Méthodes et outils pour la conception optimale des réseaux de distribution d'électricité dans les aéronefs

摘要

In the aeronautics field, the last decade has been marked by a constant and gradual increase of the electrification rate of the embedded systems. Today, the More Electric Aircraft (MEA) is seen as a major axis of improvement for the aviation industry to achieve increasingly ambitious objectives: reducing environmental impact, rationalisation of maintenance costs...In the more electrical aircraft concept, the electrical network plays a major role. Today engineers must imagine new architectural solutions to ensure the electricity supply while minimizing weight and cost. In this context, the PhD work consists in providing new methods to support the design of electrical network architectures. The PhD work is divided into 2 parts which can be seen as 2 separate studies which are introduced in the chapter 1.The 1st part, treated in the chapters 2 and 3, develops methods and tools to solve problems automatically for 2 architecture tasks: the definition of the network reconfiguration and the identification of the electrical load allocation on busbars. The formalization of these two issues highlights a common characteristic: the combinatorial explosion. As the consequence, methods from operational research area are selected to solve the 2 tasks in the frame of a general and consistent design process. The reconfiguration aspects are solved by a methodology coupling together: graph theory to model the network connectivity, an expert system capturing know-how rules and linear programming selecting the most efficient reconfiguration. The approach was successfully applied on existing aircraft electrical networks (A400M and A350) and on future architectures. The second task, related to the electrical load allocation, is solved using stochastic methods. The genetic algorithm using a niching method is the best assessed optimization method. It provides good and diversified load allocations to the electrical network architect. The 2nd part, treated in the chapter 4, focuses on a new technological concept the « modular and mutualised power electronics center ». This distribution system, closely linked to the more electrical aircraft, aims at sharing « m » power electronics modules to « c » electrical loads. The methods developed in this PhD aim at carrying out an optimal design of this new power center with 2 design variables: the number « m » of modules and the reconfigurations between the « m » modules and the « c » loads. Again, the formalization of the problem highlights that the designer must deal with a combinatorial explosion. The main objective of this study is to propose a methodological framework for solving this design problem. A heuristic-based algorithm is developed to solve this combinatorial optimization problem. A particular attention is paid to develop an organized weight estimation procedure using generic sizing models. Finally a mapping is performed to identify the best solutions and to highlight the technological elements having the most significant impact on the complete system weight