The First Selective Entropy as a Function of the State of a Scattered Generation System

Abstract

In the given paper, it is proposed to use the first selective entropy as a parameter that characterizes the variability of energy generation, consumption and storage processes in distributed generation systems with renewable energy sources, for short-term forecasting to ensure efficient operation of such systems. As an example, the first selective entropy of the power at the output of solar panels is determined, considering other environmental parameters unchanged. The method of calculation of the first sample entropy of the "ideal" daily curve of solar radiation power is given. The ideal daily curve of solar radiation for May 2, 2019 for the city of Zagreb, Republic of Croatia is given. The values of the first selective entropy of the ideal and real solar radiation power curves depending on the number of partition subintervals are given. The schedule of change of power at the output of solar panels for the week from April 27 to May 3, 2019, taken from LARES laboratory, Zagreb, Croatia, is given. The values of the first selective entropy at the output of the solar panels depending on the number of subintervals of the partition are given. The "imperfection" of the power curve at the output of solar panels leads to a decrease in the first selective entropy, which indicates an increase in the variability of the energy generation process. To correct the real power distribution at the output of solar panels and bring it closer to the ideal, it is necessary to use an electric energy storage device with an entropy equal to the modulus of entropy of solar panels, but taken with the opposite sign. The values of the first sample power entropy at the output of solar panels for the week are calculated and their change is shown on the figure. Using Newton's linear interpolation, the values of the first selective entropy for the day 8 and day 2 were predicted. The prediction error of the first selective entropy for the 8th day is 4%, and for the 2nd day - 7%. In the case of significant data deviations, the prediction error increases almost 2 times, but to reduce the magnitude of the prediction error can be performed sequentially with a correction on each prediction interval. It is shown that the implementation of predictive control of the distributed generation system using the first selective entropy, as an integral characteristic of the system state, allows to estimate and compare random processes of energy generation, consumption and accumulation, without finding the laws of their distribution.