Introduction to System Partitioning and Coordination
|
|
- Corey Harmon
- 6 years ago
- Views:
Transcription
1 Introduction to System Partitioning and Coordination James T. Allison ME 555 April 4, 2005 University of Michigan Department of Mechanical Engineering
2 Your Subsystem Optimization Problems are Complete Now What? Methods for coordinating subsystems: Sequential Optimization MDF (a.k.a. All-in-One) IDF AAO ATC
3 Can Subsystem Design Problems be Solved Independently? Not if interaction is present. An interaction is present when the response of one element is not only dependent on its own state, but also on the state of other members in the system. Result: The whole is different than the sum of the parts. Case I: Members collaborate to synergistically improve system performance, even at the expense of individual member objectives. (Teamwork) Case II: Members considering only their own performance may experience interactions that collude to degrade system performance.
4 Interaction, Coupling, and Coupling Strength Coupling Interaction Coupling indicates the existence of interaction between system elements (subspaces). Coupling Strength Interaction Strength If a subspace response is highly dependent (sensitive) on the state of another subspace, the coupling between the subspace is strong. If a subspace response is weakly dependent (insensitive) on the state of another subspace, the coupling between the subspace is weak. The coupling strength of an entire system depends on the number and strength of all couplings present.
5 Illustration of Interaction/Coupling Existence: DSM SS1 SS2 SS3 SS4
6 Physical Example: 2DOF Mass Spring System Governing Equations of Motion: [ m1 0 0 m 2 ] { ẍ1 ẍ 2 } + [ k1 + k 2 k 2 k 2 k 2 ] { x1 x 2 } = { 0 0 } Coupled in stiffness: Motion of m 1 (i.e. x 1 (t)) is dependent of the value of k 2 ( x 1 (t) depends on the state of x 2 ), and visa-versa. m 1 k 1 x 1 k 2 m 2 x 2
7 Sequential Optimization Sequential Approach: Optimization Commonly used process in industry. Deals with, but does not exploit, interactions. Design freedom is progressively reduced. Lack of parallelization and iterative optimization increases design cycle time. May not find a system optimal design, even after iteration. SS1 Optimizer SS2 Optimizer SS3 Optimizer SS1 Analysis SS2 Analysis SS3 Analysis Fix x l1, x s1 Fix x l2, x s2 Final Design
8 Sequential Optimization Example f(x 1, x 2 ) = a 1 (x 1 a 2 ) 2 + a 3 (x 2 a 4 ) 2 + a 5 x 1 x 2 Case I: No Interaction a = (1, 1, 1, 1, 0) T 6 5 (x 1-1) 2 +(x 2-1) 2 4 x x 1
9 Sequential Optimization Example f(x 1, x 2 ) = a 1 (x 1 a 2 ) 2 + a 3 (x 2 a 4 ) 2 + a 5 x 1 x 2 Case II: Interaction Present a = (1, 1, 1, 1, 1) T 6 5 (x 1-1) 2 +(x 2-1) 2 -x 1 x 2 4 x 2 3 OFAT solution 2 optimum x 1
10 What are Some Alternatives to Sequential Optimization? MDF Multidisciplinary Feasible approach a complete analysis is performed at every optimization iteration. Also known as the All-in-One approach. IDF Individual Disciplinary Feasible approach system analysis is performed simultaneously with system optimization. AAO All-at-Once approach system analysis, optimization, and determination of state variables performed simultaneously. Useful for systems involving differential equations. ATC Analytical Target Cascading each system element has its own optimizer. Parent design problems coordinate the design of child design problems. Useful for systems with a hierarchical structure.
11 x l Local design variables: Design variables that are each inputs to only one subspace. x s Shared design variables: Design variables that are each inputs to more than one subspace. y Coupling variables: Quantities passed between subspaces that are not original design variables, but rather artifacts of decomposition. MDF/IDF/AAO Terminology { s 1 x= x, x,, x x x x 1 = 1, s1 SS1 N f1, g1, h1 y 21 y 12 y j1 y 1j } System Analysis x x x 2 = 2, s2 SS2 f2, g2, h ( 1,, N ) = { 1,, N} = { 1,, N} f f f g g g h h h x = x, x N N sn SSN y Nj y jn fn, gn, hn
12 MDF Formulation and Architecture min x={x l,xs} f(x) subject to g(x) = {g 1, g 2,...} 0 h(x) = {h 1, h 2,...} = 0 { s, 1, 2} x= x x x System Analyzer ( 1, 2) = { g1, g2} = h, h } f f f g h { 1 2 Optimizer x 1, xs f, g, h x 2, xs f2, g2, h2 x f, g, h Subspace Analyzer Subspace Analyzer Analyzer s y 12 s s SS1 SS2 y 21
13 Example Problem Single Element Aeroelasticity Aeroelasticity: Requires both aerodynamic and structural analysis Application: Air-flow sensor design Implementation: Will be used to demonstrate MDF, IDF, and ATC.
14 Air-flow Sensor Analysis Structural Analysis: M = kθ = 1 2 F l cos θ Given a design (l, w) and a drag force F, solve for the corresponding deflection θ. Aerodynamic Analysis: F = CA f v 2 = Clw cos θv 2 Given a design (l, w) and a deflection θ, find the drag force F. System Analysis: Given a design (l, w), find the equilibrium values F and θ. v F 1/2 cos k cos
15 Air-flow Sensor Design Sensor Calibration: Find the sensor design that achieves a deflection θ for a given air speed v that is as close as possible to a target deflection value θ T, while ensuring the drag force remains below an upper limit, and the sensor surface area remains a constant. min l,w (θ θ T ) 2 subject to F F max 0 lw A = 0 Design Parameters: k, A, F max, C, v
16 Solution by Monotonicity For cases where meeting the deflection target requires a drag force greater than F max, the target cannot be met, and the inequality constraint will be active. This, along with the equality constraint lw = A, facilitates analytical solution for l and w. F max = Clw cos θv 2 kθ 1 2 F maxl cos θ = 0 l = ( ) θ = cos 1 Fmax CAv 2 ( ) 2k cos 1 Fmax CAv 2 CAv 2 F 2 max Now that F, θ and l are known, we can solve for w using w = A/l. This solution technique provides a check for other approaches.
