Anomaly Detection and Damage Anomaly Detection and Damage Mitigation in Complex SystemsMitigation in Complex Systems

byAmol M Khatkhate

Pennsylvania State University

Objectives

Experimental Validation of Theoretical ResearchAnomaly detectionDamage Mitigating Control

Department of Defense (DoD) Applications Rotorcraft and Armored Vehicles Jet Engines in Aircraft

Test Apparatus

Design of Test Apparutus Resonating behavior Fatigue Crack Failure crack initiation

crack propagation catastrophic failure

Role of microscope in the experiments

Implementation of Control System Data Acquisition: Acquiring sensor data from 3 LVDT’s, 2

load cells and feeding back reference input into DAQ board Real Time Control: Establishing connection through board

under RTLinux and facilitate real time control Keyboard Interfaces: Starting and stopping of system,

increasing gain and frequency of input excitation, loading and removal of damage mitigating controller in realtime

Image Capture By Microscope: Capturing crack image of all three specimens on the test apparatus

Control System ArchitectureRTLinux Kernel communicates and controls the mechanical system through DAQ board

User Space Process communicate with kernel for sending commands and receiving data

Fifos are mechanism to facilitate communication between kernel and user process

RTLinux Kernel

Mechanical System

(hardware)

FIFOS

USER SPACE PROCESS

DAQ Board

Sensor Data

Command FIFO

Software ImplementationReal Time Plotting

CO

MM

AN

D

FIFO

DA

TA F

IFO

THREADS

Keyboard Control

HA

RD

WA

RE

(DA

Q B

OA

RD

)

SER

IAL

POR

T

Microscope

Initialise module

Create Fifo Handler

Initialize Board

Command Fifo Handler

Wait for commands from keyboard

Write into data fifo

Cleanup module

Delete Fifo Handler

Mechanical System

USE

R S

PAC

E PR

OC

ESSE

S

KERNEL SPACE

GRAPHICAL USER INTERFACE

0 5 10 15 20 250

0.05

0.1

0.15

0.2

0.25

0.3

0.35

time (minutes)

stat

eVis

itPro

bDiff

Anomaly measure for crack study

# of symbols = 16, depth = 1Wavelet basis: db10

0 5 10 15 20 250

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45Anomaly measure for crack study

time (minutes)st

ateV

isitP

robD

iff

# of symbols = 8, depth = 2Wavelet basis: db10

Experimental Results & Conclusions

Anomaly detected at this point

Depth and # of symbolsChoice of wavelet basis

Future Work Apply results from the online fatigue testing

machine to study fatigue failure under flexural loading

Design of control strategy so as to enhance life (damage mitigation control) without significance loss in performance

Implementation of DES to switch between controllers (similar to rotorcraft application)

DoD Applications: Jet Engine Propulsion

2 similar Jet engines in a twin engine aircraft

Fault/anomaly detected in the operation of this engine

Data Acquisition and Real Time Control system

DES ( design the control strategy based on fault detected )

DoD Applications: Rotorcraft/Aircraft Structures

MODEL VISUALISED AS INTERACTION BETWEEN BEAMS AND MODEL VISUALISED AS INTERACTION BETWEEN BEAMS AND MASSESMASSES