Measurement Correction of a Set of Analog Sun Sensors via Neural Network

A Neural Network (NN) based method to improve the accuracy of a set of analog Sun sensors is presented. Analog Sun Sensors are commonly used on satellites due to their reduced cost, small size and low power consumption. However, especially in Earth imaging satellites, they are prone to the Earth albedo effects. Magnitude and direction of albedo change depending on the reflection characteristics of the Earth's surface, position and attitude of the satellite and position of the Sun. The albedo may deteriorate Sun direction measurements by the analog Sun sensor as much as 20°. In this study, a multilayer NN, which is trained using the Sun direction vector and available attitude information, is applied to the Sun sensor readings to correct the voltage output for the corresponding measurements. Then the corrected Sun angles in sensor x and y axes are obtained by combining NN outputs with the sensor measurements. The proposed algorithm is tested in various simulation scenarios of differing training and interrogation periods for the NN. Results show that the Sun sensor measurements can be corrected up to an accuracy of 1° using the NN approach. Generalization of the NN by tuning the parameters enables using the same trained NN for extended durations of time.