Joint Autoencoder-Classifier Model for Malfunction Identification and Classification on Marine Diesel Engine Diagnostics Data

Abstract: There has been an increasing demand on marine transportation and traveling, since the voyage of the ships are more economical and efficient than air or land-based alternatives. The propulsion of a ship is provided by a main engine system which includes the shaft, the propellers, and other auxiliary equipment. Marine diesel engine is a complex structure that the faults within these machines can cause malfunction of the whole system, which in turn inhibits the ship’s mission. It is crucial to monitor the engine and other auxiliary systems during the operation and infer their condition from their diagnostic data. In this study, we analyze monitoring data of a crude oil tanker for different ship loads and conditions. Our primary analysis includes main engine fault detection and classification for which we propose an end-to-end joint autoencoder-classifier model that contains a convolutional autoencoder, and a long-short term memory regressor connected to the latent space. Genetic algorithms optimized models gave us 93.61% accuracy for fault classification task. Further investigation on feature’s contributions to the model, we increased the accuracy up to 96%. One concern about marine transportation is the pollution of the air with greenhouse effect gases. In this study, we have developed NOx and SOx emission estimators for different faults and working conditions. Leveraging ship load, working conditions and engine faults in the models helped us to achieve 50% better estimation performance. Although there are other studies regarding gases emissions in the literature, this is the first study that took engine faults into account. We believe that the joint autoencoder-classifier model will be useful for other time series estimation task on other domains, especially where the operating condition plays a role in the process.



https://doi.org/10.36001/phme.2022.v7i1.3335

1. 1/27
Joint Autoencoder-Classifier Model for
Malfunction Identification and Classification on
Marine Diesel Engine Diagnostics Data
Kürşat İnce
HAVELSANInc.
Naval Combat Management
Technologies Center
[email protected]​
Gazi Koçak
Istanbul Technical University
Marine Engineering
Department
[email protected]
YakupGenç
GebzeTechnical University
Computer Engineering
Department
[email protected]​

2. 2/27
Agenda
•Maintenance Strategies and Predictive Maintenance
•Ship Engine Room Simulation Dataset (MC90-V)
•Fault Classification Problem
•Joint Autoencoder-Classifier Architecture
•Emission Prediction Problem
•Experimental Results
•Conclusion

3. 3/27
Maintenance Approaches
•Maintenance: The act of keeping something in good condition by
checking or repairing it regularly. (The Oxford Dictionary)
•BreakdownMaintenance
•Allow machine/equipment until it breaks down, i.e. run to failure, then replace the
defective part(s).
•Preventive Maintenance
•Carry out the repair or replacement via manufacturer specified time intervals.
•Proactive Maintenance
•Prevention of the failure: trace all failures to root cause(s). Take proactive measures to
ensure that they do not repeat. May require design change(s).
•Predictive Maintenance
•Monitor machine/equipment condition for possible performance/quality loss. Schedule
maintenance only when a functional failure is detected.

4. 4/27
Predictive Maintenance
•Benefits
•Reduction or elimination of unscheduled equipment downtime
•Increased labor utilization
•Increased production capacity
•Reduced maintenance costs
•Increased equipment lifespan
•Problems (for data-driven models)
•Availability of system monitoring data (label/unlabeled)
•Availability of maintenance records

5. 5/27
Ship Engine Room
5

6. 6/27
MC90-V Engine Room Simulator
•K-Sim: A well-known ship engine room
simulator with high fidelity.
•ERS MAN B&W 5L90MC VLCC L11-V
•Simulates a very large crude carrier
•With a MAN B&W slow speed turbo
charged diesel engine
•The model is based on real engine data
that make the dynamic behavior of the
simulator close to real engine response.
•The simulator includes Control room
operator station and panels and bridge
and steering panels, etc.

