Integrated Design Of Aircraft Electrical Power Supply System

Conventional aircraft design is a serial process from the power consumption,the power distribution,to the power generation.In this process,the designs of every subsystem are independent on each other’s,which cannot ensure the whole performance of aircraft,and even worse,lead to bad endings.Therefore,multiple iterations are required to get an acceptable design,which surely result in high cost and long period.Moreover,the electrification of aircraft,which has become the future direction of aviation technology,leads to integrated and intelligent aircraft systems.In other words,the electrification makes the systems in the aircraft more and more tightly coupled,which cannot be dealt well with by the original design method.Therefore,this dissertation is dedicated to the integrated design of the aircraft electrical power system with model and optimization-based methods to solve the problem that the traditional design approach will surely leads to massive manual iterative calculations.The dissertation seeks to propose a general design framework,build design models,solve detailed design problems,and finally realize the software implement for the design of the electrified aircraft power system.The researches and main contributions are summarized as follows.1)We first clarify the category of the design of electrified aircraft power systems,based on the summary of the key technologies of future electrified aircrafts.The design work can be identified as a multi-layer design problem including the plant design(topology,technology,and size)and the control design(power scheduling and parameter control).Therein,the plant design is the main work of our dissertation.Every layer is analyzed to demonstrate the characteristics and difficulties in the design process of electrified aircrafts.Solutions are also drawn forward to offer a guideline for the readers.2)To realize the model-based design,we propose a modular and multi-layer modeling method based on the multi-disciplinary design optimization(MDO)to reduce the computational cost and design space of the design problem.To be detailed,the dissertation improves the existing multi-physics modeling method by the modular modeling approach to describe the strong couplings between component parameters of the modeling object.To reduce the dimensionality of the model,a data mining-based model simplification method is developed to optimally select key features of the multi-physics model.Machine learning algorithm-based surrogate modeling methods are also applied to reduce the model computational cost.Comparisons with prototypes and industrial products of electric machines demonstrate that the simplified models are of accuracy up to 96.5% compared with the multi-physics model,and up to 90%compared with real performance.At the same time,the proposed models can accelerate the optimization speed by 50%,promote the convergence speed by 26%,and have better diversity performance,compared with the other models.In other words,the proposed modeling method can achieve both high precision and effective search.3)To help integrate the power supply system,we propose a function-based network topology design approach.The dissertation firstly generates the topology based on the platform-based design method;then decompose the design problem into the constraint satisfaction problem(CSP)and multi-objective problem.To solve the CSP,we build constraint models according to the experience,standards,and etc.,and integrate the traversal of single and multiple failures of the system into the model,obtaining several feasible topology solutions by constraint solvers;after that,solutions of the highest reliability are optimally selected from the feasible ones by the traversing method.A case of a system with 3 sources and 5-8 buses demonstrates the proposed method and shows that the method can shorten the design time dramatically and avoid information lapses resulting from the traditional enumeration method.4)To determine the connection forms of the system topology,we propose a quantitative analysis and optimization method of the voltage level in the future aircraft power supply system based on the built models.According to the current electrical and electronic technologies,the dissertation analyzes the optimal voltage levels of the generator,the fuel cell system,the PWM rectifier,the power cable,the solid switch,and the constituted system with high power.Compared with current studies referred to the terrestrial grids,the proposed method is of a more general theoretical significance,and can be used to evaluate unknown voltage levels of future aircrafts.5)After the determination of the topology,we study the energy allocation of electrical power system with energy storage based on the generation,reduction,and synthesis of the large-number scenarios.The dissertation proposes the network reconfiguration-based scenario generation method according to the phases and possible failures in the flight process,thus different power demands of working conditions can be taken into consideration.To facilitate the effective design,a branch-extreme-condition-based scenario reduction method is developed to reduce the scenarios for analysis.On the other hand,to synthesize the design results of power scheduling of different scenarios,a new scenario fusion method based on scenario domination is proposed to reduce the system weight and arise the capacity utilization of the system.According to analysis of a case of a distributed power system with 3 sources,5 buses,and5 energy storages,the method is proved to be able to reduce the scenarios by 95% and cover over 98%of the optimal solutions of all scenarios.6)Finally,we preliminarily study the integrated design of the electrical power system,set up an overall plan and design for the system design platform,and verify the system design results by simple experiments.The dissertation proposes three distributed optimization frameworks for the integrated design;therein,the three-layer serial structure is proved to be of the best performance via theoretical analyses and simulations.To implement the integrated design,a simple graphical user interface is developed based on Lab VIEW and MATLAB/Plat EMO,which collects the design demands,selects intelligent algorithms to solve problems,and illustrates the design results.Moreover,to demonstrate the feasibility of the researches,we explore the design space of the parameter control layer of a single-bus power supply system consisting of fuel cells,lithium batteries and supercapacitors to analyze the influences of other-layer design results on the parameter control layer.Additionally,the reconfiguration strategy used in the scenario generation is also verified by the comparisons between the results of the traversing method and those of semi-physical simulations.