Solely Magnetic Genetic/Fuzzy-Attitude-Control Algorithm for a CubeSat

Abstract

A genetic algorithm was used to optimize performance of a fuzzy inference system acting as a controller for a magnetically actuated CubeSat. A solely magnetically controlled satellite is a nonlinear, underactuated system for which the uncontrollable axis varies as a function of orbit position and attitude; variation is approximately periodic with orbit position. Therefore, controllability is not guaranteed, making solely magnetic control a less than ideal option for spacecraft requiring a high degree of pointing accuracy or spacecraft subject to relatively large disturbances. However, for small spacecraft, such as CubeSats, with modest pointing and disturbance rejection requirements, solely magnetic actuation is a good option. The genetic-algorithm-tuned fuzzy controller solution was compared to a similar linear quadratic regulator solution that was tuned to minimize the cost function used by the genetic algorithm. Both were optimized with respect to a single set of initial conditions. The genetic-algorithm-tuned fuzzy controller was found to be a lower-cost solution than the linear quadratic regulator for the optimized set of initial conditions. Additionally, a Monte Carlo analysis showed the genetic-algorithm-tuned fuzzy controller tended to settle faster than the linear quadratic regulator over a variety of initial conditions.

Investigation of Gain Tuning and Sensor Noise for CubeSat B-dot Detumbling and 3-axis PD Magnetic Attitude Control

“…An efficient automated genetic algorithm was proposed and simulated by Sorgenfrei et al [7] using a simplified magnetic field model to tune the gains for a PD-like controller that incorporated actuator saturation limits. Similarly, Walker et al [8] applied a genetic algorithm using the International Geomagnetic Reference Field (IGRF) model to optimise the gains of an LQR controller. They found their approach to be a more computationally efficient tuning solution than manual tuning.…”

Magnetic attitude tracking control of gravity gradient microsatellite in orbital transfer

“…In orbital transfer, in order to save on the cost, power, weight and complexity of the system, a magnetorquer is normally adopted to provide the control torque. Several control methods have been developed for the attitude acquisition, manoeuvre and stabilisation of magnetic actuated satellites during the last decades (1)(2)(3)(4)(5) . The well-known B-dot control law was proposed early in 1972 (6) .…”

Desarrollo, implementación y evaluación de un algoritmo de lógica difusa para el control de actitud de nanosatélites

“…A lo largo de la historia, los algoritmos de control usados en los satélites han variado desde controles tradicionales (Wingrove, 1963), que se basan en teorías clásicas de control como el PID (Sidi, 2000) o Reguladores Lineales Cuadráticos (LQR por sus siglas en inglés: Linear Quadratic Regulator) (Souza, 2006), hasta controles inteligentes basados en lógica difusa (Ortega, 1995) (Steyn, 1994) (Nagi, 2010) (Guana, 2005) (Walker, 2015) o algoritmos con aprendizaje como las redes neuronales (Zou, 2010) (Fazlyab, 2016).…”

“…Estos dos factores hacen ideal el uso de controladores basados en lógica difusa para el control de actitud en nanosatélites. Sin embargo, a pesar de todas sus ventajas, la lógica difusa ha sido implementada en pocas misiones espaciales, aunque sí ha sido estudiada y los resultados obtenidos han sido muy satisfactorios (Ortega, 1995) (Steyn, 1994) (Nagi, 2010) (Guana, 2005) (Walker, 2015).…”

“…En trabajos anteriores se ha comparado el funcionamiento de un control tradicional, un LQR con un algoritmo de lógica difusa en un CubeSat (CubeSat, [online]) a través de simulaciones (Walker, 2015), obteniéndose que el control difuso implica un coste menor que el LQR, siendo además más rápido en llegar a la solución. En este trabajo se ha diseñado un control de actitud adaptable basado en lógica difusa para la misión de un nanosatélite.…”