Knowledge-based Systems for Industrial Applications

Transcription

1 Knowledge-based Systems for Industrial Applications 1 The Topic 2 Tasks Goal: Overview of different tasks Systematic and formal characterization as a requirement for theory and implementation Script: Chap

2 Basis for Problem Solving: Knowledge about the Technologies (Re-)Design Function Components Processes Materials Recycling Manufacturing Maintenance Control Productionprocesses Disturbances Faults

3 Basis for Problem Solving: Models Representation of knowledge about a complex domain systematic (Re-)Design structured formal for inference of statements about the behavior Recycling of systems in this domain Manufacturing Not a (behavior) model: Database with product parameters Maintenance Sets of experimental Control data Equation (without equation solver) - 3

4 Model as a Substitute of a System Representation of knowledge about a complex domain systematic (Re-)Design structured formal for inference of statements about the behavior Recycling of systems in this domain Manufacturing Model as a substitute of the real systems: certain aspects not directly observable, critical situations, Maintenance Control real system not (yet) available Model as an approximation of real system behavior - 4

5 Notation ^ entails does not entail inconsistency, false ^ is consistent ^ is inconsistent?^ check for consistency - 5

6 Design Generative Design Configuration Parametric Design Design Verification Design for X Design Recycling Manufacturing Task: Maintenance Control Concept of a realizable system whose behavior fulfills a specification of goals Control Generation/ Verification - 6

7 The Design Task? GOALS Task: Concept of a realizable system whose behavior fulfills a specification of goals - 7

8 Model-based Design MODEL? GOALS Task: search for a model that is consistent with the specification of goals - 8

9 The Design Task - Formally MODEL GOALS Formally: MODEL GOALS Weaker: MODEL GOALS ^ - 9

10 Iterative Design Model Revision MODEL GOALS Task: Incremental modification of design hypotheses Precondition: good initial hypothesis - 10

11 Iterative Design - Formally Model Revision Design Alternatives MODEL GOALS Formally: MODEL GOALS ^ MODEL GOALS ^ - 11

12 Space of Design Alternatives Different Design Tasks Model Revision Design Alternatives parametric design: alternative parameters configuration: alternative component types Innovative design: alternative structures new component types - 12

13 Failure-Modes-and-Effects Analysis (FMEA) Design Failure-Modes-and- Effects Analysis Recycling Manufacturing Task: Maintenance Control Determine of the effects of (classes of) component faults under different operating conditions - 13

14 The FMEA Task Stuck-closed? BEHAVIOR/ GOALS Task: Determine of the effects of (classes of) component faults under different operating conditions - 14

15 Model-based FMEA Stuckclosed MODEL BEHAVIOR/ GOALS Task: Predict the behavior of a model of the faulty system - 15

16 The FMEA Task - Formally Stuckclosed MODEL BEHAVIOR/ GOALS Formally: MODEL fault OPERATING_CONDITIONS? MODEL fault OPERATING_CONDITIONS? EFFECT - 16

17 Manufacturing, Assembly - Tasks Recycling Design Manufacturing Production Planning, Scheduling Essential: Not the behavior of the final product, but the behavior Maintenance of the manufacturing line Control Manufacturing process subject to: - design, monitoring, control, maintenance - 17

18 Manufacturing, Assembly - Tasks Design Recycling Manufacturing Testing Provide evidence: with (few/cheap) tests, Maintenance Control That the manufactured system operates correctly i.e. has no relevant component faults - 18

19 Test Generation - Task + V mes ECU OK 3 1 OBSERVATIONS INPUT + V mes ECU OBSERVATIONS 3 1 Provide evidence: with (few/cheap) tests, That the manufactured system operates correctly i.e. has no relevant component faults - 19

20 Test Generation - Formally + V mes ECU OK 3 1 OBSERVATIONS INPUT + V mes ECU... F4 F3 F2 F OBSERVATIONS 3 1 Formally: MODEL 0 INP OBS 0 MODEL 1 INP OBS 1 OBS 0 OBS 1 ^ - 20

21 Monitoring Design Recycling Manufacturing For the given Maintenance observations: Monitoring Does the system work consistently Fault with Detection the goal specification? - 21