17 Partitioning by Discipline What are the shared design variable(s)? What are the local design variable(s)? What are the coupling variable(s)? How might this partitioned analysis be executed? (i.e., find F and θ for a given l and w. l l,w Structural Analysis Aerodynamic Analysis F
18 Matlab Demonstration Fixed Point Iteration Analysis
19 Analysis Discussion Would increasing k increase or decrease the number of required iterations? Would increasing w increase or decrease the number of required iterations?
20 MDF Design Solution MDF: Complete analysis for every optimization iteration. Monotonicity Solution: x = [l, w ] = [0.3651, ] Matlab Demonstration MDF
21 MDF Discussion Advantages: Intuitive procedure/no specialized knowledge required Easy to incorporate existing models Disadvantages: Proven that FPI will not converge in many cases If multiple analysis solutions exist, MDF may not choose the best one Cannot be parallelized Decision making is centralized
22 IDF Formulation and Architecture min x={x l,xs},y f(x, y) subject to g(x, y) = {g 1, g 2,...} 0 h(x, y) = {h 1, h 2,...} = 0 f 1, g 1, h 1, y i1 Optimizer f 2, g 2, h 2, y i2 h aux (x, y) = y(x, y) y = 0 x 1, x s1, y 1j x 2, x s2, y 2j Analyzer Analyzer s s
23 IDF Formulation of Air Flow Sensor Problem min l,w,θ,f (θ θ T ) 2 subject to F F max 0 lw A = 0 θ θ(l, F ) = 0 F F (l, w, θ) = 0 Structural Analysis System Optimizer l, F l, w, F Aero Analysis
24 IDF Design Solution IDF: Analysis and Design performed simultaneously Monotonicity Solution: x = [l, w ] = [0.3651, ] Matlab Demonstration IDF
25 IDF Discussion Advantages: Exploits the power of optimization algorithms to drive analysis (another example is regression) optimizer controls both x and y All subspaces executed in parallel Analysis and Design performed simultaneously reduces computation time In many cases will converge when MDF will not, and may find better solutions than MDF Faster convergence for tightly coupled systems Disadvantages: Increased dimension of the optimization problem over MDF Increased centralization of decision making Must meet hard equality constraints
26 AAO Formulation and Architecture min x={x l,xs},y,s f(x, y) subject to g(x, y, s) = {g 1, g 2,...} 0 h(x, y, s) = {h 1, h 2,...} = 0. { } y(x, y, s) y h aux (x, y, s) = w(x, y, s) Optimizer f 1, g 1, h 1, y i1, w 1 f 2, g 2, h 2, y i2, w 2 x 1, x s1, y 1j, s 1 x 2, x s2, y 2j, s 2 = 0. Evaluator Evaluator
27 Advantages: AAO Discussion (No Demo) Exploits the power of optimization algorithms to drive analysis and DEQ solution optimizer controls x, y and s Convenient for systems of PDE s All subspaces executed in parallel Highly efficient process Excellent convergence properties Faster convergence for tightly coupled systems Disadvantages: Increased dimension of the optimization problem over IDF Highly centralized decision making increased communication requirements Must meet hard equality constraints Difficult to integrate simulations
28 February 23, :42 Proceedings of DETC ASME Design Engineering Technical Conferences Long Beach, California, USA, September 24-28, 2005 DRAFT DETC More details on MDF, IDF, and AAO are available in this upcoming DETC publication, which is now available on the course website. ON THE IMPACT OF COUPLING STRENGTH ON COMPLEX SYSTEM OPTIMIZATION FOR SINGLE-LEVEL FORMULATIONS James Allison, Michael Kokkolaras, and Panos Papalambros Optimal Design Laboratory Department of Mechanical Engineering University of Michigan, Ann Arbor, Michigan ABSTRACT Design of modern engineering products requires complexity management. Several methodologies for complex system optimization have been developed in response. Single-level strategies centralize decision-making authority, while multi-level strategies distribute the decision-making process. This article studies the impact of coupling strength on single-level Multidisciplinary Design Optimization formulations, particularly the Multidisciplinary Feasible (MDF) and Individual Disciplinary Feasible (IDF) formulations. The Fixed Point Iteration solution strategy is used to motivate the analysis. A new example problem with variable coupling strength is introduced, involving the design of a turbine blade and a fully analytic mathematical model. The example facilitates a clear illustration of MDF and IDF and provides an insightful comparison between these two formulations. Specifically, it is shown that MDF is sensitive to variations in coupling strength, while IDF is not. KEYWORDS Complex System Design, Multidisciplinary Design Optimization, Coupling Strength, Multidisciplinary Feasible, Individual Disciplinary Feasible, Fixed Point Iteration 1 INTRODUCTION This article endeavors to illustrate the implementation of single-level formulations for complex system optimization via an analytical example. Specifically, the effect that subsystem interdependence (coupling) has on the performance of these formulations is explored. Background in complex system optimization is provided, and the mathematical model for the illustrative example is developed and presented. This work aims at improving general understanding of techniques for complex system optimization, and it is part of a greater effort to review single and multi-level MDO formulations [1, 2]. Design of products classified as complex systems poses substantive challenges to both analysis and optimization, necessitating specialized solution techniques. A complex system is defined as an assembly of interacting members that is difficult to understand as a whole. An interaction between members exists if some aspect of one member affects how the system responds to changes in another member. A system is difficult to understand if an individual cannot understand the details of all members and all interactions between members. A system may qualify as complex due to its large scale (large number of members or inputs), or due to strong interactions. These interactions complicate optimization, but provide opportunity to exploit synergy between system members. Analysis of complex systems as an undivided whole can be inefficient, if not intractable. An alternative is to partition the system into smaller subsystems (or subspaces). Wagner [3] identified four categories of system partitioning methods: by object, by aspect, sequential, or matrix. The aspect (discipline) partitioning paradigm is used in this article, and the term subspace is used to refer to system members. System partitioning is also characterized by the structure of its communication pathways. A non-hierarchic system has no restrictions on these pathways 1 Copyright c 2005 by ASME
29 Introduction to Analytical Target Cascading ATC: Target setting methodology intended for product development. Based on a hierarchical problem structure. Typically partitioned by physical object (subsystems/components). Early product development tool. Can be used for either target setting or for design optimization. Originally intended for problems with unidirectional information flow, but can be used where feedback coupling is present. Multilevel optimization methodology each subspace has its own optimizer (rather than a single centralized optimizer).
