and the least common was the monthly scale (18%). Based on the regional studies, the highest number of solar radiation papers originated from Asia, with Europe in second place and African countries in third place. An increasing trend in the number of papers from 2011 to 2015 was noted, and the second peak started from 2018 till the present. Under each section, a summary of findings is provided. The paper concludes with future thoughts and directions on solar radiation modeling.
environmental sciences,green & sustainable science & technology,engineering, environmental