Aplikasi Sinyal Fibrasi Untuk CMB Condition Based Maintenance

8/10/2019 Aplikasi Sinyal Fibrasi Untuk CMB Condition Based Maintenance

1/33

Cost & Complexity Trade-offs inPrognostics

Dr. Ir. M. Sabri, MT

Condition Base Maintenance

(CBM)


2/33

The Opportunity

Condition BasedMaintenance (CBM)

promises to deliverimproved maintainabilityand operational availabilityof naval systems whilereducing life-cycle costs

The Challenge

Prognostics is the Achilles heel of CBM systems - predicting thetime to failure of critical machines requires new and innovativemethodologies that will effectively integrate diagnostic results with

maintenance scheduling practices

Condition-Based Maintenance


3/33

Prognostics

Objective Determine time window over which maintenance must be

performed without compromising the systems operationalintegrity


4/33

The CBM Architecture

5. FeatureExtraction




Hardware Chiller Sensors DAQ

Hardware Engine Sensors DAQ

3. ModeEstimator /

Usage Pattern Identification

3. ModeEstimator /

Usage Pattern Identification

Central DB Event Dispatch

Event Dispatch

DatabaseManagement

DatabaseManagement

DWNN

Virtual Sensor (WNN)

CPNN

failure dimension

6. Prognostics

Classifier (WNN)

Classifier (WNN)

Classifier (Fuzzy)

Classifier (Fuzzy)

5. Diagnostics

PEDS Software System Architecture

2. DataPreprocessing

2. DataPreprocessing

1. DDL

1. DDL Interface


5/33

The CBM Architecture (continued)

Diagnostic Models: Fuzzy Logic Based Wavelet Neural Network Model Rough Set Theory based NN Model

Prognostic Models: Dynamic Wavelet Neural Network Model Confidence Prediction Neural Network Model

Physical Models of Failure Mechanisms


6/33

Objec t Oriented HybridSystem Models

Diagnostic Algorithms

Prognostic Algorithms

Static and DynamicCase Library

S o

f t w a r e

R e p o s i

t o r y

System (FMTV, PLS,

etc.)

Sensors DAQ/CPU

H a r d w a r e

Interface

Army Vehicle Systems

Assessment Module

Maintainer

Designer

Statistics Optimization Performance

Prescription Maintenance

Plan

Communityof Agents Multiagent System

M ul t i a g e n t S y s t e m

Intelligent SelectionLayer

Decision Support Layer

Interface

Layer

I n t e l l i g e n t A g e n t

Preliminary Diagnostics

Online

Offline

/Components

Model-Based CBM Architecture


7/33

Centralized Control & KBArchitectures

UUT

A Generic Central Control and Knowledge Base Framework

Diagnostic

Algorithm

Prognostic Algorithm

Control

KnowledgeBase

Diagnosis

PrognosisSensors

Events

Preprocessing

Data-mining

FeatureExtraction

UUTSensors

Events

UUTSensors

Events


8/33

Distributed Control & KBArchitectures

U U T

UUT

U U T

Distributed Control and KnowledgeBase Framework

Diagnosis

Prognosis

Diagnostic Algorithms

Prognostic Algorithms

CentralControl

CentralKnowledge

Base

LocalDiagnostic Algorithms

LocalPrognostic Algorithms

LocalControl

LocalKB

L o c a l D

i a g n o s t i c

A l g o r i t h

m s

L o c a l P

r o g n o s

t i c A l g o r

i t h m s

L o c a l C

o n t r o l

L o c a l K

B

Lo c a l D i a g n o s t i c Al g o r i t h m s Lo c a l P r o g n o s t i c Al g o r i t h m s

Lo c a l C o n t r o l

Lo c a l K B

Sensors

Events

S e n s o r s E v e n t s

S e n s o r s

E v e n t s

Knowledge

Fusion


9/33

Case-Based Reasoning &Learning

CBR - an episodic memory of past experiences CBR - initial cases by examples CBR Methodology:

Indexing (generate indices for classification and categorization)Retrieval (retrieve the best past cases from the memory)

Adaptation (modify old solution to conform to new situation)Testing (did the proposed solution work)

Learning (explain failed & store successful solutions)

