[EN]

Precise orbit positioning for a satelite using sensor fusion and model-based estimation techniques

The project PROFET aims at developing and implementing a better solution compared to the existing ones regarding the precise positioning of a satellite onto its orbit. The efficiency of our approach comes from the low frequency of interaction between the space object and the GNSS (Global Navigation Satellite System) network. To achieve this, the definition of mission requirements, altogether with data retrieval from the orbital satellite is needed. For such an object, current solutions for precise positioning exist and some are based on machine learning techniques. However, even if these approaches are somehow computationally efficient, they are strongly dependent on the neural network architecture, with an inherent probability of quite large estimation errors. Our project proposes the use of analytical equations, derived from the first principles of mechanical physics, to provide a classical model in state-space form, readily usable for efficient numerical algorithms, much more appealing in either classical or advanced position estimation techniques. The successful design of the estimation algorithms will be followed by the definition of the hardware requirements and the design of an electronic module, the estimation software being implemented at the firmware level, afterwards. The project PROFET will succeed by proving through a hardware in the loop first, followed by the satellite interfacing of the electronic module.

[RO]

Pozitionarea precisa pe orbita a unui satelit folosind tehnici de fuziune de date si estimare bazata pe model

Proiectul PROFET isi propune sa dezvolte si sa implementeze o solutie mai buna, comparativ cu cele existente deja, privind pozitionarea precisa a unui satelit pe orbita. Eficienta abordarii propuse in acest proiect este generata de o solutie de pozitionare care necesita o frecventa redusa a interactiunii dintre obiectul spatial si reteaua GNSS (Sistemul Global de Navigare cu Sateliti). In acest scop, sunt necesare definirea tehnica a misiunii impreuna cu achizitia datelor reale de la satelit. Pentru un astfel de obiectiv, exista solutii curente disponibile de pozitionare precisa, dintre care unele sunt bazate pe tehnici de inteligenta artificiala cu invatare automata. Totusi, desi aceste abordari sunt relativ eficiente din punct de vedere computational, acestea sunt puternic dependente de arhitectura retelei neurale, cu o probabilitate inerenta de a obtine erori destul de mari de estimare. Acest proiect isi propune folosirea ecuatiilor analitice, derivate din principiile fundamentale ale mecanicii clasice, pentru a propune un model clasic in reprezentare pe stare, mai usor de folosit in cadrul unor algoritmi numerici eficienti, mult mai atractivi de integrat in tehnici clasice si chiar avansate de estimare a pozitiei. Incheierea cu succes a proiectarii algoritmilor de estimare va fi urmata de definirea cerintelor tehnice pentru echipamentul hardware si proiectarea modulului electronic, continuata de implementarea software a procedurilor de estimare si la nivel firmware.

• Registration code: PN-III-P2-2.1-PED-2021-1626
• Contract number: 680PED
• Duration: 24 months (July 2022 – June 2024)
• Funding: 648795 RON (approx. 131673 euro)