A simulation environment for UAV-based real-time condition monitoring of solar tower power plants

Abstract

The development and testing of unmanned aerial vehicle (UAV)-based condition monitoring systems is time consuming, costly and poses safety risks. While numerous examples show that simulation environments are well suited to support the development process, existing environments fall short of simulating quantities specific to the condition monitoring of solar tower power plants. To bridge this gap, we present a simulation environment that provides quantities necessary to investigate such systems in simulation, prior to their application in real solar tower power plants. The presented environment models the state of the solar field and computes observations of the field and reflections of a point light source, as seen from a virtual camera. In addition, it allows for the navigation of a simulated UAV in the virtual solar field in response to realistic UAV control signals. The simulated concentrator corner points were found to match the concentrator corner points determined by a bundle adjustment measurement up to an RMSE=23.7mm before and RMSE=4.6mm after accounting for translational, rotational and scale errors. The simulated reflections of a point light source were found to match the measured reflections up to an RMSE of 2.25mrad in X-direction and 2.09mrad in Y-direction in the concentrator coordinate system. After eliminating errors in the camera position estimate, concentrator orientations and mirror surface slope errors, the remaining RMSE is 0.35mrad in X-direction and 0.22mrad in Y-direction. We conclude that the proposed simulation environment is a valuable tool for the development of UAV-based condition monitoring systems of solar tower power plants.