An Easy-to-Understand Expert System for the Prediction of Direct, Diffuse and Global Solar Radiations in Indian Region

Abstract

Building is a modifier of indoor climate to provide comfort to its inhabitant. The building walls and roof often have sloppy surfaces, and the thermal performance evaluation of the building requires simultaneous values of direct, diffuse and global solar radiation. Such values are available only for limited locations. Consequently, the thermal evaluation of building in regions of complex climate and design of efficient air-conditioned system has continually suffered a setback. This paper presents an easy-to-understand expert system for the prediction of direct, diffuse and global solar radiation in the Indian region. The approach is based on a multi-frame output model of the artificial neural network analysis. The computational algorithm includes estimation of direct, diffuse and global components of solar radiation through clear sky conditions. The deviations of these estimates from measurements are considered to be due to random weather phenomena characterized by atmospheric clearness indices which are determined by an artificial neural network analysis based on interrelationship of direct, diffuse and global solar radiation. The target values of atmospheric clearness index achieved as a result of ANN analysis are expressed by an explicit polynomial representation which in turn enables the determination of atmospheric clearness index as a function of latitude and longitude of location, time of the day and month of the year. The whole computational scheme is embedded in a graphical user interface which is designed to assist a wide spectrum of users such as energy planner, engineers, architects, scientists and researchers to predict direct, diffuse and global solar radiation.