FPGA-based implementation of intelligent predictor for global solar irradiation, Part I: Theory and simulation

Recent advances in artificial intelligent techniques embedded into a Field Programmable Gate Array (FPGA) allowed the application of such technologies in real engineering problems (robotic, image and signal processing, control, power electronics, etc.), however, the application of such technologies in the solar energy field is very limited. The embedded intelligent algorithm into FPGA can play a very important role in energy and renewable energy systems for control, monitoring, supervision, etc. In this paper, the software as well as the implementation of intelligent predictors for solar irradiation on reconfigurable FPGA is described. FPGA technology was employed due to its development, flexibility and low cost. An experimental dataset of air temperature, solar irradiation, relative humidity and sunshine duration in a specific area is used; this database has been collected from 1998 to 2002 at Al-Madinah (Saudi Arabia). Initially, a Mul-tiLayer Perceptron (MLP) is trained by using a set of 1460 patterns and then a set of 365 patterns are used for testing and validating the MLP-predictor. Six MLP-predictors (configurations) are proposed and developed by varying the MLP inputs data, while the output is always the global solar irradiation for different Subsequently, the different MLP-predictors developed are written and simulated under the Very High Speed Integrated Circuit Hardware Description Language (VHDL) and ModelSim Ò. The best designed architecture for different MLP-predictors is then implemented under the Xilinx Ò Virtex-II FPGA (XC2v1000). The developed hardware devices permit the prediction of global solar irradiation using available air temperature, relative humidity and sunshine duration; therefore, the designed configurations are very suitable especially in areas, where there are no instruments for measuring the solar irradiation data. Global solar irradiation is considered as the most important parameter in the design of renewable and solar energy systems, particularly for the sizing of photovoltaic (PV) systems. Numerous studies in the literature have shown the possibility to find a correlation between solar irradiation and other meteorological parameters such as air temperature, sunshine duration, relative humidity Global solar irradiation is usually presented as a time series and is especially challenging to predict its values in situations where an underlying model for generating data is not known. Frequently, these data (measured data) are not always available, especially in remote areas because of the absence of meteorological stations or measuring instruments. For this purpose, several models have been developed in order to generate these data based on stochastic models such as AR (auto-regressive), ARMA …