predictive maintenance & machine learning
TRANSCRIPT
![Page 1: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/1.jpg)
PREDICTIVE MAINTENANCE &
MACHINE LEARNINGA comprehensive and integrated vision – our Fleet Control Room
Daniele BottazziChief Commercial Officer – IB
![Page 2: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/2.jpg)
SUMMARY
• IB Presentation
• Introduction to new technologies
• Predictive maintenance
• Fleet Control Room – IB Vision
![Page 3: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/3.jpg)
1. IB Presentation
![Page 4: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/4.jpg)
Who we are
4
IB Company
Deployment software consultants
Software Developers
115/115
55/115
35/115
The skilled and knowledgeable staff at IB provides systems and services for Maintenance & Operations improvements, with
a special focus on technologies, methodologies and re-engineering of processes.
![Page 5: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/5.jpg)
Certifications
CONTINOUS IMPROVEMENT
IB's Quality Management System integrates perfectly into the management system of a company, and aims to the
maintain and fulfill quality requirements in compliance with all of the Labor and Environmental Regulations in force.
ISO27001
SA8000
ISO9001
On June 2017, IB has awarded the prestigious ISO
27001 certification, which attests the company's
commitment to information security at every level
of the company.
1998 as the first ISO 9001 certification
until obtaining in 2017 the new
conformity ISO 9001: 2015
Since 2004, the standard of social
responsibility & ethics, that lead
us
![Page 6: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/6.jpg)
InfoSHIPsuite
®
![Page 7: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/7.jpg)
Headquarters and IB Offices
EUROPE
ITALY16035 Rapallo (GE) - ItalyVia Cerisola, 37/2Ph: +39 0185 273088
AMERICAS
FLORIDA33327 Weston - USA2900 Glades Circle, Suite 250Ph: +1 305 998 2745
ASIA &PACIFIC
HONG KONGWanchai, Hong Kong145 Hennessy RoadHong Commercial Building, Room 28A, 23/5
WestonIB USA HQ
Seattle
StavangerNorway
HamburgGermany
Rapallo Italy
IB Europe HQ
Singapore
Taipei(Taiwan)
Hong KongIB Asia HQ
LimassolCyprus
Shanghai
![Page 8: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/8.jpg)
2. Introduction to new technologies
![Page 9: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/9.jpg)
Milestones• Advent of digital era is now• Not only marine… all industries are impacted by
this revolution (or evolution)
Slogans• Artificial Intelligence• Predictive Maintenance• Machine Learning• Digital Twin• Internet Of Things (Iot)• Etc.
Are those empty words or do theyhave any practical applicability?
Digitalization towards vessel operators
![Page 10: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/10.jpg)
Artificial Intelligence
What It IsArtificial Intelligence is a collection of advanced technologies that allow machines to sense, comprehend, act and learn.AI is adopted in several applications.
![Page 11: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/11.jpg)
![Page 12: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/12.jpg)
![Page 13: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/13.jpg)
![Page 14: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/14.jpg)
It is closer than it seems
![Page 15: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/15.jpg)
Definitions
MACHINE LEARNING (ML) is a field of artificial intelligence that uses statistical techniques to give computer systems the ability to "learn" (e.g., progressively improve performance on a specific task) from data, without being explicitly programmed but using trained data sets.
REGRESSION
CLUSTERING
CLASSIFICATION
Predicting a continuous-valued attribute associated with an object. E.g. Expected performances vs real-time performances. Support Vecting Regressors, Lasso, etc.
Identifying which category an object belongs to. With Classification, the predicted output (class) is categoric, i.e. it can assume only fewvalues. E.g. Component Failure probability: High, Medium, Low. Support VectingMachines, Random Forests, etc.
Automatic grouping of similar objects into sets. K-means, mean-shift, etc.
![Page 16: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/16.jpg)
Statistic learning
There are two different phases in automatic learning:
• Training phase: it is based on meaningful data sets; e.g. for hydrodynamic – after a drydock / propeller polishing,
for thermodynamic – after an engine overhaul; → calculation of targets
• Prediction phase: from new data sets, computer can predict thanks to the models (physical or statistical).
Two types of learning:
• Supervised Learning: The computer is presented with example inputs and their desired outputs, given by a "teacher", and the
goal is to learn a general rule that maps inputs to outputs. As special cases, the input signal can be only partially available,
or restricted to special feedback
• Unsupervised Learning: No labels are given to the learning algorithm, leaving it on its own to find structure in its input.
Unsupervised learning can be a goal in itself (discovering hidden patterns in data) or a means towards an end
![Page 17: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/17.jpg)
Why Machine Learning can help fleet operations?
Understand the conditions (considering the various variables which can impact the output) in order to take the proper
decision might be a very complicated duty.