30 Analysis Structure Amenable to ATC System-level design variables Subsystem Responses: Subsystem performance Geometry System Analysis: Vehicle System Integration Packaging Control Subsystem-level design variables Subsystem Analysis: Powertrain integration and simulation Component Responses: Engine map Geometry Component-level design variables Component Analysis: Engine Simulation......
31 Using ATC for Product Development 1. Specify top-level targets. 2. Propagate targets through the system. (a) Develop low-fidelity analysis models. (b) Partition the system. (c) Formulate the target cascading problem. (d) Solve the target cascading problem with a coordination strategy. 3. Perform subspace detail design to meet targets. 4. Verify system consistency and system level target matching.
32 How is the Target Cascading Problem Formulated and Solved? 1. Choose a partitioning structure 2. Identify the input/output relations 3. Identify shared values 4. Formulate the optimization problem for each system element (a) Minimize local objective function (b) Create penalties for shared values that do not match (c) Subject to local design constraints 5. Choose a strategy for executing element optimization problems (explained later) 6. Run the process until system inconsistency meets convergence criteria
33 ATC Formulation Air Flow Sensor Problem What values are shared between subspaces? l Structural Analysis F l,w Aerodynamic Analysis
34 ATC Formulation Air Flow Sensor Problem What values are shared between subspaces? l, θ, and F l Structural Analysis F l,w Aerodynamic Analysis
35 ATC Formulation Air Flow Sensor Problem Structural Problem (arbitrarily chosen as master problem) min l,f (f 1 f ) 2 + (l 1 l 2 ) 2 + (F 1 F 2 ) 2 + (θ 1 θ 2 ) 2 subject to F F max 0 Aerodynamic Problem (arbitrarily chosen as subproblem) min l,w,θ (l 1 l 2 ) 2 + (F 1 F 2 ) 2 + (θ 1 θ 2 ) 2 subject to lw A = 0 where f = (θ θ T ) 2, and f is the known optimal value of f.
36 ATC Implementation 1. Begin with a guess for subproblem responses to solve the master problem. 2. Use the master problem responses to solve the subproblem. 3. Iterate until system consistency stops changing.
37 ATC Implementation 1. Begin with a guess for subproblem responses to solve the master problem. 2. Use the master problem responses to solve the subproblem. 3. Iterate until system consistency stops changing. Note: Other coordination strategies are possible when more than two levels exist. Penalty function weights may be required to improve ATC convergence. In theory, if all targets are attainable, ATC will converge to an exactly consistent system. If targets are unattainable, some type of weighting update method is required.
38 Matlab Demonstration ATC
39 Additional ATC Example Problem Illustrative example with slightly increased complexity to demonstrate additional aspects of ATC. Statically Indeterminate Structural Analysis (SISA) of an anchoring system. Can be expanded to an arbitrarily large number of subspaces.
40 L Ø d nb beam n b rod n b -1 Ø d r(nb-1) l r(nb-1) Ø d 3 beam 3 Ø d r2 rod 2 l r2 Ø d 2 beam 2 Ø d r1 Ø d 1 rod 1 l r1 beam 1 F 1
41 Anchor Design Problem min x={x l,xs} f(x) = δ 1 (x) subject to g 1i (x) = σ bi (x) σ allow 0 i = 1... n b g 2j (x) = σ aj (x) σ allow 0 j = 1... n r g 3 (x) = n b i=1 m bi + nr i=1 m rj m allow 0 g 4i (x) = F t i F t allow 0 i = 1... n b x = {d 1 d 2 d 3 d r1 d r2 } T
42 Analytical Target Cascading Example Analysis structure manipulated into a sequential form with only one feedback. Analysis becomes a root-finding problem: ˆF1 (F 3 ) F 1 = 0 Present two-level partitioning scheme. F 3 3 a d r2 f 2 b F 2 e r1 g ˆ 1 F 1 c
43 F 1 Subspace 1 d 1, d r1 F Subspace 2 Subspace 3 d 2 d 3, d r2 F 3 F 3 subspace 3 subspace 2 subspace 1 F 3 3 a d r2 f 2 b F 2 e r1 g ˆ F 1 1 c
44 min x 11 =[d 1,d r1,f 2,F 3,δ 2,m 2,m 3 ] T (δ 1(11) 1 0)2 + (F 1(11) 1 F 1(11) 0 )2 + (F 2(22) 1 F 2(22) 2 )2 SS1-Top +(δ 1 2(23) δ2 2(23) )2 + (m 1 2(22) m2 2(22) )2 +(m 1 3(23) m2 3(23) )2 + (F 1 3(21) F 2 3(22) )2 + (F 1 3(21) F 2 3(23) )2 subject to g 11 (x 11 ) = σ b1 (x 11 ) σ allow 0 g 21 (x 11 ) = σ a1 (x 11 ) σ allow 0 g 3 (x 11 ) = m 1 (x 11 ) + m 2 + m 3 m allow 0 g 41 (x 11 ) = F t 1 (x 11 ) F t allow 0 min x 22 =[d 2,F 3,δ 2 ] T (F 2(22) 1 F 2(22) 2 )2 + (m 1 2(22) m2 2(22) )2 SS2-Bottom +(F 1 3(21) F 2 3(22) )2 + (δ 1 2(23) δ2 2(23) )2 subject to g 12 (x 22 ) = σ b2 (x 22 ) σ allow 0 g 42 (x 22 ) = F t 2 (x 22 ) F t allow 0 min x 23 =[d 3,d r2,f 3 ] T (δ 2(23) 1 δ2 2(23) )2 + (m 1 3(23) m2 3(23) )2 + (F 3(21) 1 F 3(23) 2 )2 SS3-Bottom subject to g 13 (x 23 ) = σ b3 (x 23 ) σ allow 0 g 22 (x 23 ) = σ a2 (x 23 ) σ allow 0 g 43 (x 23 ) = F t 3 (x 23 ) F t allow 0
Lecture Notes: Introduction to IDF and ATC
Lecture Notes: Introduction to IDF and ATC James T. Allison April 5, 2006 This lecture is an introductory tutorial on the mechanics of implementing two methods for optimal system design: the individual
More informationIntroduction to System Optimization: Part 1
Introduction to System Optimization: Part 1 James Allison ME 555 March 7, 2007 System Optimization Subsystem optimization results optimal system? Objective: provide tools for developing system optimization
More informationOn Selecting Single-Level Formulations for Complex System Design Optimization
