応用および計算数学ジャーナル

Combining Theoretical Imaging and Computer Vision Canting Heliostats

Abstract

Nick Cercone

Solar Power Tower technology requires accurate techniques to ensure the optical performance of the heliostats both in commissioning and operation phases. This paper presents a technique based on target reflection to detect and correct canting errors in heliostat facets. A camera mounted on the back of a target heliostat sees an object heliostat and the target facets in reflection. The pixels difference between detected and theoretical borders determines the canting errors. Experiments in a lab scale testbed show that canting errors can be corrected up to an average value of around as low as 0.15 mrad. Experiments were also performed on a real heliostat at Plataforma Solar de Almería. As a result, canting errors (up to 5 mrad) have been reduced below 0.75 mrad. Mirror slope errors, which can be noticeable in large facets, becomes the largest source of inaccuracy in the presented method.