7. 7/27
MC90-V Engine Room Simulator

8. 8/27
Ship Engine Room Simulation Dataset
•Initial Conditions
Condition Value
Ship Speed/Load Full Ahead Loaded (FAL)
Full Ahead Unloaded (FAU)
Sea Water Temperature 20
o
C,
25
o
C,
28
o
C
Sea condition (Beauf) 0
4
6

9. 9/27
Ship Engine Room Simulation Dataset – Cont.ed
•Fault Classes
•Normal condition (M000)
•Cyl 1 injection valve nozzle wear (M2503)
•Cyl 1 injection valve nozzle clogged (M2508)
•Cyl 1 piston ring stiction (M2520)

10. 10/27
Ship Engine Room Simulation Dataset – Cont.ed
•Dataset construction
•No of Fault Classes: 4
•Initial Conditions: 18
•53 runs /per fault class /per IC
•35 for training
•18 for test
•1000 to 1400 data point per run at 1 Hz

11. 11/27
Ship Engine Room Simulation Dataset – Cont.ed
VariableUnit/Range Description Type
P01600 barME air receiver press Real
T01601 degCME air receiver temp Real
G02051 ton/hME cyl 1 air flow Virtual
T02042 degCME cyl 1 air inlet temp Real
P02065 barME cyl 1 combustion press (pmax) Real
P02066 barME cyl 1 compression press (pcompr) Real
T02040 degCME cyl 1 exh outlet temp Real
T02041 degCME cyl 1 exh outlet temp deviation Real
G02050 kg/hME cyl 1 FO flow Virtual
E02056 kW ME cyl 1 indicated power (IKW) Real
P02072 barME cyl 1 injection max press (pinjm) Real
P02071 barME cyl 1 injection open press (pinjo) Real
X02074 deg ME cyl 1 length of injection (linj) Virtual
P02055 barME Cyl 1 mean effective pressure (mip) Real
G02052 ton/hME cyl 1 oil flow Virtual
T02044 degCME cyl 1 oil outlet temp (piston) Real
… … … …

12. 12/27
MC90-V Dataset Research Challenges
•Challenges related to predictive maintenance
•Fault identification and classification: Identifying the state of the
machinery, whether it is operating normal, and predicting the
fault type if it is not.
•Health index generation: Predicting the health state of the
machinery, i.e. Cyl1.
•Remaining useful life prediction (RUL): Predicting time to failure
for Cyl 1.
•Challenges related to emissions
•NOx, SOx and Smoke Content

13. 13/27
Initial Study on MC90-V Dataset
•Fault Detection with Joint Autoencoder-Classification Model
•NOx and SOx gases emission prediction
More info

14. 14/27
General Framework

15. 15/27
Fault Detection
Problem: Develop a model that will predict fault type
•M0000: Normal condition
•M2503: Cyl 1 injection valve nozzle wear
•M2508: Cyl 1 injection valve nozzle clogged
•M2520: Cyl 1 piston ring stiction
Inputs:
•Initial conditions
•Sensor values
Condition Value
Ship Speed/Load Full Ahead Loaded (FAL)
Full Ahead Unloaded (FAU)
Sea Water Temperature 20
o
C 25
o
C 28
o
C
Sea condition (Beauf) 0 4 6

16. 16/27
Joint Autoencoder-Classification Architecture
•Inputs:
•X(t): Sensor readings at time (t)
•OC(t): Initial conditions for the experiment.
•Outputs:
•෠??????(�): Predicted sensor values at time (t)
•෣??????????????????��(�): Predicted fault class of the experiment
CNN
LSTM
CNN

17. 17/27
Fault Classification ResultsTest 20% 40% 60% 80% 100%
Train ACC F1 ACC F1 ACC F1 ACC F1 ACC F1
20% 79.77 79.83 85.39 85.39 73.22 72.88 65.35 64.78 61.78 60.56
40% 78.87 79.13 90.55 90.65 90.42 90.45 85.67 85.25 82.34 81.32
60% 79.15 79.12 90.68 90.68 94.00 94.00 93.70 93.69 91.78 91.78
80% 73.98 74.01 88.37 88.58 92.51 92.63 94.48 94.55 91.14 91.12
100% 73.17 73.23 88.01 88.25 92.28 92.41 94.31 94.39 93.61 93.63