James T. Allison Ph.D. Candidate e-mail: optimize@umich.edu Michael Kokkolaras Associate Research Scientist Mem. ASME e-mail: mk@umich.edu Panos Y. Papalambros Professor Fellow ASME e-mail: pyp@umich.edu
More informationOn the Use of Analytical Target Cascading and Collaborative Optimization for Complex System Design
6 th World Congress on Structural and Multidisciplinary Optimization Rio de Janeiro, 30 May 03 June, 005 Brazil x On the Use of Analytical Target Cascading and Collaborative Optimization for Complex System
More informationMultidisciplinary design optimisation methods for automotive structures
Multidisciplinary design optimisation methods for automotive structures Rebecka Domeij Bäckryd, Ann-Britt Rydberg and Larsgunnar Nilsson The self-archived version of this journal article is available at
More informationMulti-level hierarchical MDO formulation with functional coupling satisfaction under uncertainty, application to sounding rocket design.
11 th World Congress on Structural and Multidisciplinary Optimisation 07 th -12 th, June 2015, Sydney Australia Multi-level hierarchical MDO formulation with functional coupling satisfaction under uncertainty,
More informationMultidisciplinary System Design Optimization (MSDO)
Multidisciplinary System Design Optimization (MSDO) Numerical Optimization II Lecture 8 Karen Willcox 1 Massachusetts Institute of Technology - Prof. de Weck and Prof. Willcox Today s Topics Sequential
More informationCE 191: Civil and Environmental Engineering Systems Analysis. LEC 05 : Optimality Conditions
CE 191: Civil and Environmental Engineering Systems Analysis LEC : Optimality Conditions Professor Scott Moura Civil & Environmental Engineering University of California, Berkeley Fall 214 Prof. Moura
More informationOptimal Multilevel System Design under Uncertainty
Optimal Multilevel System Design under Uncertainty 1 M. Kokkolaras (mk@umich.edu) Department of Mechanical Engineering, University of Michigan, Ann Arbor, Michigan Z.P. Mourelatos (mourelat@oakland.edu)
More informationCompositional Safety Analysis using Barrier Certificates
Compositional Safety Analysis using Barrier Certificates Department of Electronic Systems Aalborg University Denmark November 29, 2011 Objective: To verify that a continuous dynamical system is safe. Definition
More informationConstrained optimization: direct methods (cont.)
Constrained optimization: direct methods (cont.) Jussi Hakanen Post-doctoral researcher jussi.hakanen@jyu.fi Direct methods Also known as methods of feasible directions Idea in a point x h, generate a
More informationPareto Set Analysis: Local Measures of Objective Coupling in Multi-objective Design Optimization
8 th World Congress on Structural and Multidisciplinary Optimization June -5, 009, Lisbon, Portugal Pareto Set Analysis: Local Measures of Objective Coupling in Multi-objective Design Optimization Bart
More informationAdvanced Space Vehicle Design Taking into Account Multidisciplinary Couplings and Mixed Epistemic/Aleatory
Advanced Space Vehicle Design Taking into Account Multidisciplinary Couplings and Mixed Epistemic/Aleatory Uncertainties Mathieu Balesdent, Loïc Brevault, Nathaniel Price, Sébastien Defoort, Rodolphe Le
More informationA beam reduction method for wing aeroelastic design optimisation with detailed stress constraints
A beam reduction method for wing aeroelastic design optimisation with detailed stress constraints O. Stodieck, J. E. Cooper, S. A. Neild, M. H. Lowenberg University of Bristol N.L. Iorga Airbus Operations
More informationResearch Article An Enhanced Collaborative Optimization Approach with Design Structure Matrix Algorithms to Group and Decouple Multidisciplines
Mathematical Problems in Engineering Volume 2016, Article ID 4340916, 8 pages http://dx.doi.org/10.1155/2016/4340916 Research Article An Enhanced Collaborative Optimization Approach with Design Structure
More informationMulti-objective design and tolerance allocation for singleand
DOI 10.1007/s10845-011-0608-3 Multi-objective design and tolerance allocation for singleand multi-level systems Tzu-Chieh Hung Kuei-Yuan Chan Received: 25 April 2010 / Accepted: 6 March 2011 Springer Science+Business
More informationOptimal Partitioning and Coordination Decisions in System Design Using an Evolutionary Algorithm
7 th World Congress on Structural and Multidisciplinary Optimization COEX Seoul, 1 May - May 7, Korea Optimal Partitioning and Coordination Decisions in System Design Using an Evolutionary Algorithm James
More informationImproved System Identification for Aeroservoelastic Predictions
Master's Thesis Defense Improved System Identification for Aeroservoelastic Predictions Presented by Charles Robert O'Neill School of Mechanical and Aerospace Engineering Oklahoma State University Time
More informationLecture 24 November 27
EE 381V: Large Scale Optimization Fall 01 Lecture 4 November 7 Lecturer: Caramanis & Sanghavi Scribe: Jahshan Bhatti and Ken Pesyna 4.1 Mirror Descent Earlier, we motivated mirror descent as a way to improve
More informationAero-Propulsive-Elastic Modeling Using OpenVSP
Aero-Propulsive-Elastic Modeling Using OpenVSP August 8, 213 Kevin W. Reynolds Intelligent Systems Division, Code TI NASA Ames Research Center Our Introduction To OpenVSP Overview! Motivation and Background!
