A relaxation approach for hybrid multi-objective optimal control: application to multiple debris removal mission

Research output: Contribution to conferenceProceeding

Abstract

This paper presents a novel approach to the solution of multi-phase multi-objective hybrid optimal control problems. The proposed solution strategy extends previous work which integrated the Direct Finite Elements Transcription (DFET) method to transcribe dynamics and objectives, with a memetic strategy called Multi Agent Collaborative Search (MACS). The problem is reformulated as two non-linear programming problems: a bi-level and a single level one. In the bi-level formulation the outer level, handled by MACS, generates trial control vectors that are then passed to the inner level, which enforces the feasibility of the solution. Feasible control vectors are then returned to the outer level to evaluate the corresponding objective functions. The single level formulation is also run periodically to ensure local convergence to the Pareto front. In order to treat mixed integer problems, the heuristics of MACS have been modified in order to preserve the discrete nature of integer variables. For the single level refinement and the inner level of the bi-level approach, discrete variables are relaxed and treated as continuous. Once a solution to the relaxed problem has been found, a smooth constraint is added to systematically force the relaxed variables to assume integer values. The approach is first tested on a simple motorised travelling salesmen problem and then applied to the mission design of a multiple debris removal mission.

Conference

Conference29th AAS/AIAA Space Flight Mechanics Meeting
CountryUnited States
CityKa'anapali, Hawaii
Period13/01/1917/01/19

Fingerprint

Debris
Traveling salesman problem
Nonlinear programming
Transcription

Keywords

  • multi-objective optimal control
  • DFET

Cite this

Ricciardi, L. A., & Vasile, M. (2019). A relaxation approach for hybrid multi-objective optimal control: application to multiple debris removal mission. 1-20. 29th AAS/AIAA Space Flight Mechanics Meeting, Ka'anapali, Hawaii, United States.
@conference{903836fee8e246fea4b8c80ce06f99e6,
title = "A relaxation approach for hybrid multi-objective optimal control: application to multiple debris removal mission",
abstract = "This paper presents a novel approach to the solution of multi-phase multi-objective hybrid optimal control problems. The proposed solution strategy extends previous work which integrated the Direct Finite Elements Transcription (DFET) method to transcribe dynamics and objectives, with a memetic strategy called Multi Agent Collaborative Search (MACS). The problem is reformulated as two non-linear programming problems: a bi-level and a single level one. In the bi-level formulation the outer level, handled by MACS, generates trial control vectors that are then passed to the inner level, which enforces the feasibility of the solution. Feasible control vectors are then returned to the outer level to evaluate the corresponding objective functions. The single level formulation is also run periodically to ensure local convergence to the Pareto front. In order to treat mixed integer problems, the heuristics of MACS have been modified in order to preserve the discrete nature of integer variables. For the single level refinement and the inner level of the bi-level approach, discrete variables are relaxed and treated as continuous. Once a solution to the relaxed problem has been found, a smooth constraint is added to systematically force the relaxed variables to assume integer values. The approach is first tested on a simple motorised travelling salesmen problem and then applied to the mission design of a multiple debris removal mission.",
keywords = "multi-objective optimal control, DFET",
author = "Ricciardi, {Lorenzo A.} and Massimiliano Vasile",
year = "2019",
month = "1",
day = "17",
language = "English",
pages = "1--20",
note = "29th AAS/AIAA Space Flight Mechanics Meeting ; Conference date: 13-01-2019 Through 17-01-2019",

}

Ricciardi, LA & Vasile, M 2019, 'A relaxation approach for hybrid multi-objective optimal control: application to multiple debris removal mission' 29th AAS/AIAA Space Flight Mechanics Meeting, Ka'anapali, Hawaii, United States, 13/01/19 - 17/01/19, pp. 1-20.

A relaxation approach for hybrid multi-objective optimal control : application to multiple debris removal mission. / Ricciardi, Lorenzo A.; Vasile, Massimiliano.

2019. 1-20 29th AAS/AIAA Space Flight Mechanics Meeting, Ka'anapali, Hawaii, United States.

Research output: Contribution to conferenceProceeding

TY - CONF

T1 - A relaxation approach for hybrid multi-objective optimal control

T2 - application to multiple debris removal mission

AU - Ricciardi, Lorenzo A.

AU - Vasile, Massimiliano

PY - 2019/1/17

Y1 - 2019/1/17

N2 - This paper presents a novel approach to the solution of multi-phase multi-objective hybrid optimal control problems. The proposed solution strategy extends previous work which integrated the Direct Finite Elements Transcription (DFET) method to transcribe dynamics and objectives, with a memetic strategy called Multi Agent Collaborative Search (MACS). The problem is reformulated as two non-linear programming problems: a bi-level and a single level one. In the bi-level formulation the outer level, handled by MACS, generates trial control vectors that are then passed to the inner level, which enforces the feasibility of the solution. Feasible control vectors are then returned to the outer level to evaluate the corresponding objective functions. The single level formulation is also run periodically to ensure local convergence to the Pareto front. In order to treat mixed integer problems, the heuristics of MACS have been modified in order to preserve the discrete nature of integer variables. For the single level refinement and the inner level of the bi-level approach, discrete variables are relaxed and treated as continuous. Once a solution to the relaxed problem has been found, a smooth constraint is added to systematically force the relaxed variables to assume integer values. The approach is first tested on a simple motorised travelling salesmen problem and then applied to the mission design of a multiple debris removal mission.

AB - This paper presents a novel approach to the solution of multi-phase multi-objective hybrid optimal control problems. The proposed solution strategy extends previous work which integrated the Direct Finite Elements Transcription (DFET) method to transcribe dynamics and objectives, with a memetic strategy called Multi Agent Collaborative Search (MACS). The problem is reformulated as two non-linear programming problems: a bi-level and a single level one. In the bi-level formulation the outer level, handled by MACS, generates trial control vectors that are then passed to the inner level, which enforces the feasibility of the solution. Feasible control vectors are then returned to the outer level to evaluate the corresponding objective functions. The single level formulation is also run periodically to ensure local convergence to the Pareto front. In order to treat mixed integer problems, the heuristics of MACS have been modified in order to preserve the discrete nature of integer variables. For the single level refinement and the inner level of the bi-level approach, discrete variables are relaxed and treated as continuous. Once a solution to the relaxed problem has been found, a smooth constraint is added to systematically force the relaxed variables to assume integer values. The approach is first tested on a simple motorised travelling salesmen problem and then applied to the mission design of a multiple debris removal mission.

KW - multi-objective optimal control

KW - DFET

UR - https://www.space-flight.org/docs/2019_winter/2019_winter.html

M3 - Proceeding

SP - 1

EP - 20

ER -

Ricciardi LA, Vasile M. A relaxation approach for hybrid multi-objective optimal control: application to multiple debris removal mission. 2019. 29th AAS/AIAA Space Flight Mechanics Meeting, Ka'anapali, Hawaii, United States.