Typical approach is based on:
ML and, more in general, computer based systems, does not replace these activities, but, thanks to their computational power, can facilitate the decision makers to process the data and understand the conditions in order to provide them more
precise/accurate/objective elements to take the proper decisions.
Taking in consideration possible constraints/obligations (economical, operational, compliance, contractual, etc.)
SOURCES
Internal discussionsPersonal knowledge
Risk ManagementInsight
Previous Experiences / ExpertiseEtc.
SOURCES
Expert Judgements (internal, external)Analysis on available data
![Page 18: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/18.jpg)
Example: is my ship consumption good?
E.g. Admiralty CoefficientWhere S stands for Speed, H for power (horsepower) and D
for Displacement
• Sea state
• Wind State
• Currents
• Hull conditions
• Aging of equipment
• Bunker quality
• Navigation behavior
• Trim
• Etc.
It is an empirical formula. It does not take in consideration for example:
Machine Learning is the only way to understand the performances in their steady conditions
![Page 19: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/19.jpg)
Mandatory steps towards Artificial Intelligence
Data Availability
Data Storage
Data Standardization and harmonization
Data Reliability →Golden Rule: garbage
in garbage out
![Page 20: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/20.jpg)
ARTIFICIAL INTELLIGENCE AND MAINTENANCE
Artificial Intelligence can be beneficial to maintenance for:
• Detect the early stages of a failure
• Predict when a failure will occur
• Calculation of equipment RULs (Remaining Useful Life)
DataLegacy Systems
PredictionsAlerts/Alarms,
What-if, Optimizations,
Simulations, etc.
DataHistorian
DataReference
Values
Raw Datafrom Equipment
DataComparisonBenchmarks
DATA!!!
![Page 21: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/21.jpg)
DATA RELIABILITY Crucial point
TECHNICIAN DATA (HUMAN CENTERED):
not homogeneous information, «poor» or
useless, not properly encoded, no cross
relations
EQUIPMENT SENSOR DATA (AUTOMATION CENTERED):
Qualitative and quantitative poor information (availability,
historian, no. of tags or sensors, lack of cross relation,
sensors failures)
THREATS
![Page 22: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/22.jpg)
DATA COLLECTOR
Vessel kit - Components
Flowmeters
Inclinometers
Automation
Navigation
Additional Sensor(e.g. BWACS, OW-ACS, etc.)
ECDIS
GPS
Gyro
Anemometer
Echo Sounder
![Page 23: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/23.jpg)
![Page 24: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/24.jpg)
![Page 25: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/25.jpg)
List of signalsList of signals
![Page 26: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/26.jpg)
AUTOMATIC DATA COLLECTION False positive mitigation– self quality check
3 levels of sensor value quality check:
Signal Missing Signal out ofExpected range
Specific controls(e.g. STW – SOG > 5 knots)--> 1 of the 2 sensors is not working properly
ALERT BOT
![Page 27: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/27.jpg)
MANUAL DATA COLLECTION → OPERATION DATA
![Page 28: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/28.jpg)
MANUAL DATA COLLECTION → OPERATION DATA
![Page 29: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/29.jpg)
MANUAL DATA COLLECTION → CMMS
All measurements values should be
stored in a db
![Page 30: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/30.jpg)
MANUAL DATA COLLECTION → CMMS
![Page 31: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/31.jpg)
MANUAL DATA COLLECTION
![Page 32: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/32.jpg)
3. Predictive Maintenance
![Page 33: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/33.jpg)
Definition
MaintenancePolicies
ImprovementMaintenance
Preventive Maintenance
InspectionsPeriodic
Maintenance
Fixed TimeFixed Cycle
(running hours, age, etc…)
Condition BasedMaintenance
PredictiveMaintenance
CorrectiveMaintenance
Deferred
Reactive
UNPLANNEDPLANNED
![Page 34: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/34.jpg)
PREDICTIVE MAINTENANCE
Predictive maintenance (PdM) techniques are designed to help determine the condition of in-service equipment in order to predict when maintenance should be performed.
![Page 35: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/35.jpg)
![Page 36: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/36.jpg)
Predictive maintenance techniques
• Vibration monitoring
• Oil Analysis
• Infrared
• Corrosion
• Ultrasonic
• Acoustic
• Maintenance Techniques (RBI, RCM, FMEA, etc.)
• Condition Monitoring in EAM/CMMS
• Predictive Software Modeling
![Page 37: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/37.jpg)
![Page 38: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/38.jpg)
Predictive maintenance
techniques evolution
From a Gartner Group Research
![Page 39: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/39.jpg)
Predictive MaintenanceMeasurement of Effectiveness
• Vibration monitoring
• Oil Analysis
• Infrared
• Corrosion
• Ultrasonic
• Acoustic
• Maintenance Techniques (RBI, RCM, FMEA, etc.)
