Transcript Fault Prediction in Electrical Valves Using - LaPSI

UFRGS
Fault Prediction in Electrical
Valves Using Temporal
Kohonen Maps
Luiz F. Gonçalves, Eduardo L. Schneider, Jefferson L. Bosa, Renato Ventura B.
Henriques, Paulo M. Engel, Marcelo S. Lubaszewski
11th LATW
Punta del Leste - March 28-31 2010
OUTLINE
Introduction
 Maintenance scheme

 Mathematical
model
 Signal processing
 Temporal Kohonen maps
Experimental results
 Conclusions

2
INTRODUCTION
3
INTRODUCTION


The prediction of certain phenomena, processes or failures
(or time series prediction) is particularly interesting and
useful in many cases
It has been the subject of research in several areas:





Medicine (saving lives)
Meteorology (predicting the rain precipitation)
Engineering (increasing equipment reliability)
Economics (predicting changes in the stock market)
Main motivation: is the need to predict the future conditions
and to understand the underlying phenomena and processes of
the system under study
Building models of the system using the
knowledge and information that is available
4
INTRODUCTION


Many methods for system prediction have been
developed with very different approaches
Statistics:



Autoregressive
Autoregressive Moving Average
Neural networks:



Multi-Layer Perceptrons
Radial Basis Networks
Self-Organizing Maps (SOM)
In the last years, models based on self-organizing
maps have been raising much interest
5
INTRODUCTION


Self-organizing map algorithms perform a vector
quantization of data, leading to representatives in
each portion of the space
The temporal models, built from SOM such as:



Temporal Kohonen maps (TKM)
Merge self-organizing maps (MSOM)
Recurrent self-organizing maps (RSOM)
Use a leaky integrator
memory to preserve the
temporal context of the input
signals
6
INTRODUCTION

In this work, a proactive maintenance scheme is
proposed for fault prediction in electrical valves
Electrical valves
Model
Proactive
maintenance
scheme
Signals of torque and position
Wavelet packet
transform
Oil distribution
network
&
Temporal
Kohonen maps
Predicting the faults
7
MAINTENANCE
SCHEME
8
PROACTIVE MAINTENANCE

Recent advances in:



To automate and integrate proactive
(also know as intelligent) maintenance
tasks into embedded system
Electronics
Computing
Proactive
≠
corrective, preventive or predictive
Focuses on fault prediction
and diagnosis based on
component lifetimes and on
system on-line monitoring
That are based either on post-failure
correction or on off-line periodic
system checking
9
MAINTENANCE SCHEME
Mathematical model
Temporal Kohonen maps
Wavelet packet transform
10
MATHEMATICAL MODEL

Electrical actuator
Main components
Forces
11
MATHEMATICAL MODEL

Electrical actuator model
Differential and algebraic
equations
Position
Torque
Fault
injection
12
SIGNAL PROCESSING

Wavelet packet transform




Energy (spectral density)



Preserves timing and spectral information
Suitable for the analysis of non-stationary signals
Capable of decomposing the signal in frequency bands
Torque
Position
Divided into N frequency
bands
The energy is used by the self-organizing maps
The WPT runs in a
PC station during the
training phase
During on-line testing,
the WPT shall be part of
the embedded system
13
SELF-ORGANIZING MAPS


SOM or Kohonen maps (class of neural networks)
Unsupervised learning paradigm based on:



Competition (search the winner neuron)
Cooperation (identify direct neighbors)
Adaptation (update synaptic weights)
The goal of a SOM is,
after trained, mapping
any input data from a
Rn space representation
into R2 lattice-like matrix
Synaptic weight vector
Energy vector
14
TEMPORAL KOHONEN MAPS

The temporal Kohonen map (TKM)




Unsupervised approach for prediction derived from the SOM algorithm
Uses leaky integrators to maintain the activation history of each neuron
These neurons gradually loose their activity and are added to the
outputs of the other normal competitive units
These integrators, and consequently the decay of activation,
are modeled through the difference equation:
Temporal
Activation
Where:
Euclidean
Distance
15
TEMPORAL KOHONEN MAPS

The internal processing of SOM and TKM algorithms
can be simplified and divided in three different steps:
1. Start up
2. Training
3. Recovery

Except for the determination of the winner
neurons (recovery step), all other steps of
the TKM are the same as in the SOM
Winner neuron:


