Uncertainty-based multidisciplinary design optimization of lunar CubeSat missions

Transcription

1 4th Interplanetary CubeSat Workshop Uncertainty-based multidisciplinary design optimization of lunar CubeSat missions Xingzhi Hu 3 rd year PhD xh269@cam.ac.uk, huxingzhi@nudt.edu.cn Supervisor: Prof. Geoffrey T. Parks Prof. Xiaoqian Chen

2 Outline Introduction Problem Formulation Optimization Methodology Results & Discussions Conclusions 2

3 Introduction Cost-effective shift Large-scale to small-scale Successful heritage, easy launch, much lower cost, faster development Dual INSPIRE CubeSats by NASA 3U, 2U, 1U cubic cells 3U CubeSat by PocketSpacecraft 3

4 Introduction Goals of our research Multidisciplinary design optimization (MDO) under uncertainty Uncertainty quantification and probabilistic optimization for lunar CubeSats Cost risk minimization with regard to robustness and reliability Impact of event Performance loss Catastrophe No engineering applications Robust design optimization (RDO) Reliability-based design optimization (RBDO) Reliability is not an issue Small perturbation Extreme event Uncertain event 4

5 Problem Formulation Uncertainty f ( f s,) Surrogate model Design variables 5

6 Problem Formulation Lunar trajectory and orbit design Edelbaum s analysis ΔV = V 0 cosβ 0 V 0 sinβ 0 tan( π 2 Δi + β 0 β(t) = tan 1 V 0 sinβ 0 ( V 0 cosβ 0 a T t V(t) = V 0 2 2V 0 a T tcosβ 0 + a T 2 t 2 2 a t V cos i 1 T 0 0 (t) [tan ( ) 0] V0sin 0 2 m p = m total (1 e Δ V gi sp Simulated spiral trajectory 6

7 Problem Formulation Subsystem definition Propulsion: Miniature Xenon Ion (MiXI) Engine Thrust (mn) Mass (kg) Isp (s) Power (W) Efficiency (%) MiXI ~50 Key payload: CCD camera Source: hsb it Dr xvn d 2 s q d 0 h / f s S R h R R i 1 w 2 e sin [sin x( e) / e] x / sin c Thermal: passive control Q in Q 0 out T sc 4 4 I [ cos( ) ] int ( ), ) s As Fs s AAAFA Q sp Asp RAR FspT sp IR sc IR AIR FIRT IR ( sp Asp R AR ) Fsp IR, scir AIR FIR 7

8 Problem Formulation Uncertainty definition Discipline Parameter Notation Distribution Orbit Orbit altitude (km) h Truncated Nor. Lunar orbit inclination (deg) i Normal Payload CCD Focus length (mm) f c Normal Multiple Mission cycle (year) T life Normal Efficiency η t Normal Propulsion Input power (kw) P t Normal Isp (s) I sp Normal Solar energy-conversion efficiency η a Normal Power Solar array energy-mass density (W h/kg) γ a Normal Average discharge depth DOD Normal Multiple System mass margin ε m Interval System power margin ε p Interval 8

9 Problem Formulation Reliability-based robust design optimization (RBRDO) find X [ h i f ] c Tlife min,, ( ) f d f d f d f d g1 : Pr{ ds 30m} 0.99 s.t. g2 : Pr{ F 1} 0.99 str g3 : Pr{ Vsat 3 U} 0.99 g4 : Pr{ NBa 10, 000} km h 600km, 80 i 90 20mm f 200mm, 2a 6a c Tlife 2 2 C M C C 365LT fc = Csat / ( DrTw ) M c f M = M sub, i 9

10 Optimization Methodology Optimization workflow Begin System modeling Uncertainty modeling System decoupling Optimization under uncertainty Uncertainty quantification No Convergence Yes Robust and reliable optimal solution Disciplinary analysis End 10

11 Optimization Methodology In-loop uncertainty quantification Previous methods: Monte Carlo; polynomial chaos, compressed sensing Our method: identify a one-dimensional active subspace Applied to 3U lunar CubeSat 11

12 Optimization Methodology Multi-objective solver Previous algorithms: MOGA, NSGA-II, MOPS.. Our method: Multi-objective alliance algorithm (MOAA) Begin Parameter initialization Initial samples Pareto-optimal (PO) solutions Yes Convergence No Solution generation End Formation of alliances Creation of new tribes 12

13 Results & Discussions RBRDO solutions 2881 of 4001 Pareto front 13

14 Results & Discussions Pareto comparison MOAA vs. NSGA-II 14

15 Results & Discussions Deterministic vs. Nondeterministic 15

16 Conclusions Take home messages 16

17 Conclusions MOAA and active subspaces work. RBRDO worthwhile for lunar CubeSats. Reference for conceptual design and parameter control. Further perfected and demonstrated in near missions. 17

18 CubeSat is revolutionizing aerospace science and engineering.