More informationTarget Cascading: A Design Process For Achieving Vehicle Targets
Target Cascading: A Design Process For Achieving Vehicle Targets Hyung Min Kim and D. Geoff Rideout The University of Michigan May 24, 2000 Overview Systems Engineering and Target Cascading Hierarchical
More information2.3 Linear Programming
2.3 Linear Programming Linear Programming (LP) is the term used to define a wide range of optimization problems in which the objective function is linear in the unknown variables and the constraints are
More informationCOMPLETE CONFIGURATION AERO-STRUCTURAL OPTIMIZATION USING A COUPLED SENSITIVITY ANALYSIS METHOD
COMPLETE CONFIGURATION AERO-STRUCTURAL OPTIMIZATION USING A COUPLED SENSITIVITY ANALYSIS METHOD Joaquim R. R. A. Martins Juan J. Alonso James J. Reuther Department of Aeronautics and Astronautics Stanford
More informationPenalty and Barrier Methods General classical constrained minimization problem minimize f(x) subject to g(x) 0 h(x) =0 Penalty methods are motivated by the desire to use unconstrained optimization techniques
More informationBilevel multiobjective optimization of vehicle layout
10 th World Congress on Structural and Multidisciplinary Optimization May 19-24, 2013, Orlando, Florida, USA Bilevel multiobjective optimization of vehicle layout Paolo Guarneri 1, Brian Dandurand 2, Georges
More informationCopyrighted Material. 1.1 Large-Scale Interconnected Dynamical Systems
Chapter One Introduction 1.1 Large-Scale Interconnected Dynamical Systems Modern complex dynamical systems 1 are highly interconnected and mutually interdependent, both physically and through a multitude
More informationA New Approach to Multidisciplinary Design Optimization via Internal Decomposition
13th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference 13 15 September, 2010, Fort Worth, Texas, United States A New Approach to Multidisciplinary Design Optimization via Internal Decomposition
More information5 Handling Constraints
5 Handling Constraints Engineering design optimization problems are very rarely unconstrained. Moreover, the constraints that appear in these problems are typically nonlinear. This motivates our interest
More informationCHAPTER 2: QUADRATIC PROGRAMMING
CHAPTER 2: QUADRATIC PROGRAMMING Overview Quadratic programming (QP) problems are characterized by objective functions that are quadratic in the design variables, and linear constraints. In this sense,
More informationHypersonic Vehicle (HSV) Modeling
Hypersonic Vehicle (HSV) Modeling Carlos E. S. Cesnik Associate Professor of Aerospace Engineering University of Michigan, Ann Arbor HSV Concentration MA Kickoff Meeting Ann Arbor, 29 August 2007 Team
More informationUncertainty quantifications of Pareto optima in multiobjective problems
DOI 0.007/s085-0-060-9 Uncertainty quantifications of Pareto optima in multiobjective problems Tzu-Chieh Hung Kuei-Yuan Chan Received: 5 May 0 / Accepted: 9 November 0 Springer Science+Business Media,
More informationAn Introduction to Algebraic Multigrid (AMG) Algorithms Derrick Cerwinsky and Craig C. Douglas 1/84
An Introduction to Algebraic Multigrid (AMG) Algorithms Derrick Cerwinsky and Craig C. Douglas 1/84 Introduction Almost all numerical methods for solving PDEs will at some point be reduced to solving A
More informationBECAS - an Open-Source Cross Section Analysis Tool
BECAS - an Open-Source Cross Section Analysis Tool José P. Blasques and Robert D. Bitsche Presented at DTU Wind Energy stand at the EWEA 2012 conference, Copenhagen, 16.4.2012 BECAS-DTUWind@dtu.dk Motivation
More informationCOMP 633: Parallel Computing Fall 2018 Written Assignment 1: Sample Solutions
COMP 633: Parallel Computing Fall 2018 Written Assignment 1: Sample Solutions September 12, 2018 I. The Work-Time W-T presentation of EREW sequence reduction Algorithm 2 in the PRAM handout has work complexity
More informationSimulating Two-Dimensional Stick-Slip Motion of a Rigid Body using a New Friction Model
Proceedings of the 2 nd World Congress on Mechanical, Chemical, and Material Engineering (MCM'16) Budapest, Hungary August 22 23, 2016 Paper No. ICMIE 116 DOI: 10.11159/icmie16.116 Simulating Two-Dimensional
More informationImprovements to Benders' decomposition: systematic classification and performance comparison in a Transmission Expansion Planning problem
Improvements to Benders' decomposition: systematic classification and performance comparison in a Transmission Expansion Planning problem Sara Lumbreras & Andrés Ramos July 2013 Agenda Motivation improvement
More informationOptimization Problems with Constraints - introduction to theory, numerical Methods and applications
Optimization Problems with Constraints - introduction to theory, numerical Methods and applications Dr. Abebe Geletu Ilmenau University of Technology Department of Simulation and Optimal Processes (SOP)
More informationConstrained Optimization
1 / 22 Constrained Optimization ME598/494 Lecture Max Yi Ren Department of Mechanical Engineering, Arizona State University March 30, 2015 2 / 22 1. Equality constraints only 1.1 Reduced gradient 1.2 Lagrange
More informationComponents for Accurate Forecasting & Continuous Forecast Improvement
Components for Accurate Forecasting & Continuous Forecast Improvement An ISIS Solutions White Paper November 2009 Page 1 Achieving forecast accuracy for business applications one year in advance requires
More informationExtreme Point Solutions for Infinite Network Flow Problems
Extreme Point Solutions for Infinite Network Flow Problems H. Edwin Romeijn Dushyant Sharma Robert L. Smith January 3, 004 Abstract We study capacitated network flow problems with supplies and demands
More informationAlgorithms for Constrained Optimization
1 / 42 Algorithms for Constrained Optimization ME598/494 Lecture Max Yi Ren Department of Mechanical Engineering, Arizona State University April 19, 2015 2 / 42 Outline 1. Convergence 2. Sequential quadratic
More informationValidating Expensive Simulations with Expensive Experiments: A Bayesian Approach
Validating Expensive Simulations with Expensive Experiments: A Bayesian Approach Dr. Arun Subramaniyan GE Global Research Center Niskayuna, NY 2012 ASME V& V Symposium Team: GE GRC: Liping Wang, Natarajan
