An MPC-based algorithm for estimating the spatial DNI of cloud shaded regions using a robotic sensor system

Abstract

This paper presents an original algorithm based on the Model Predictive Control strategy for estimating the direct normal irradiance of cloud shaded regions using a mobile robotic sensor system to improve the control of a solar thermal power plant. This new algorithm generates the waypoints of the robot team solving a minimisation problem where the objective function combines several criteria, including the measurements taken by the team. The novel method has been tested by simulation with groups of different numbers of unmanned aerial vehicles using the shape of real cloud shadows projected on the ground extracted from images and it improves the estimation error and the estimation time of previous algorithms.