Game theory and evolutionary algorithms applied to MDO in the AGILE European project

Abstract

In this paper, an optimization technique in aircraft design field, based on game theory and evolutionary algorithms to define the key variables for Multi-Disciplinary aircraft Optimization (MDO) into AGILE (Aircraft 3rd Generation MDO for Innovative Collaboration of Heterogeneous Teams of Experts) European project, is presented. This work represents one of the contributions given by UniNa (University of Naples “Federico II”) research group within the AGILE project, which is coordinated by the DLR and funded by EU through the project HORIZON 2020 that aims to create an evolution of MDO, promoting a novel approach based on collaborative remote design and knowledge dissemination among various teams of experts. Since the aircraft design field is very complex in terms of number of involved variables and the dimension of the space of variation, it is not feasible to perform an optimization process on all the design parameters; this leads to the need to reduce the number of the parameters to the most significant ones. A multi-objective optimization approach allows many different variables, which could be a constraint or an objective function for the specific investigation; thus, setting the constraints and objectives to reach, it is possible to perform an optimization process and control which parameters significantly affect the final result. Within AGILE project, UniNa research group aims to perform wing optimization processes in a preliminary design stage, coupling Nash game theory (N) with typical genetic evolutionary algorithm (GA), reducing computational time and allowing a more realistic association among objective functions and variables, to identify the main ones that significantly affect final result and that consequently must be considered by the partners of the AGILE consortium to perform MDO in the final part of project, applying the proposed optimization technique to novel aircraft configuration

Enhancing optimization capabilities using the AGILE collaborative MDO framework with application to wing and nacelle design

“…Use case focused on the improvement of MDO strategies with the development and integration of new design competences in terms of optimization algorithms and surrogates modelling. This use case and the results are presented in [11,12]. 2.…”

“…The approach developed by the UniNa research group couples the Nash game theory (N) to a typical genetic evolutionary algorithm GA, reducing computational time and allowing a more realistic association among variables and objective functions [47]. A detailed description of NGA optimization used in the AGILE project is given in [12]. In the game theory approach, a multi-objective problem is considered as a game with n players, each one characterized by a pay-off.…”

“…The idea to apply NGA equilibrium solutions to the aircraft design and optimization leads to the chance to avoid a more arbitrary and less physically based variables association among the different objective functions, using, instead, a more engineering reliable variables assignment based on well-known parameter association [47,12]. In the NGA optimization approach, variables cards can be assigned to players (objective functions) in a unique case, assigning in a static manner these variables, or in multiples combinations, choosing cases to be optimized (until to the maximum number of possible combinations).…”

A clustered and surrogate-based MDA use case for MDO scenarios in AGILE project

“…1) Use case focused on the improvement of MDO strategies with the development and integration of new design competences in terms of optimization algorithms and surrogates modelling. These investigations are presented in [10,11].…”

A Simulation-Based Performance Analysis Tool for Aircraft Design Workflows

“…JPAD is an application programming interface (API) able to support civil transport aircraft designers in the need for building multi-disciplinary, simulation-based analysis and optimization workflows. The API is a Java software library containing classes and utility functions that can be used to build software systems running MDAO procedures [5,[28][29][30][31][32][33]. JPAD comes in different interdependent software modules providing modeling, simulation, and analysis features.…”