More informationAnalysis and Design of an Electric Vehicle using Matlab and Simulink
Analysis and Design of an Electric Vehicle using Matlab and Simulink Advanced Support Group January 22, 29 23-27: University of Michigan Research: Optimal System Partitioning and Coordination Original
More informationOptimization and Complexity
Optimization and Complexity Decision Systems Group Brigham and Women s Hospital, Harvard Medical School Harvard-MIT Division of Health Sciences and Technology Aim Give you an intuition of what is meant
More information822. Non-iterative mode shape expansion for threedimensional structures based on coordinate decomposition
822. Non-iterative mode shape expansion for threedimensional structures based on coordinate decomposition Fushun Liu, Zhengshou Chen 2, Wei Li 3 Department of Ocean Engineering, Ocean University of China,
More informationLecture 3: Lagrangian duality and algorithms for the Lagrangian dual problem
Lecture 3: Lagrangian duality and algorithms for the Lagrangian dual problem Michael Patriksson 0-0 The Relaxation Theorem 1 Problem: find f := infimum f(x), x subject to x S, (1a) (1b) where f : R n R
More informationMethods for solving recurrences
Methods for solving recurrences Analyzing the complexity of mergesort The merge function Consider the following implementation: 1 int merge ( int v1, int n1, int v, int n ) { 3 int r = malloc ( ( n1+n
More informationConstrained dynamics
Constrained dynamics Simple particle system In principle, you can make just about anything out of spring systems In practice, you can make just about anything as long as it s jello Hard constraints Constraint
More informationSupport Vector Machines, Kernel SVM
Support Vector Machines, Kernel SVM Professor Ameet Talwalkar Professor Ameet Talwalkar CS260 Machine Learning Algorithms February 27, 2017 1 / 40 Outline 1 Administration 2 Review of last lecture 3 SVM
More informationMath 671: Tensor Train decomposition methods II
Math 671: Tensor Train decomposition methods II Eduardo Corona 1 1 University of Michigan at Ann Arbor December 13, 2016 Table of Contents 1 What we ve talked about so far: 2 The Tensor Train decomposition
More informationA Robust Controller for Scalar Autonomous Optimal Control Problems
A Robust Controller for Scalar Autonomous Optimal Control Problems S. H. Lam 1 Department of Mechanical and Aerospace Engineering Princeton University, Princeton, NJ 08544 lam@princeton.edu Abstract Is
More informationMulti-Point Constraints
Multi-Point Constraints Multi-Point Constraints Multi-Point Constraints Single point constraint examples Multi-Point constraint examples linear, homogeneous linear, non-homogeneous linear, homogeneous
More informationComputational Optimization. Augmented Lagrangian NW 17.3
Computational Optimization Augmented Lagrangian NW 17.3 Upcoming Schedule No class April 18 Friday, April 25, in class presentations. Projects due unless you present April 25 (free extension until Monday
More informationMethods of Analysis. Force or Flexibility Method
INTRODUCTION: The structural analysis is a mathematical process by which the response of a structure to specified loads is determined. This response is measured by determining the internal forces or stresses
More informationCONVERGENCE PROPERTIES OF ANALYTICAL TARGET CASCADING
9th AIAA/ISSMO Symposium on Multidisciplinary Analysis and Optimization 4-6 September 2002, Atlanta, Georgia AIAA 2002-5506 CONVRGNC PROPRTIS OF ANAYTICA TARGT CASCADING Nestor Michelena Hyungju Park Panos
More informationDynamic Redesign of a Flow Control Servovalve Using a Pressure Control Pilot 1
Perry Y. Li Department of Mechanical Engineering, University of Minnesota, 111 Church St. SE, Minneapolis, MN 55455 e-mail: pli@me.umn.edu Dynamic Redesign of a Flow Control Servovalve Using a Pressure
More informationDynamic Redesign of a Flow Control Servo-valve using a Pressure Control Pilot
Dynamic Redesign of a Flow Control Servo-valve using a Pressure Control Pilot Perry Y. Li Department of Mechanical Engineering University of Minnesota Church St. SE, Minneapolis, Minnesota 55455 Email:
More informationRecoverable Robustness in Scheduling Problems
Master Thesis Computing Science Recoverable Robustness in Scheduling Problems Author: J.M.J. Stoef (3470997) J.M.J.Stoef@uu.nl Supervisors: dr. J.A. Hoogeveen J.A.Hoogeveen@uu.nl dr. ir. J.M. van den Akker
More informationNOTES ON FIRST-ORDER METHODS FOR MINIMIZING SMOOTH FUNCTIONS. 1. Introduction. We consider first-order methods for smooth, unconstrained
NOTES ON FIRST-ORDER METHODS FOR MINIMIZING SMOOTH FUNCTIONS 1. Introduction. We consider first-order methods for smooth, unconstrained optimization: (1.1) minimize f(x), x R n where f : R n R. We assume
More informationA COUPLED-ADJOINT METHOD FOR HIGH-FIDELITY AERO-STRUCTURAL OPTIMIZATION
A COUPLED-ADJOINT METHOD FOR HIGH-FIDELITY AERO-STRUCTURAL OPTIMIZATION Joaquim Rafael Rost Ávila Martins Department of Aeronautics and Astronautics Stanford University Ph.D. Oral Examination, Stanford
More informationVibration Transmission in Complex Vehicle Structures
Vibration Transmission in Complex Vehicle Structures Christophe Pierre Professor Matthew P. Castanier Assistant Research Scientist Yung-Chang Tan Dongying Jiang Graduate Student Research Assistants Vibrations
More informationOptimal Approximations of Coupling in Multidisciplinary Models
Optimal Approximations of Coupling in Multidisciplinary Models Ricardo Baptista, Youssef Marzouk, Karen Willcox, Massachusetts Institute of Technology, Cambridge, MA and Benjamin Peherstorfer University
More informationLecture Notes: Geometric Considerations in Unconstrained Optimization
Lecture Notes: Geometric Considerations in Unconstrained Optimization James T. Allison February 15, 2006 The primary objectives of this lecture on unconstrained optimization are to: Establish connections
More informationCHAPTER 6 A STUDY ON DISC BRAKE SQUEAL USING DESIGN OF EXPERIMENTS
134 CHAPTER 6 A STUDY ON DISC BRAKE SQUEAL USING DESIGN OF EXPERIMENTS 6.1 INTRODUCTION In spite of the large amount of research work that has been carried out to solve the squeal problem during the last
