1 / 35

KBE-System and Case-Library for UG-Structures EUUG Usermeeting 2003 Peter Binde

C ase B ased E ngineering. KBE-System and Case-Library for UG-Structures EUUG Usermeeting 2003 Peter Binde. Content. - Dr. Wallner Group - Motivation for a new Assistant System - Case Study: Automotive-Generator - Feature-Processing - Concepts of the CBE-Assistant.

lundy
Download Presentation

KBE-System and Case-Library for UG-Structures EUUG Usermeeting 2003 Peter Binde

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. CaseBasedEngineering KBE-System and Case-Library for UG-Structures EUUG Usermeeting 2003 Peter Binde

  2. Content - Dr. Wallner Group - Motivation for a new Assistant System - Case Study: Automotive-Generator - Feature-Processing - Concepts of the CBE-Assistant

  3. Dr. Wallner Group Service around CAD, CAE, PDM Focus on EDS-PLM Products Competences: - Process-analyses and -syntheses - Trainings and workshops - User-support and hotline - Integration-solutions and software-development

  4. CaseBasedEngineering Dr. Wallner Group: - Simulation Focus on Design-Embedded Product-Simulation Training, Consulting, Support for CAE Software for Knowledge-Processing Projects shared by Industry & Research

  5. Dr. Wallner Group: - The Founder‘s Vision Jens Wallner died this year by a car accident His vision: - High Quality Service, for CAD/CAE/PDM – will continue in the four Wallner-companies,that belong to the Group.

  6. Motivation for a new Assistant System

  7. Original Scenario Original Scenario Bördelung Scenario Bead =80N/mm2 u=0,9mm Vergleich = 72N/mm2 u=0,8mm =71N/mm2 u=0,4mm Present Situation More and more FEA becomes the designers daily work VDI Report about Adam Opel Project: - Designers do FEA themselves -

  8. Deficits - Often special-knowledge is necessary - Different methods for same problems are used - Knowledge stays in the heads of specialists and key users - Examples and summarised methods are only static - Databases organize data, but do not know about the content ?

  9. Storage of a new solution Retrieval of most appropriate cases Case Base Adapting knowledge to a new Solution Solution: - Store all knowledge anyone has learned in past-cases. - Reuse that knowledge Very similar to human learning Method: - Case Based Reasoning (CBR) Superior to application of rules

  10. All we need is: - Two new buttons in our UG-Scenario system:

  11. Questions: - How to store a solution? - How to adapt knowledge from a similar solution? - How to find the most similar cases? Store Find Case Base Answers: - Extensively use feature-processing - Store the feature-structure of successfully solved cases - Find similar cases by pattern recognition of feature-structures - Let the user interpret and adapt knowledge Adapt

  12. Case Study: Automotive Generator Analysis Goals:Find Characteristics, that Contain Knowledge Develop a Structure for that Knowledge

  13. Contact-Condition Solver-Parameter Design-context Part for analysis & part-properties Form-feature at result: Cutout, Blended Result-type: Stress & validation Mesh-property at result, Geometry-preparation Centrifugal-Analysis of Fan: Characteristics, that Contain Knowledge Load-type: Rotation

  14. Contact-Condition Design-context Solver-Parameter Support-Condition Form-feature at result Result-type: & Validation Part for analysis & part-properties Mesh-property at result Generator-Axle under Bending-Load

  15. Solver-Parameter Design-context Part for analysis & part-properties Form-feature at result: Blended Result-type: Stress & validation Mesh-property at result Acceleration-Analysis of Housing Forces, that result from contacting parts Load-type: Acceleration& Geometry applied to

  16. Adjoining Parts - Form-feature - Form-feature … Adjoining Parts - Form-feature - Form-feature … … - Form-feature - Form-feature Analysis-part - Form-feature - Form-feature … … - Form-feature - Form-feature Contact Contact Face-FaceCoupling Screw (4x) Contact Stress-Validation (requested) Screw (4x) Description of the Situation and Analysis-Task by Computable Elements Effect: Acceleration

  17. Project Structure for Storage of a Task Analysis-scenario Part Material Property-aggregation... Form-feature Contact Form-feature Part Form-feature Fix Load-Effects Solution-Group Property-aggregation... Form-feature Part Material Analysis-scenario

  18. Description of a Solution by Computable Elements A solution contains idealization-methods and the result • Idealization of screw-connections • Neglected pre-load • Rigid Type, … • Idealization of face-coupling • Restriction of DOFs • Mesh-property in critical area • Number, quality, type of elements • Idealization of acceleration • To be neglected Result: -location, -type, -validation and -consequence • Idealization of adjoining parts • Acceleration-Force on contacting faces • Geometry-Idealization • Small feature removal • Idealization of material • Isotropic, linear Solver-Parameter

  19. Feature-Processing

  20. Features Model Design Characteristics A Feature is a Container for a Design Characteristic Example Feature Weld point: included information: - Group of involved elements - Connectivity to geometry - Strength requirements - Tolerances - Costs - Manufacturing processes ... Advantage: Weld point can be processed as a „weld point“.

  21. Knowledge Management using Semantics and Features We try to build up models only by use of features This method automatically captures our knowledge and provides standard-solutions CAD Features: (1) Sketch (2) Revolved (3) Pocket (4) Pocket (5) Hole (6) Threads FEM Features: Fixed Face Centrifugal Load Concentrated Mass Stress Calculation Fatigue Calculation rot

  22. This,Because the hole can be recognizedautomatically by all following processes Feature-Processing: Simple Example ? ? Which feature-description of a hole is preferred

  23. Knowledge is captured automatically Knowledge is captured manually Extensive Usage of Feature-Processing • -The user builds a clear, short feature-structure • The user inserts further knowledge by assigning names, groups, descriptions, comments, … . Insert as much knowledge as possible to the model. Manually definition of features is also possible.

  24. CaseBasedEngineering Concepts of the CBE-Assistant

  25. Feature Recognition Extracts and classifies all available feature-information to an extended feature-navigator:

  26. Transparency Example: Part-connections are shown completely: “Fan” is connected to “Rotator”. The involved form-features as well as contact properties are shown.

  27. Feature-Attributes Properties of features can be edited

  28. Manual Feature Definition Features, not supported by UG-scenario can be added: Stress-Results are validated on a Blend Form-Feature of the “Fan”.This feature is “extra” inserted by the user.

  29. Meets the “designers language” Tasks / Solutions are Separated CBE feature-information merely includes task (situation) -descriptions Corresponding solutions are given in the found cases Problem-Description: Search for non-linear contact Search for contacting parts, that can separate

  30. Storage of Successful Solved Cases The feature-Information is stored in a xml database CAD/FEM-System Analysis Case CBE-System prt xml Case Libary

  31. Definition of a Problem: Two Possibilities 1. Build up features in the CAD/FEM-system 1. Build up features in the CBE-system Advantage:This method can be used with any CAD/FEM-system Advantage:All existing features are used automatically 2. Start CBE-system3. Optional insert further knowledge4. Start search 2. Start search

  32. Similarity-Measurements: Method A: Find “Similar Features” Incremental ranking of a requested feature-class based on a user-weighted “Manhattan-Distance”: Example:Requested feature-class: Partn=10 Attributes + Weights Example:Number of requested results: 6

  33. Similarity-Measurements: Method B: Find “Similar Feature-Structures” Exact sub-graph matching algorithm Developed by: Dennis Shasha, Courant Institute,New York University & Rosalba Giugno, Math & Computer Science, University Catania

  34. Test-pattern: P containedIn(P,A) = true A Example Stress-analysis on a sheet-metal-bead Task-pattern:

More Related