Case LibraryFailure Mode i

SymptomsCase #

S1 S2 SmTests Prescription

123


10/33

On-board/Off-board Diagnostics

OperationalPerformanceMonitoring

PeriodicHealthCheck

Operator

Embedded Diagnostics

Maintainer Pre-Diagnostic

Session

Diagnostic Session

Post-DiagnosticSession

AbnormalPerformance

Detection


11/33

DiagnosticsManager

Platform FamilyCase Library

Platform Data

Platform HistoricalRecords Database

Current DiagnosticsSession Database

Case BasedDiagnostics Reasoner

Knowledge FusionModule

Case-Based Reasoning Architecture


12/33

Sensor

Pressure Sensor

Thermocouple

Current Shunt

Fuel

Subsystem

Starter

Battery

EmbeddedDiagnostics Processor

EmbeddedDiagnostics

Interface

Etc..Etc..

Etc.. J-1708 1553

1939 DCA

PortableDiagnostics &Maintenance

Aid

SPORT

MSD

Support AreaDatabase

Army CentralDatabase

EmbeddedDiagnostics

Interface

EmbeddedDiagnostics

Data Collector

DiagnosticTest

Updates (brief case model)

?

* *

*

*

Platform

Legend: ( Unless OtherwiseAnnotated )

Has

Is

* 1 or more

? 0 or more

Interfaces

Selects

Uses

Controls

CAN

InteractiveElectronic

Technical Manual(GUI)

Feeds

Displays *

MIMOSA

DiagnosticsDatabase

Updates

DiagnosticsManager

Platform FamilyCase Library

Case BasedDiagnostics Reasoner Knowledge Fusion

Module

Feeds

UsesUses

Uses

Uses

At Platform Diagnostics SessionTopology (Legacy)


13/33

Performance Measures(How to Compare and )

Measures

Precision for Prognosis a measure of the narrowness of an interval in which the remaining lifefalls

Reliability how the system responds to individual component failures

Extensibility or Scalability how the system can be extended if new components are added

Robustness how the system tolerates uncertainty

Reuse or Portability how easy or hard it is to use this system in another problem domain

Accuracy how an agent improves true positives and true negatives as a result of

learning, self-organization, and active diagnosis Entropy a measure of how the system learns and organizes over time.

Decreasing entropy signifies increasing order in a multi-agent system,resulting in more accurate and timely diagnoses

Network Activity how much network related activity results if the framework isimplemented for distributed systems


14/33

Implementation IssuesEmbedded Distributed Diagnostic Platform (EDDP)

Hardware: Modular I/O (e.g. NIs FieldPoint System, or MAX -IO). Embedded PC (e.g. MPC - Matchbox PC of TIQIT or MAX-

PC of Strategic-Test).

Network (e.g. Ethernet, PROFIBUS, CAN).

Software: Windows CE, Linux, QNX, VxWorks, or OsX operating

systems. Embedded databases (like Polyhedra). RAD tools (like eMbedded Visual Studio of Microsoft).


15/33

A Possible Agent Node

Sensors Sensors Sensors

Distributed I/O System(FieldPoint)

Network (Ethernet, CAN, Profibus)

A Small PC(MPC, MAX-PC)

An Operator Interface(LCD Display)


16/33

PM1 PM2 PM3

Algorithm #1 * * *

Algorithm #2 * * *

Algorithm #3 * * *

Performance

Assessment Matrix:

CBM Performance Assessment

Objective: To assess the technical and economic feasibility of various

prognostic algorithms

Technical Measures:

Accuracy, Speed, Complexity, Scalability Overall Performance Measure:

w1Accuracy + w 2Complexity + w 3Speed + (w i - weighting factors)


17/33


18/33

computation timetime to failure

p d

pf

t t complexity E E

t

Overall Performance = w 1accuracy + w 2complexity + w 3cost + .

Complexity/Cost-benefit Analysis

Complexity Measure

Cost/Benefit Analysis frequency of maintenance downtime for maintenance dollar cost

etc. Overall Performance


19/33

Cost/Benefit Analysis

Establish Baseline Condition - estimate cost of breakdown or time-based preventive maintenance frommaintenance logs

A good percentage of Breakdown Maintenance costs

may be counted as CBM benefits If preventive maintenance is practiced, estimate how

many of these maintenance events may be avoided.The cost of such avoided maintenance events is countedas benefit to CBM.