More informationRobust Optimization: Design in MEMS. Peter Josef Sedivec. B.S. (University of California, Berkeley) 2000
Robust Optimization: Design in MEMS by Peter Josef Sedivec B.S. (University of California, Berkeley) 2000 A report submitted in partial satisfaction of the requirements for the degree of Master of Science,
More informationAM 205: lecture 19. Last time: Conditions for optimality, Newton s method for optimization Today: survey of optimization methods
AM 205: lecture 19 Last time: Conditions for optimality, Newton s method for optimization Today: survey of optimization methods Quasi-Newton Methods General form of quasi-newton methods: x k+1 = x k α
More informationInstitute of Structural Engineering Page 1. Method of Finite Elements I. Chapter 2. The Direct Stiffness Method. Method of Finite Elements I
Institute of Structural Engineering Page 1 Chapter 2 The Direct Stiffness Method Institute of Structural Engineering Page 2 Direct Stiffness Method (DSM) Computational method for structural analysis Matrix
More informationA Comprehensive Robust Design Approach for Decision Trade-Offs in Complex Systems Design. Kurt Hacker
Proceedings of DETC 99: 1999 ASME Design Engineering Technical Conferences September 115, 1999 Las Vegas, Nevada DETC99/DAC8589 A Comprehensive Robust Design Approach for Decision TradeOffs in Complex
More informationIntegrated reliable and robust design
Scholars' Mine Masters Theses Student Research & Creative Works Spring 011 Integrated reliable and robust design Gowrishankar Ravichandran Follow this and additional works at: http://scholarsmine.mst.edu/masters_theses
More informationCSC 411: Lecture 04: Logistic Regression
CSC 411: Lecture 04: Logistic Regression Raquel Urtasun & Rich Zemel University of Toronto Sep 23, 2015 Urtasun & Zemel (UofT) CSC 411: 04-Prob Classif Sep 23, 2015 1 / 16 Today Key Concepts: Logistic
More informationLEAST ANGLE REGRESSION 469
LEAST ANGLE REGRESSION 469 Specifically for the Lasso, one alternative strategy for logistic regression is to use a quadratic approximation for the log-likelihood. Consider the Bayesian version of Lasso
More informationStochastic Analogues to Deterministic Optimizers
Stochastic Analogues to Deterministic Optimizers ISMP 2018 Bordeaux, France Vivak Patel Presented by: Mihai Anitescu July 6, 2018 1 Apology I apologize for not being here to give this talk myself. I injured
More informationMIT Manufacturing Systems Analysis Lecture 14-16
MIT 2.852 Manufacturing Systems Analysis Lecture 14-16 Line Optimization Stanley B. Gershwin Spring, 2007 Copyright c 2007 Stanley B. Gershwin. Line Design Given a process, find the best set of machines
More informationAM 205: lecture 19. Last time: Conditions for optimality Today: Newton s method for optimization, survey of optimization methods
AM 205: lecture 19 Last time: Conditions for optimality Today: Newton s method for optimization, survey of optimization methods Optimality Conditions: Equality Constrained Case As another example of equality
More informationIntegration of measured receptance into a time domain simulation of a Multi Body Model using SIMPACK
Fakultät Maschinenwesen Professur für Dynamik und Mechanismentechnik Integration of measured receptance into a time domain simulation of a Multi Body Model using SIMPACK Dipl.-Ing. Johannes Woller Prof.
More information2D Decision-Making for Multi-Criteria Design Optimization
DEPARTMENT OF MATHEMATICAL SCIENCES Clemson University, South Carolina, USA Technical Report TR2006 05 EW 2D Decision-Making for Multi-Criteria Design Optimization A. Engau and M. M. Wiecek May 2006 This
More informationMechanical Vibrations Chapter 6 Solution Methods for the Eigenvalue Problem
Mechanical Vibrations Chapter 6 Solution Methods for the Eigenvalue Problem Introduction Equations of dynamic equilibrium eigenvalue problem K x = ω M x The eigensolutions of this problem are written in
More informationCONSTRAINED OPTIMIZATION OVER DISCRETE SETS VIA SPSA WITH APPLICATION TO NON-SEPARABLE RESOURCE ALLOCATION
Proceedings of the 200 Winter Simulation Conference B. A. Peters, J. S. Smith, D. J. Medeiros, and M. W. Rohrer, eds. CONSTRAINED OPTIMIZATION OVER DISCRETE SETS VIA SPSA WITH APPLICATION TO NON-SEPARABLE
More informationMultibody dynamics of mechanism with secondary system
University of Iowa Iowa Research Online Theses and Dissertations Spring 212 Multibody dynamics of mechanism with secondary system Jun Hyeak Choi University of Iowa Copyright 212 Jun Choi This thesis is
More informationNeural networks COMS 4771
Neural networks COMS 4771 1. Logistic regression Logistic regression Suppose X = R d and Y = {0, 1}. A logistic regression model is a statistical model where the conditional probability function has a
More informationOptimal dynamic operation of chemical processes: Assessment of the last 20 years and current research opportunities
Optimal dynamic operation of chemical processes: Assessment of the last 2 years and current research opportunities James B. Rawlings Department of Chemical and Biological Engineering April 3, 2 Department
More informationComputational Stiffness Method
Computational Stiffness Method Hand calculations are central in the classical stiffness method. In that approach, the stiffness matrix is established column-by-column by setting the degrees of freedom
More informationPrecision Attitude and Translation Control Design and Optimization
Precision Attitude and Translation Control Design and Optimization John Mester and Saps Buchman Hansen Experimental Physics Laboratory, Stanford University, Stanford, California, U.S.A. Abstract Future
More informationk 21 k 22 k 23 k 24 k 31 k 32 k 33 k 34 k 41 k 42 k 43 k 44
CE 6 ab Beam Analysis by the Direct Stiffness Method Beam Element Stiffness Matrix in ocal Coordinates Consider an inclined bending member of moment of inertia I and modulus of elasticity E subjected shear
More informationBasics of Uncertainty Analysis
Basics of Uncertainty Analysis Chapter Six Basics of Uncertainty Analysis 6.1 Introduction As shown in Fig. 6.1, analysis models are used to predict the performances or behaviors of a product under design.