18. 18/27
Confusion Matrices (for 100% test data)
20% training data 40% training data 60% training data
80% training data Full training data

19. 19/27
Important Features
Feature Description Unit
P02072ME cyl 1 injection max press (pinjm)bar
T02044ME cyl 1 oil outlet temp (piston)degC
Z02013ME exhaust gas smoke content%
T01351Main LO temp outlet ME degC
Z01970ME exh NOx content final g/kWh
T02040ME cyl 1 exh outlet temp degC
P02071ME cyl 1 injection open press (pinjo)bar
E02056ME cyl 1 indicated power (IKW)kW
P01600ME air receiver press bar
Z00518ME exh SOx content g/kWh
P02055
ME Cyl 1 mean effective pressure
(mip)
bar
T02042ME cyl1 air inlettemp degC
P02066
ME cyl 1 compression press
(pcompr)
bar
T01601ME air receiver temp degC

20. 20/27
Gas Emission Prediction
Problem:
•As the ships run on diesel fuel, and produce several gases while
burning, their emissions becomes a global concern → Increase
greenhouse effect and global warming
•Develop a model to predict NOx and SOx emissions.
Inputs:
•Initial conditions
•Sensor values
Model: Gradient Boosting model optimized with Optuna

21. 21/27
NOx Plots

22. 22/27
NOx Prediction Results
Ship LoadALL Conditions FAL FAU
Data in use MAE RMSE MAE RMSE MAE RMSE
FULL Data 0.02130.04780.00940.03370.00920.0276
M0000 0.00210.00300.00170.00230.00180.0028
M2503 0.00790.01150.00660.00930.00560.0076
M2508 0.01640.05540.01510.05750.02080.0562
Ship LoadALL Conditions FAL FAU
Data in useMEAN STD MEAN STD MEAN STD
FULL Data 14.18541.069915.06640.502713.26220.6433
M0000 13.92471.089315.02750.020712.84960.0390
M2503 14.62471.035815.44610.522813.79890.7186
M2508 13.98180.914114.70400.459313.11760.4605

23. 23/27
SOx Plots

24. 24/27
SOx Prediction Results
Ship LoadALL Conditions FAL FAU
Data in use MAE RMSE MAE RMSE MAE RMSE
FULL Data 0.00530.01610.00310.00840.00270.0111
M0000 0.00050.00090.00040.00060.00050.0007
M2503 0.00680.00980.00530.00750.00280.0038
M2508 0.00390.01880.00300.01140.00410.0201
Ship LoadALL Conditions FAL FAU
Data in useMEAN STD MEAN STD MEAN STD
FULL Data
13.05890.405013.16460.476412.94780.2718
M0000
12.86190.058412.92080.004112.80430.0039
M2503
13.42520.502913.63870.566613.21010.3037
M2508
12.86500.102712.91790.063712.80130.1046

25. 25/27
Conclusion and Summary
•Fault Classification
•Joint autoencoder-classifier model
•Autoencoder: CNN. Effective for extracting time series features
•Classifier: LSTM. Does the actual classification
•Successfully added IC knowledge into the model.
•Gas Emission Prediction
•Gradient boosting model
•Reasonable results for NOx and SOx emissions.
•The joint autoencoder-classifier model will be useful for other
time series estimation task on other domains, especially where
the operating condition plays a role in the process.

26. 26/27
Future Plans
•The MC90-V dataset has much more initial conditions than we
have used in this study. We will be inspecting other scenarios in
the future studies.
•We also plan developing a remaining useful life prediction model
which will predict when the failure will occur in the main engine
Cyl. 1.
•We will analyze other types of gas emissions, such as smoke
content, from the engine.

27. 27/27
Thank you…
Kürşat İnce
[email protected]
[email protected]