20/33

Cost/Benefit Analysis (contd)

Intangible benefits - Assign severity index to impact ofBM on system operations

Estimate the projected cost of CBM, i.e. $ cost ofinstrumentation, computing, etc.

Aggregate life-cycle costs and benefits from theinformation detailed above


21/33

CINCLANTFLT Study

Question: What is the value of prognostics?

Summary of findings:(1) Notional Development and Implementation for Predictive CBM Based on

CINFCLANTFLT I&D Maintenance Cost Savings(2) Assumptions

CINCLANTFLT Annual $2.6B [FY96$] I&D Maintenance Cost Fully Integrated CBM yields 30% reduction Full Realization Occurs in 2017, S&T sunk cost included Full Implementation Costs 1% of Asset Acquisition Cost IT 21 or Equivalent in place Prior to CBM Technology

(3) Financial Factors Inflation rate: 4% Investment Hurdle Rate: 10% Technology Maintenance Cost: 10% Installed Cost

(4) Financial Metrics: 15 year 20 year NPV $337M $1,306M IRR 22% 30%


22/33

Concluding Remarks

CBM/PHM are relatively new technologies - sufficienthistorical data are not available

CBM benefits currently based on avoided costs Cost of on-board embedded diagnostics primarily

associated with computing requirements Advances in prognostic technologies (embedded

diagnostics, distributed architectures, etc.) and lowerhardware costs (sensors, computing, interfacing, etc.)

promise to bring CBM system costs within 1-2% of atypical Army platform cost


23/33

The Dynamic Case-BasedReasoning Architecture

Model-based reasoner

New case constructor

Failure driven learning

Indexing rules

Phase matching evaluator

Feature interpretation(static, dynamic, composite)

Case indexing

AS path PD path

Case retrieval

Case adaptation

Test/evaluation

Propagation evaluator

Indexing path selection

Case memoryactive inactive

Model base

Analytical Models and algorithms

Sensory data

Case similarity calculation

Remembrance calculation


24/33


25/33


26/33

)exp()( R y DC accuracy real

predictedreal

lower bound

upper bound

time pt f t

predictedreal

DC

R

Accuracy


27/33

An Example

A defective bearing with a crack causes the machine tovibrate abnormally

Vibrations can be caught with accelerometers whichtranslate mechanical movement into electrical signals

Bearing crack faults may be prognosed by examining and predicting their vibration signals


28/33

An Experimental Setup


29/33

0 100 200 300 400 500 600 700 800 900 1000-0.6

-0.4

-0.2

0

0.2

0.4Figure 1 Original signals: normal & defective

0 100 200 300 400 500 600 700 800 900 1000-4

-2

0

2

4

6

0 20 40 60 80 100 120 1400

0.05

0.1

0.15

0.2Figure 2 Spectra: good & defective

0 20 40 60 80 100 120 1400

2

4

6

8

10

Vibration Signals from a good and a defective bearing PSDs of the vibration signals

Vibration Signals Power Spectrum Densities

Bearing Vibration Data


30/33

Growth of bearing crack fault


31/33

0 20 40 60 80 1000

1

2

3

4

5

6

Time Window

P S D

0 20 40 60 80 1000

1

2

3

4

5

6

7

Time Window

P S D

Small variations are added


32/33

80 85 90 95 100 1050

1

2

3

4

5

6

Time Window

P S D

80 85 90 95 100

0

1

2

3

4

5

6

Time Window

P S D

Prediction by AR Prediction by WNN

Prediction


33/33

Table: Performances of the AR predictor and the WNN predictor

Performance Measures

TTF Error Rate

Dynamic Error

Time Dynamic

Error

Similarity Error

Output Error

Total Error

Scaling Factor

1.0 100.00 1.0 0.1 1.0 N/A

Weighting Coefficients

0.20 0.20 0.20 0.20 0.20 1.0

AR Performance

0.4275 0.5200 0.4074 0.3448 0.3200 2.0197

WNN Performance 0.1855 0.5500 0.2684 0.2857 0.3200 1.6096

Overall Performance Error:

---- 2.0197 for the AR predictor

---- 1.6096 for the WNN predictor

Thus, the WNN outperforms the AR in this case

Performance

Aplikasi Sinyal Fibrasi Untuk CMB Condition Based Maintenance

Documents