Improved solar tracking system with linear regression error correction
Date
1996-07
Authors
Wood, Norman B., author
Cox, Stephen K., author
Cornwall, Christopher L., author
Journal Title
Journal ISSN
Volume Title
Abstract
A motor driven two-axis optical mount combined with PC-based solar position and correction software is easily set up and provides highly accurate tracking of the sun. During setup, the tracker needs only to be approximately aligned, then a series of simple, manual corrections can be made which lead to automatic correction for misalignment. With accurate manual corrections, the tracking accuracy can approach 0.1 degrees. Various tracking options allow the user to scan repeatedly across the solar disc, track a position offset from the sun, or reflect the solar image into another instrument. The system includes a two-axis mount driven by servo motors with optical encoder position indication; servo amplifiers; a personal computer equipped with a two-axis motor controller; software for calculating solar position; and error correction software. The optical encoders have a resolution of 0.1 arcseconds per step, and the solar position software agrees to within 1.25 arcminutes with U.S. Naval Observatory calculations of solar position. The error correction software applies linear regression via singular value decomposition to a series of manual tracking corrections. The regression creates a best-fit compensation for misalignment of the mount. Several factors are evaluated for their influence on tracker performance. These factors include the initial misalignment of the tracker, errors in the manual corrections, and the frequency of the manual corrections. Performance is largely insensitive to the magnitude and orientation of the initial misalignment, but sensitive to the accuracy and frequency of manual corrections. Based on this sensitivity, a manual correction scheme is developed which improves the performance of the correction software. The performance of the tracker, employing the correction scheme, is evaluated using both a computer simulation of the tracker and field testing. Computer tests with simulated random manual correction errors show that the correction algorithm can achieve accuracy within two times the standard deviation of the correction errors. This accuracy is maintained following final manual correction for test periods as long as 54 hours. In the field test, highly accurate manual corrections were made by reflecting the solar image to a wall about 100 feet from the tracker. The observed tracking errors are 0.11 +/- 0.05 degrees for 48 hours following the final manual corrections.
Description
July 1996.
Rights Access
Subject
Solar radiation -- Measurement
Solar activity -- Measurement