Title: Solution for the Third Global Trajectory Optimization Competition
1Solution for the Third Global Trajectory
Optimization Competition
Workshop 3 GTOC 27 June 2008 TORINO
- Team Politecnico di Milano - M.Massari, R.
Armellin, G. Bellei, P. Di Lizia, M. Lavagna, G.
Mingotti, S. Tonetti, F. Topputo - Department of Aerospace Engineering
- Politecnico di Milano
2Outline
- Optimization strategy
- Problem Analysis
- Global Optimization
- Solution Refinement
- Results
3Two Phase Approach
- The proposed problem is not a typical global
trajectory optimization, it is a sequence of - a combinatorial problem on the asteroids sequence
- a continuous problem on the trajectory between
asteroids - A two phase approach has been followed
- The Asteroid Selection and preliminary trajectory
definition - An asteroids pruning
- A Stochastic algorithm has been used
- The Trajectory Refinement
- Parametric Optimization Problem
- Multiple Shooting and Collocation
- A first look to the objective function allows to
identify the conditions that would characterize
the global optimum - the stay time needs to be maximized
- the thrust time needs to be reduced.
4Problem AnalysisAsteroid selection
- The list of asteroids has been pruned by
analyzing - Semi-major axis AU
- Eccentricity
- Inclination deg.
- The constraints on the orbital parameters can
guide to a smart selection of the asteroids to
visit. - After the pruning process only six asteroids
remain 49, 61, 76, 85, 88, 96.
5Problem ModelizationPreliminary Trajectory
Definition
- The Problem of Global Trajectory Optimization has
been modeled using for each transfer - Multiple Revolution Lamberts problem solution
- Lamberts problem for Exponential Sinusoids
solution (for multi-revolution transfer) - Solution of an optimal control problem by means
of an indirect method formulation. - No gravity assists of the Earth have been
considered in the preliminary phase. - The Problem is completely identified considering
- selection of the asteroids IDs
- determination of the departure epoch, the four
transfer times, and the three stay times - choice of the number of revolutions for the
Lambert and the exponential sinusoid first guess - the exponential sinusoid characterization
parameter k2.
6Global Optimization Particle Swarm Optimization
- The Problem has been solved using a stochastic
search method. - The Particle Swarm Optimization
- Is based on the idea of swarms
- Initially the particles are randomly initialized
- Particles move in the search space on the basis
of a velocity which is influenced by - Inertia
- Personal Best Solution
- Global Best Solution
7Solution Refinement
- The solution refinement is necessary as the
solution found by the global optimizer does not
satisfy the problem constraints - The trajectory refinement can be formulated as
- the solution of an optimal control problem in
which the objective function must be maximized, - subject to
- the differential constraints given by the
dynamics, - The boundary constraints deriving from rendezvous
conditions, - the path constraints deriving from the threshold
on the available thrust. - The solution found with the global optimization
can be used as initial guess in the local
optimization process
8Solution RefinementOptimal Control Problem
- The Optimal control problem has been solved
- Transcribing the continuous variables in
parametric variables - Expressing the differential constraints as
algebraic constraints on the parametric variables - Solving the resulting NLP problem with a
Sequencial Quadratic Programming Solver - Two different transcription techniques have been
applied - Multiple Shooting method
- Collocation method
- Earths flyby have been included based on
energetic considerations.
9Results
- Solution sequence
- Phase 1 Earth Asteroid 2001 GP2 (GTOC3 N. 96)
- Phase 2 Asteroid 2001 GP2 Earth
- Phase 3 Earth Asteroid 1991 VG (GTOC3 N. 88)
- Phase 4 Asteroid 1991 VG Asteroid 2000 SG344
(GTOC3 N. 49) - Phase 5 Asteroid 2000 SG344 Earth
- Departure epoch 58169 MJD
- Arrival epoch 61693.21 MJD
- Total time of flight 9.6487 years
- Minimum stay time 100.0 days
- Initial s/c mass 2000 kg
- Final s/c mass 1663.1148536687001 kg
- Propellant mass used 336.89 kg
- Objective function value 0.83758069856604
10Results
11Results
- Phase 1
- Hyperbolic excess velocity 0.4999 km/s
- Departure epoch 58169.0 MJD
- Time of flight 569.65 days
- Arrival epoch 58738.65 MJD
- Initial s/c mass 2000 kg
- Final s/c mass 1907.70 kg
Phase 3 Fly-By radius 6878.63 km Departure
epoch 59142.33 MJD Time of flight 531.67
days Arrival epoch 59674 MJD Initial s/c mass
1884.87 kg Final s/c mass 1776.05 kg
Phase 2 Stay time at Asteroid 2001 GP2 110.34
days Departure epoch 58849 MJD Time of flight
293.33 days Arrival epoch 59142.33 MJD Initial
s/c mass 1907.70 kg Final s/c mass 1884.87 kg
Phase 4 Stay time at Asteroid 1991 VG 110
days Departure epoch 59784 MJD Time of flight
390 days Arrival epoch 60174 MJD Initial s/c
mass 1776.05 kg Final s/c mass 1709.37 kg
Phase 5 Stay time at Asteroid 2000 SG344 110
days Departure epoch 60284.00 MJD Time of
flight 1409.21 days Arrival epoch 61693.21
MJD Initial s/c mass 1709.37 kg Final s/c mass
1663.148536687001 kg
12Results
13Team 21 - Politecnico di Milano
- Mauro Massari
- Roberto Armellin
- Gabriele Bellei
- Pierluigi Di Lizia
- Michéle Lavagna
- Giorgio Mingotti
- Stefania Tonetti
- Francesco Topputo
- Aerospace Engineering Department
- Politecnico di Milano
- Via La Masa, 34
- 20156 Milano, Italy
- Ph. 39 02 2399 8308
- Fax 39 02 2399 8028
- Contact Person
- Mauro Massari