22 Model-based Monitoring Observations MODEL? ^ GOALS For the observations given : Is the model consistent with the goal specification? - 22

23 Monitoring - Formally Observations MODEL? ^ GOALS Formally: MODEL OBS GOALS? ^ If system design is correct: MODEL GOALS then MODEL OBS? ^ suffices - 23

24 Diagnosis Design Recycling Manufacturing Task: Maintenance Control Determine based on the given observations: What is happening in the system? Fault Detection/ Identification - 24

25 The Diagnosis Task? OBS Task: Determine based on the given observations: What is happening in the system? - 25

26 Model-based Diagnosis MODEL? OBS Task: Determine the models that are consistent with the observations - 26

27 The Diagnosis Task - Formally MODEL? OBS Formally: MODEL OBS ^ - 27

28 Iterative Diagnosis Model Revision MODEL OBS Task: Incremental modification of fault hypotheses Initial hypothesis: system as designed (all components OK) - 28

29 Iterative Diagnosis - Formally Model Revision Disturbances MODEL OBS Formally: MODEL OBS ^ MODEL OBS ^ - 29

30 Testing for Diagnosis - Task Design Recycling Manufacturing Testing Task: Maintenance Determine with (few/cheap) tests, which fault occurred. Control - 30

31 Test Generation for Diagnosis + V mes ECU F1 3 1 OBSERVATIONS INPUT + V mes ECU F OBSERVATIONS 3 1 Formally: MODEL F1 INP OBS 1 MODEL F2 INP OBS 2 OBS 1 OBS 2 ^ - 31

32 Therapy Design Recycling Manufacturing Task: Maintenance Re-establishment of goal functionality as far as (Self-) possible Reconfiguration, Repair Recovery Control - 32

33 Therapy Task hydraulic unit brake pedal Therapy F ^ GOALS front left wheel rear right wheel Task: Re-establishment of goal functionality as far as possible - 33

34 Iterative Model-based Therapy Proposal Model Revision Therapy Actions MODEL ^ GOALS Task: Modification of the model by actions until it is (or will be) consistent with the goals - 34

35 Therapy Proposal - Formally Model Revision Therapy Actions MODEL ^ GOALS Formally: MODEL GOALS ^ MODEL ACTIONS GOALS ^ - 35

36 Therapy Needs Time! Model Revision Therapy Actions MODEL ^ GOALS Formally: MODEL GOALS ^ MODEL ACTIONS GOALS ^ Time point t 1 Time point t 2 > t 1-36

37 Different Therapy Tasks Model Revision Therapy Actions Repair (in the narrow sense): Replacement of faulty components Reconfiguration: exploit structural redundancy (Nets) exploit analytical redundancy (new point of work) Process therapy : structural changes - 37

38 Diagnosis, Therapy,... Model Revision Therapy Actions MODEL ^ GOALS Model Revision Disturbances MODEL ^ OBS - 38

39 Consistency-based Problem Solving: Situation Assessment Inconsistency inconsistent Predictor Model Solution consistent Observations - 39

40 Consistency-based Problem Solving: Therapy Proposal Inconsistency inconsistent Predictor Model Therapy consistent Goal - 40

41 Elements of Model-based Problem Solvers: Predictor Model-based prediction and consistency check Criteria : Observations Goals Inconsistency inconsistent Predictor Model Solution consistent Criterion - 41

42 Elements of Model-based Problem Solvers: Model Revision Revision of inconsistent models Model Revision Inconsistency inconsistent Predictor Model Solution consistent Criterion - 42

43 Model Revision Revisions: Component faults External disturbances Unexpected objects... Revision at a conceptual level! Library Model Revision Inconsistency inconsistent Model Predictor Solution consistent Criterion - 43

44 Model Revision? - at a Conceptual Level! Model Revision conceptual Conceptual Model? Model Composer Inconsistency inconsistent Predictor Behavior Model Model mathematical Solution consistent Criterion - 44

45 Requirements on Modeling conceptual modeling executable behavior models qualitative models (automatic) model composition behavior prediction multiple modeling Conceptual Model Model Composer Inconsistency inconsistent Predictor Behavior Model mathematical Solution consistent Criterion - 45