1 / 24

Solving Interval Constraints in Computer-Aided Design

Solving Interval Constraints in Computer-Aided Design. Yan Wang Department of Industrial Engineering NSF Center for e-Design University of Pittsburgh. Outline . Parametric geometric modeling Interval geometric modeling Constraint solving. Parametric Geometric Modeling. Geometric model

opeel
Download Presentation

Solving Interval Constraints in Computer-Aided Design

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Solving Interval Constraints in Computer-Aided Design Yan Wang Department of Industrial Engineering NSF Center for e-Design University of Pittsburgh

  2. Outline • Parametric geometric modeling • Interval geometric modeling • Constraint solving

  3. Parametric Geometric Modeling • Geometric model • Geometry • Topology • Attributes • Constraint solver • Numerical • Symbolic • Graph-based / Constructive • Rule-based reasoning • Visualization

  4. Fixed-Value Parameter vs. Interval-Value Parameter • Fixed-value parameters may generate inconsistency errors from floating-point arithmetic. • Fixed-value constraints bring up conflicts easily at later design stages. • Fixed-value parameters make the development of Computer-Aided Conceptual Design difficult. • Interval parameters improve robustness of geometry computation. • Interval parameters capture the uncertainty and inexactness. • Interval parameters directly represent boundary information for optimization. • Intervals provide a generic representation for geometric constraints.

  5. Application of IA in CAD/CAE • Computer graphics: rasterizing [Mudur and Koparkar], ray tracing [Toth, Kalra and Barr], collision detection [Moore and Wilhelms, Von Herzen et al., Duff, Snyder et al.]. • CAD: curve approximation [Sederberg and Farouki, Patrikalakis et al., Chen and Lou, Lin et al.], shape interrogation [Maekawa and Patrikalakis], robust boundary evaluation [Patrikalakis et al., Wallner et al.] • CAE: finite element formulation [Muhanna and Mullen] • System design: set-based modeling [Finch and Ward],structural analysis [Rao et al.]

  6. Display Interactivity Tolerance • equivalence: • nominal equivalence: • strictly greater than or equal to: • strictly greater than: • strictly less than or equal to: • strictly less than: • inclusion: Nominal Intervals in IGM Given that A =[aL, aN, aU], B =[bL, bN, bU],

  7. Sampling Relation between Real Number and Interval Number Strict relations • strict equivalence: • strictly greater than or equal to: • strictly greater than: • strictly less than or equal to: • strictly less than: Global relations • global equivalence: • greater than or equal to: • greater than: • less than or equal to: • less than:

  8. Set vs. Individuals • Global relations are default relations in IA. • Global relations ensure the feasibility of interval arithmetic operations and solutions. • Global relations make global solution and optimization of interval analysis possible. • Strict relations exhibit the rigidity of RA. • Strict relations specify constraints between variables directly.

  9. Preference, Specification, & Interval Constraint • Improve specification interoperability for design life-cycle • Represent soft constraint • Capture the uncertainty of design • Model incompleteness and inexactness especially during conceptual design • Model a set of design alternatives • Represent tolerance and boundary information for global optimization • Improve robustness of computation

  10. Under-, Over-, & Well-Constrained

  11. Special Considerations of Interval Linear Equations for CAD • Matrix-based methods are not for under- or over-constrained problems • Iteration-based methods (e.g. Jacobi iteration, Gauss-Seidel iteration) are more general and useful in CAD constraint solving

  12. X A Y Extended Gauss-Seidel Method

  13. Solving Interval Nonlinear Equations based on Linear Enclosure • Transform to separable form; • Find linear enclosure; • Solve linear enclosure equations; • Update variable values • If stop criteria not satisfied, go to step 2; otherwise stop. Start Transform to Separable Form Find Linear Enclosure Solve Linear Enclosure Equations Update Variable Values N Stop Criteria Satisfied? Y End

  14. 1. Separable Form • Function f(x1, x2, …, xn) is said to be separable iff f(x1, x2, …, xn) = f1(x1) + f2(x2) + … + fn(xn). Yamamura’s algorithm[Yamamura,1996]: +, , , /, sin, exp, log, sqrt, ^, etc. For example: f = f1 f2f = (y2 f12 f22)/2 y = f1 + f2 f = f1 / f2 f = (y2 f121/ f22)/2 y = f1 + 1/f2 f = (f1)f2f =exp(y1) y1= (y22 (log(f1))2 f22)/2 y2 = log(f1)+ f2

  15. 2. Linear Enclosure Extending Kolev’s work: Let Xj0 = [xLj, xNj, xUj] Linear Enclosure is defined as: such that

  16. 3. Solve Linear Enclosure Equations If fij(x) is continuous within interval Xj0, solve using root isolation [Collins et al.] and Secant method. Suppose xjp (p=1, 2, …, P) is the pth solution of the above equation, and xj0=xLj. Let Bij=[bLij, bNij, bUij], where

  17. 4. Update Variable Values • Suppose Yj is the jth variable solution of linear enclosure equations in the kth iteration, update Xj for (k+1)th iteration by • If an empty interval is derived, the original system has no solution within the given initial intervals. • If the stop criterion is not met, iterate.

  18. Solving Interval Inequalities • Adding slack variables to translate inequalities into equalities. • Solving linear/nonlinear equations with previous methods.

  19. Interval Subdivision • Subpaving divides a hyper-cube into multiple smaller hyper-cubes recursively • Implemented as order elevation of power interval P(m, n) = [X1, X2, …, Xm]

  20. Constraint Re-Specification • Need to differentiate active and inactive constraints. • For a constraint set p = {f(X) = Y and g(X) = Z}, the subset f(X) = Y with respect to a solution DX is inactive if f(D) Y and g(D) Z. (a) f – inactive, g – active (b) f – active, g – active (c) f – active, g – inactive

  21. An Example

  22. subdivide up to Level 3, and some sub-regions are eliminated. Refinement - subdivision

  23. What can interval provide for design? • The decisions to fix values of parameters can be postponed to later design stages. • Variation and uncertain are inherent in the process of design. • Soft constraint-driven geometry modeling • Support under- and over-constrained problem • Integrated linear, nonlinear equations, and inequality solving

  24. Thank you!

More Related