More informationCBE495 LECTURE IV MODEL PREDICTIVE CONTROL
What is Model Predictive Control (MPC)? CBE495 LECTURE IV MODEL PREDICTIVE CONTROL Professor Dae Ryook Yang Fall 2013 Dept. of Chemical and Biological Engineering Korea University * Some parts are from
More informationHaris Malik. Presentation on Internship 17 th July, 2009
Haris Malik University of Nice-Sophia Antipolics [math.unice.fr] Presentation on Internship 17 th July, 2009 M. Haris University of Nice-Sophia Antipolis 1 Internship Period Internship lasted from the
More informationSolving Quadratic Equations Using Multiple Methods and Solving Systems of Linear and Quadratic Equations
Algebra 1, Quarter 4, Unit 4.1 Solving Quadratic Equations Using Multiple Methods and Solving Systems of Linear and Quadratic Equations Overview Number of instructional days: 13 (1 day = 45 minutes) Content
More informationSection 11.1 Sequences
Math 152 c Lynch 1 of 8 Section 11.1 Sequences A sequence is a list of numbers written in a definite order: a 1, a 2, a 3,..., a n,... Notation. The sequence {a 1, a 2, a 3,...} can also be written {a
More informationSoftware for Integer and Nonlinear Optimization
Software for Integer and Nonlinear Optimization Sven Leyffer, leyffer@mcs.anl.gov Mathematics & Computer Science Division Argonne National Laboratory Roger Fletcher & Jeff Linderoth Advanced Methods and
More informationMultiobjective Optimization Applied to Robust H 2 /H State-feedback Control Synthesis
Multiobjective Optimization Applied to Robust H 2 /H State-feedback Control Synthesis Eduardo N. Gonçalves, Reinaldo M. Palhares, and Ricardo H. C. Takahashi Abstract This paper presents an algorithm for
More informationOptimization methods
Lecture notes 3 February 8, 016 1 Introduction Optimization methods In these notes we provide an overview of a selection of optimization methods. We focus on methods which rely on first-order information,
More informationDistributed Optimization: Analysis and Synthesis via Circuits
Distributed Optimization: Analysis and Synthesis via Circuits Stephen Boyd Prof. S. Boyd, EE364b, Stanford University Outline canonical form for distributed convex optimization circuit intepretation primal
More informationKaisa Joki Adil M. Bagirov Napsu Karmitsa Marko M. Mäkelä. New Proximal Bundle Method for Nonsmooth DC Optimization
Kaisa Joki Adil M. Bagirov Napsu Karmitsa Marko M. Mäkelä New Proximal Bundle Method for Nonsmooth DC Optimization TUCS Technical Report No 1130, February 2015 New Proximal Bundle Method for Nonsmooth
More informationAutomatic Control II Computer exercise 3. LQG Design
Uppsala University Information Technology Systems and Control HN,FS,KN 2000-10 Last revised by HR August 16, 2017 Automatic Control II Computer exercise 3 LQG Design Preparations: Read Chapters 5 and 9
More informationMacro 1: Dynamic Programming 1
Macro 1: Dynamic Programming 1 Mark Huggett 2 2 Georgetown September, 2016 DP Warm up: Cake eating problem ( ) max f 1 (y 1 ) + f 2 (y 2 ) s.t. y 1 + y 2 100, y 1 0, y 2 0 1. v 1 (x) max f 1(y 1 ) + f
More informationInfeasibility Detection and an Inexact Active-Set Method for Large-Scale Nonlinear Optimization
Infeasibility Detection and an Inexact Active-Set Method for Large-Scale Nonlinear Optimization Frank E. Curtis, Lehigh University involving joint work with James V. Burke, University of Washington Daniel
More informationSIMPLEX LIKE (aka REDUCED GRADIENT) METHODS. REDUCED GRADIENT METHOD (Wolfe)
19 SIMPLEX LIKE (aka REDUCED GRADIENT) METHODS The REDUCED GRADIENT algorithm and its variants such as the CONVEX SIMPLEX METHOD (CSM) and the GENERALIZED REDUCED GRADIENT (GRG) algorithm are approximation
More informationAlgorithms for constrained local optimization
Algorithms for constrained local optimization Fabio Schoen 2008 http://gol.dsi.unifi.it/users/schoen Algorithms for constrained local optimization p. Feasible direction methods Algorithms for constrained
More informationD R A F T. P. G. Hänninen 1 M. Lavagna 1 Dipartimento di Ingegneria Aerospaziale
Multi-Disciplinary Optimisation for space vehicles during Aero-assisted manoeuvres: an Evolutionary Algorithm approach P. G. Hänninen 1 M. Lavagna 1 p.g.hanninen@email.it, lavagna@aero.polimi.it 1 POLITECNICO
More information