SOM: the neuron with the shortest distance
TKM: the neuron with the highest activation
16
TEMPORAL KOHONEN MAPS


For fault prediction, in recovery step, the map is colored such
that the distance between neighboring neurons can be seen
The distance is given by the difference between the synaptic
weights of neighboring neurons



Closer neurons will appear clustered in the map and will be assigned
the same color
Different colors will denote neurons under different operation
conditions: normal, degraded or faulty
Once the winner neuron is computed for a particular input
vector, E, the current system status can be identified in the
colored map and, in deviated behavior, the degradation
trajectory can be visualized in the map
17
TEMPORAL KOHONEN MAPS

In the TKM the system state can be visualized as a trajectory on
the map and it is possible to follow the dynamics of the process
This trajectory is
described based on
the winning neurons
In a normal operation
mode, the winners
ought to follow a path
inside the normal
behavior region
When a failure
occurs, the winner
will deviate from the
normal region
18
EXPERIMENTAL
RESULTS
19
EXPERIMENTAL RESULTS

Steps to generate the results:
1. Generate data (W) for normal (N), degraded (D)
and faulty (F) behavior (obtained from the model)
2. Obtain the classification map (N, D and F data)
using temporal Kohonen maps
3. Generate new N, D and F data (E) for three faults
4. Obtain the prediction map for each kind of faulty
20
EXPERIMENTAL RESULTS


A lot of simulations is performed to obtain typical
values of torque and opening position under N, D and
F valve operation to train the fault prediction map
The fault simulation is needed to generate the F and D
data (some parameters are gradually incremented)
KR simulates the
degradation of the
internal valve worm
gear, till it breaks
KM deviations simulate the
elasticity loss of the valve
spring along time
100 operation cycles
Ca deviations simulate an
increase of friction between
the valve stem and seal 21
MODEL RESULTS

Fault simulation
100
300
• •
• •
•
Torque [Nm]
•
100
50
0
0
Torque
•
•
•
•
•
•
• • • •
•
• 10• • 20
30
40
50
60
Time [s]
a)
• • •
70
•
•
60
•
•
50
•
•
40
•
•
30
•
•
20
•
10
• • • •
•
• •
0
0 • •
10 • 20
30
40
50
60
70
80
Normal
80
•
200
150
• Faulty
•
• Faulty
90
Position [%]
250
Normal
70
• •90 • 100
80
90
•
100
Time [s]
b)
Position
22
CLASSIFICATION RESULTS

Fault classification map of faults in KR, KM and Ca
23
CLASSIFICATION RESULTS

Fault classification map of faults in KR, KM and Ca
24
CLASSIFICATION RESULTS

Fault classification map of faults in KR, KM and Ca
25
CLASSIFICATION RESULTS

Fault classification map of faults in KR, KM and Ca
26
CLASSIFICATION RESULTS

Fault classification map of faults in KR, KM and Ca
Each cluster is
assigned a
different color
During the on-line
testing phase, a
winner neuron
computed for a
measured input
vector can be
easily located in
this map
27
PREDICTION RESULTS

Fault prediction map of faults in KR
28
PREDICTION RESULTS

Fault prediction map of faults in KM
29
PREDICTION RESULTS

Fault prediction map of faults in Ca
30
PREDICTION RESULTS



It can be seen in these figures, three different paths
(one for each simulated fault)
The trajectories started from neurons classified as
normal, passed through neurons classified as
degradation, and arrived to a neuron that represents
the failure
It is noteworthy that in this work, the temporal
Kohonen map is just used as a visualization tool
31
CONCLUSION
32
CONCLUSIONS



A proactive maintenance scheme is proposed for the
prediction of faults in electrical valves, used for flow
control in an oil distribution network
This is the first attempt to apply a proactive
maintenance methodology to this sort of valves
A implementation of temporal Kohonen maps is
proposed to solve the valve maintenance problem
33
CONCLUSIONS


An system implements these maps for the prediction
of faults in this valves
This technique can clearly be extended to any type of
maintenance scheme including the on-line testing of
heterogeneous chip with some kind of electromechanical systems (sensors or actuators) or other, for
example
34
CONCLUSIONS


The results obtained point out to a promising solution
for the maintenance in electrical valves
Acknowledgements
CNPq
 CAPES
 Petrobrás

35
Thank you!
[email protected]