Control for Satellites Formation Flying Yunjun Xu 1 ; Norman Fitz-Coy 2 ; Rick Lind 3 ; and Andrew Tatsch 4 Abstract: In this paper, a controller is designed for a satellite formation flying system around the Earth based on an uncertainty model derived from a nonlinear relative position equation. In this model, nonzero eccentricity and varying semimajor axis are included as parametric uncertainties. J 2 perturbation, atmospheric drag, and actuation and sensor noise are bounded by functional uncertainties. The controller design based on the nominal mission an 800 km altitude circular reference orbitis capable of achieving desired perfor- mance, is robust to uncertainties, and satisfies fuel consumption requirements even in a challenge nonnominal mission a 0.1 eccentricity and 7,978 km semimajor axis elliptic reference orbitwith the same control gain. In this nonnominal mission, the designed controller is able to keep formation with almost the same level of the V budget 43.86 m/s/yearas used in the nominal mission 39.65 m/s/year. For comparison, linear quadratic regulator LQRand sliding mode controllers SMCare developed and extensively tuned to get the same V consumption as that of the designed controller for the nominal mission. However, as shown in the simulation, the designed linear robust controller LQRand nonlinear robust controller SMChave a serious V consumption penalty 1.72 km/s/year for SMC or are unstable for LQRin the nonnominal mission. DOI: 10.1061/ASCE0893-1321200720:110 CE Database subject headings: Satellites; Dynamics; Models; Control systems. Introduction Formation flying systems FFShave been investigated and pro- posed for military or nonmilitary services. Due to their potential advantages over the conventional large size monolithic satellite, such as the low cost, flexibility, improved observation efficiency, increased reliability, and enhanced survivability, many missions are considering the use of satellite or spacecraft FFS to achieve goals that are difficult for the conventional large-size single sys- tem. An ongoing FFS mission list includes the New Millennium Program DS1, DS2, EO-1, EO-3, ST-5, and ST-6, micro- arcsecond X-ray imaging mission MAXIMLuquette and San- ner 2003, terrestrial planet finder TPFLuquette and Sanner 2003, Stellar Imager Luquette and Sanner 2003, SPHERES Miller et al. 2003, ACE Dahl et al. 1999, TerraSAR-X/ TanDEM-X synthetic aperture radar SAR D’ Amico et al. 2005, Gemini Gill et al. 2001, and GRACE Kirschner et al. 2004. The typical distance among space satellites in a formation ranges from 30 m to 200 km, with the control precision in a range from 1 m to 50 km. In relative dynamic models for FFS, the Clohessy-Wiltshire CWequation is most widely used Redding and Adams 1989; Kapila et al. 1999; Robertson et al. 1999; Morton and Weininger 1999; Aorpimai et al. 1999; Vassar and Sherwood 1985; Sedwick et al. 1999; Yan et al. 2000; Yedavalli and Sparks 2000. Based on this linear model, Redding and Adams 1989and Kapila et al. 1999designed LQ controllers for a pulse-based thruster, whereas Robertson et al. 1999implemented a PD controller with a thruster mapping. Sedwick et al. 1999derived an analyti- cal expression for an exact cancellation of differential J 2 for a satellite formation in a polar, circular orbit. Yan et al. 2000 designed a pulse-based control for the formation satellites’ peri- odic motion which guarantees the global stability. Yedavalli and Sparks 2000proposed a hybrid control scheme to achieve a balance between discrete control efforts and continuous dynamics performance. Ulybyshev 2003designed a discrete LQR control- ler for an 800 km altitude Earth orbit formation mission with fuel consumption on the order of V =50 m/s/year for formation sat- ellites station keeping within a 20–40 kmcontrol precision. All the above-mentioned controllers are designed using the CW equation. However, the CW equation is a linear approxima- tion of the real system, and does not contain atmospheric drag, high order harmonic terms, and uncertainties. Furthermore, the CW equation is only applicable in the near circular reference orbit case. Among these effects, higher order harmonic terms espe- cially J 2 , which changes the orbit period, drifts the perigee, and changes the nodal precession rate Alfriend et al. 2000 and ec- centricity are the dominant ones which affect the drifting of the CW equation. Therefore, mitigating the relative distance drift ef- fects coming from J 2 and elliptic reference orbits have been in- vestigated by many researchers. Schweighart and Sedwick 2001studied the effect of J 2 on the orbital elements, and a linearized J 2 form is added to the right of the CW equation. Yeh et al. 2000designed a Lyapunov-type nonlinear controller for the J 2 perturbation rejection based on the CW equation; the V budget is on the order of 100 m/s for only 1 Assistant Professor, School of Aerospace and Mechanical Engineering, Univ. of Oklahoma, Norman, OK 73019. E-mail: yjxu@ou.edu 2 Associate Professor, Mechanical and Aerospace Engineering, Univ. of Florida, Gainesville, FL 32611. E-mail: nfc@ufl.edu 3 Assistant Professor, Mechanical and Aerospace Engineering, Univ. of Florida, Gainesville, FL 32611. E-mail: ricklind@ufl.edu 4 Research Assistant, Mechanical and Aerospace Engineering, Univ. of Florida, Gainesville, FL 32611. E-mail: atatsch@ufl.edu Note. Discussion open until June 1, 2007. Separate discussions must be submitted for individual papers. To extend the closing date by one month, a written request must be filed with the ASCE Managing Editor. The manuscript for this paper was submitted for review and possible publication on April 15, 2005; approved on November 18, 2005. This paper is part of the Journal of Aerospace Engineering, Vol. 20, No. 1, January 1, 2007. ©ASCE, ISSN 0893-1321/2007/1-10–21/$25.00. 10 / JOURNAL OF AEROSPACE ENGINEERING © ASCE / JANUARY 2007