The problem of Fault Detection and Isolation (FDI), is of much concern among the control engineering community. In addition, the space industry is pushing towards more compact spacecraft solutions (e.g. Cubesats) that have the advantage of being relatively low cost and easily configurable, posing however new constraints in terms of mass and power available onboard. The FDI problem is directly influenced by these new trends, that limit the amount of sensors carriable on board, forcing the community to come up with innovative ways to guarantee the fulfillment of those new requirements. With this in mind, the thesis carries out the development of a tool for detecting and isolating specific kinds of faults, that can occur in thruster-controlled space missions’ maneuvers, making use of only accelerometer and gyroscope information. This kind of approach is particularly suitable for smaller spacecraft which, having more stringent requirements in terms of mass and power consumption, can benefit from the minimal number of sensors needed. In addition, this tool can be used soon after the spacecraft is released by the carrier, when not all the sensors are active, and usually information comes from simple sensor like an IMU unit. The method consists in three-stages (one for fault detection, two for fault identification), and combines simple sensor information monitoring with model-based approaches, leading to the design of multiple Nonlinear Unknown Input Observers. The implementation is carried out using Basilisk, the astrodynamics simulation software developed by the University of Colorado AVS Lab and the Laboratory for Atmospheric and Space Physics (LASP). It makes use of Python’s ease of scripting and reconfiguration, together with the execution speed of low level programmed C/C++ modules, making the developed algorithms faster with respect to classical MATLAB simulations, and directly usable on spacecraft’s onboard computers.
Il problema di Fault Detection and Identification (FDI) è di grande interesse nel campo di ingegneria del controllo. In più, l’industria spaziale si sta indirizzando verso l’utilizzo di satelliti sempre più compatti (come ad esempio i Cubesat). Essi hanno il vantaggio di essere relativamente economici e facilmente configurabili, creando tuttavia nuovi vincoli in termini di massa e consumo disponibile a bordo. La FDI è direttamente influenzata da questi nuovi trends, che limitano la quantità di sensori trasportabili a bordo, forzando quindi gli ingegneri a trovare nuove innovative soluzioni che garantiscano l’adempimento a questi nuovi vincoli. Tenendo conto di tutto ciò, la tesi propone il design di un nuovo metodo per captare ed isolare specifici tipi di malfunzionamenti che possono avvenire durante manovre nello spazio che utilizzano propulsori di tipo "Cold Gas Thrusters", facendo uso esclusivamente di misurazioni di accelerometri e giroscopi. Questo tipo di approccio è particolarmente utile per piccoli veicoli che, avendo vincoli di massa e consumo elettrico piu stringenti, beneficiano della possibilità di utilizzare solo questi due sensori. Ancora, questo strumento può essere utilizzato appena dopo il rilasio del satellite da parte del vettore, quando non tutti i sensori sono ancora attivi, e le misurazioni arrivano solo da sensori quali IMU. Il metodo consiste in tre stadi (uno per fault detection e due per fault identification), e combina il semplice monitoramento dell’accelerazione con tecniche "Model-Based", portando al design di Nonlinear Unknown Input Observers. L’implementazione è portata a termine in Basilisk, il software di simulazione astrodinamica creato nell’Università del Colorado Boulder, dall’AVS Laboratory ed il Laboratory for Atmospheric and Space Physics (LASP). Questo software sfrutta la semplicità di riconfigurazione e scripting di Python, insieme con la velocità di esecuzione di moduli codificati nei linguaggi di basso livello C e C++. Ciò rende gli algoritmi qui presenati di più veloce esecuzione rispetto alle classiche simulazioni Matlab, e direttamente utilizzabili nei computer di bordo di veri satelliti.
A model-based thruster fault detection and isolation tool with implementation in basilisk astrodynamics simulation framework
NAPOLITANO, GIULIO
2019/2020
Abstract
The problem of Fault Detection and Isolation (FDI), is of much concern among the control engineering community. In addition, the space industry is pushing towards more compact spacecraft solutions (e.g. Cubesats) that have the advantage of being relatively low cost and easily configurable, posing however new constraints in terms of mass and power available onboard. The FDI problem is directly influenced by these new trends, that limit the amount of sensors carriable on board, forcing the community to come up with innovative ways to guarantee the fulfillment of those new requirements. With this in mind, the thesis carries out the development of a tool for detecting and isolating specific kinds of faults, that can occur in thruster-controlled space missions’ maneuvers, making use of only accelerometer and gyroscope information. This kind of approach is particularly suitable for smaller spacecraft which, having more stringent requirements in terms of mass and power consumption, can benefit from the minimal number of sensors needed. In addition, this tool can be used soon after the spacecraft is released by the carrier, when not all the sensors are active, and usually information comes from simple sensor like an IMU unit. The method consists in three-stages (one for fault detection, two for fault identification), and combines simple sensor information monitoring with model-based approaches, leading to the design of multiple Nonlinear Unknown Input Observers. The implementation is carried out using Basilisk, the astrodynamics simulation software developed by the University of Colorado AVS Lab and the Laboratory for Atmospheric and Space Physics (LASP). It makes use of Python’s ease of scripting and reconfiguration, together with the execution speed of low level programmed C/C++ modules, making the developed algorithms faster with respect to classical MATLAB simulations, and directly usable on spacecraft’s onboard computers.File | Dimensione | Formato | |
---|---|---|---|
MastersThesisGiulioNapolitano.pdf
solo utenti autorizzati dal 21/05/2021
Descrizione: Thesis
Dimensione
9.4 MB
Formato
Adobe PDF
|
9.4 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/165549