Fleet Usage Spectrum Evaluation & Mission Classification

Transcription

1 Fleet Usage Spectrum Evaluation & Mission Classification 1

2 Background Design Process C-130 Application Results Summary 2

3 3 Background

4 Usage Description Measurement of parameters which are representative of the stress cycle and can be related to crack length. Usage Descriptions: Mission By Mission Mission Segment By Mission segment Time In Usage Category Strain History Damage Parameter 4

5 Mission By Mission Based on pilot log data Rapid method for converting usage data to damage. Most common on cargo fleets. 5

6 Mission Segment - Recognition 6 USAF Damage Tolerance Design Handbook

7 Mission Segment - Regime Recognition Requires substantial amounts of calculation effort. 7

8 Time In Usage Category 8 USAF Damage Tolerance Design Handbook

9 Strain History 9 USAF Damage Tolerance Design Handbook

10 Damage Parameter 10 USAF Damage Tolerance Design Handbook

11 Goal Proof of concept for automated Fleet Usage Spectrum Evaluation FUSE and mission classification 11

12 12 Design Process

13 Pattern Recognition Approach Flight science FUSE & Mission Classification DSP Image Processing Multi Disciplinary Approach Knowledge Fusion Implementation Smoothing Feature Extraction Loads Artificial Intelligence Classification Statistics 13

14 Operational Data Segmentation Smoothing Integral Feature Extraction Statistical 14 Mission Spectrum Mission Classification

15 15 Segmentation

16 16 Smoothing

17 17 = dxdy y y x x y x f q p pq ' ', μ = dxdy y x f xdxdy y x f x,, ' = dxdy y x f ydxdy y x f y,, ' λ μ μ 00 pq pq = 1 2 = q p λ Moment Invariants Moment Invariants

18 18 { } { } { } { } { } φ φ φ φ φ φ φ φ = = = = = = = = Moment Invariants Moment Invariants

19 19 Clustering

20 20 Clustering

21 21 = = = > =< n r j r i r j i j k j j k i n x a x a Cluster x d d Cluster x x a x a x a x , min..., Nearest Cluster : Classification Classification

22 22 C-130 Application

23 C-130 IAT Improvements CDR PILOTS REPORT FLIGHT- HOURS SYSTEM Mission Classification OPERATIONAL DATA NDI DATA CONFIGURATION DATA REPAIRS DATA IAT SYSTEM DTA ANALYSIS DATA NDI FORECAST DAMAGE STATUS MISSION SPECTRUM MANAGEMENT REPORT 5 YEARS REPORT COMPONENT REPORT 23

24 IAT upgrade current status Installation of CDR systems. Code development for transforming raw CDR data to a data base Complete. Phase 1 Cycle and Exccedance Counting algorithm Completed. Phase 2 Mission definition and categorization based on CDR data Due 3/03. 24

25 25 Alt & Vel vs. Time

26 26 Mission Scatter

27 Global Mission Vs. Sub Mission Altitude T 27

28 Sub Missions Reduce number of basic mission Alt_mini Alt_Final0 Value התיחנ הארמה Can effect on the inspection interval because it requires different Ground-Air Ground cycle Alt_mini Alt_Final0 Value התיחנ הארמה 28

29 Hieratical Clustering Algorithm Calculate similarity measure between data points Place each point near its most similar point Group similar data points to cluster if they are close based on the similarity measure. Increase similarity measure threshold and reconsider merging clusters. 29

30 Traditional H.C. Divergence 30

31 Clustering nearest Cluster 31

32 32 Natural Stopping Criteria

33 33 Cluster Representation

34 34 3D Classification spheres

35 Classification Decision Surface 35 - Clusters congruency

36 3D Distorted Classification spheres 36

37 37 Results

38 38 Results Analysis

39 39 Results Analysis

40 40 1#

41 41 3#

42 42 4#

43 43 Summary

44 Summary The automated FUSE concept was proved to be feasible: The typical missions are identified automatically. automatically capture the number of typical missions. Mission classification is relatively simple task when using pattern recognition approach. Enable the fleet managers to easily identify usage changes. The Operational Data Recorders can be use to increase the reliability of IAT systems. 44

45 Summary The automated FUSE concept was proved to be feasible: The typical missions are identified automatically. automatically capture the number of typical missions. Mission classification is relatively simple task when using pattern recognition approach. Enable the fleet managers to easily identify usage changes. The Operational Data Recorders can be use to increase the reliability of IAT systems. 45