How to Deal with Mixed-Variable Optimization Problems: An Overview of Algorithms and Formulations

2019 IEEE Congress on Evolutionary Computation (CEC)

Filippi

et al. 2019

This paper addresses the problem of autonomous scheduling of space objects' observations from a network of tracking stations to enhance the knowledge of their orbit while respecting allocated resources. This task requires the coupling of a state estimation routine and an optimisation algorithm. As for the former, a sequential filtering approach to estimate the satellite state distribution conditional on received indirect measurements has been employed. To generate candidates, i.e. observation campaigns, a Structured-Chromosome Genetic Algorithm optimiser has been developed, which is able to address the issue of handling mixed-discrete global optimisation problems with variable-size design space. The search algorithm bases its strategy on revised genetic operators that have been reformulated for handling hierarchical search spaces. The presented approach aims at supporting the space sector by tracking both operational satellites and non-collaborative space debris in response to the challenge of a constantly increasing population size in the near Earth environment. The potential of the presented methodology is shown by solving the optimisation of a tracking window schedule for a very low Earth satellite operating in a highly perturbed dynamical environment.

Section: The Algorithmmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Autonomous Generation of Observation Schedules for Tracking Satellites with Structured-Chromosome GA Optimisation

Greco

2019 IEEE Congress on Evolutionary Computation (CEC)

Filippi

et al. 2019

“…In classical global optimisation, this often relies on techniques that aim at maximising the information-gain distributing samples in the search space following some strategy. However, none of them can be carelessly adopted in variable-size optimisation problems [9]. In the SCGA an iterative algorithm that creates feasible candidates has been developed.…”

Section: Initial Populationmentioning

confidence: 99%

“…Genetic Algorithms (GAs) can overcome most of these issues by means of an appropriated encoding. As stated in [9], GAs is one of the most suited classes of optimisers for variable-size optimisation.…”

Section: Introductionmentioning

confidence: 99%

Structured-chromosome GA optimisation for satellite tracking

Proceedings of the Genetic and Evolutionary Computation Conference Companion

Greco

Minisci

et al. 2019

This paper presents a novel optimisation approach, called Structured-Chromosome Genetic Algorithm (SCGA), that addresses the issue of handling variable-size design space optimisation problems. This is based on variants of standard genetic operators able to handle structured search spaces. The potential of the presented methodology is shown by solving the problem of defining observation campaigns for tracking space objects from a network of tracking stations. The presented approach aims at supporting the space sector in response to the constantly increasing population size in the around-Earth environment. The test case consists in finding the observation scheduling that minimises the uncertainty in the final state estimation of a very low Earth satellite operating in a highly perturbed dynamical environment. This is evaluated by coupling the optimiser with an estimation routine based on a sequential filtering approach that estimates the satellite state distribution conditional on received indirect measurements. The solutions found by employing SCGA are finally compared to the ones achieved using more traditional approaches. Namely, the problem has been reformulated to be faced using standard Genetic Algorithm and another variable-size optimiser, the "Hidden-genes" Genetic Algorithm variant.

“…This problem is even more critical when variables of a different type, like continuous (numerical) and categorical (nominal), are used to simultaneously encoding topological and dimensional features. Among the different algorithms, one of the most suited to face this particular kind of problems are Genetic Algorithms (GAs) [8].…”

Section: Introductionmentioning

confidence: 99%

High-Lift Devices Topology Optimisation using Structured-Chromosome Genetic Algorithm

2020 IEEE Congress on Evolutionary Computation (CEC)

Morales

Quagliarella

et al. 2020

This paper addresses the problem of including the choice of the High-Lift Devices (HLDs) configuration as a decision variable of an automatic optimisation tool. This task requires the coupling of an estimation routine and an optimisation algorithm. For the former, SU2 flow solver has been used. The Structured-Chromosome Genetic Algorithm (SCGA) optimiser has been employed to search for the optimal HLD. SCGA can overcome the limitations dictated by standard fixed-size continuous optimisation algorithms. Indeed, using hierarchical formulations, it can manage configurational decisions that are conventionally the responsibility of expert designers. The search algorithm bases its strategy on revised genetic operators conceived for handling hierarchical search spaces. The presented research not only shows the practicability of delegating to a specialised optimisation algorithm the complete HLD design but is intended to be a proof of concept for the whole field of multidisciplinary design optimisation. Indeed, the aerospace sector as a whole would benefit by reducing human intervention from the decision process.