• Condition Monitoring in EAM/CMMS
• Predictive Software Modeling
FEEDBACKS FROM FIELD
SYSTEMIC APPROACH
CROSS REFERENCE
![Page 40: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/40.jpg)
Analytics are essentials for a predictive maintenance adoption.
• DESCRIPTIVE ANALYTICS is the failure root cause detection. Diagnosis identifies the «patterns of behaviour»
• PREDICTIVE ANALYTICS permits to build a model able to forecast when something will fail.
• PRESCRIPTIVE ANALYTICS automates the actions to prevent the failure or malfunction.
MACHINE LEARNING can support all aforementioned cases.
Maintenance policy evolution – from PMS to Prescriptive
![Page 41: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/41.jpg)
4. Fleet Control Room IB vision
![Page 42: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/42.jpg)
Alarm / Alert definition
Alarm/Alert Rule Definition
Set DiagnosisHypothesis
Feedbacks from field
Analysis
• Revision of thresholds• Cleansing of Diagnosis Hypothesis• Adjust data sources (add sensors, add information, etc.)
Every cycle adds new alarms/alerts rules more sophisticated reducing the no. of diagnosis hypothesis per each alarm
Theoretically trend to 1 (Pure determinism)
MACHINE LEARNING
![Page 43: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/43.jpg)
The primary scope is to have less people in
front of monitors looking at data they don’t
really care, but ask the computer to drive
the analysis according to the conditions
and provide precise duties to crew
personnel
![Page 44: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/44.jpg)
![Page 45: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/45.jpg)
![Page 46: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/46.jpg)
![Page 47: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/47.jpg)
High level alerts ready “off the shelf"
• Sfoc
• Propulsive power (ship is requesting more power than expected)
• Propeller slip
• Consumption
• No. of dg running compared to optimum
• Trim
• Speed optimization
• Hull Degradation
• Emerging propeller (propeller is not fully submerged)
• Engine performance degradation
• Engine overhaul effectiveness
• Voyage schedule effectiveness
• Bad weather impact on ship route
• Charter contract clauses
![Page 48: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/48.jpg)
EXAMPLE – SFOC OVER THRESHOLD
SFOC DEFINITION: It stands for Specific Fuel Oil Consumption and represents one of the most important indicators for the health and efficiency of
the engine.
SFOC (g/kwh) = Mass of fuel consumed per hour / Power developed in kW
ALERT RULE: if my current SFOC is greater than the target (calculated by a Machine Learning Regressor Real Time) by
10 points for more than 20 minutes in steady conditions, then alert.
![Page 49: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/49.jpg)
ALARM RULE
![Page 50: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/50.jpg)
ALARM RULE
![Page 51: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/51.jpg)
DIAGNOSIS HYPOTHESIS DEFINITION
• Maintenance issue
• Equipment issue
• Procedure issue
ROOT CAUSE ANALYSIS
CHECK1st Level 2nd Level• Check current ship position
and activity• Fuel manual inputs• Check source data signals• Check sister ships for
common issue
PRELIMINARY CHECK• Maintenance last jobs • Maintenance planned
• Lubrication
• Turbine
• Compressor
• Cylinder head
• Missing or wrong data…
• Maintenance done on engine parts• Planned maintenance on engine
parts
• Lubrication system setup
• Contaminated lube oil
• Dirty turbine blades
• Exhaust gas leaking
• Waste gate setup
• Compressor filter cleaning
• Valve leaking• Gasket issue
• Fuel data issue
n Level• Shore: Check Last Maintenance
Reports• Shore: Check Planned Maintenance
• Ship: Check Lube System Setup• Shore: Check Lube Oil Analysis• Ship: Check Lube Oil Filters
• Shore: Check Performance• Ship: Clean Procedure
• Ship: Visual Check
• Ship: Check Parameters
• Shore: Check Last Done• Ship: Check Filter Status
• Shore: Check Performance• Ship: Visual Check
• Shore: Check Fuel Analysis
ALERT BOT
![Page 52: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/52.jpg)
DISPLAY ALERTS OVER THE MAPS
SHIP 1
![Page 53: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/53.jpg)
ALERT / ALARM CONTROL ROOM
![Page 54: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/54.jpg)
/SHIP 1 SFOC ALERT
PREVIOUS JOB DONE on 15/09/2018
NEXT PLANNED JOB on15/11/2018
RUNNING HOURS5671
47% TO OVERHAUL
DIAGNOSIS HYPOTHESISLUBE OIL ANALISYS
OK
WATER ANALISYSOK
TRENDPOWER [kW] 18156
CONS [t/h] 3.62
SCATTERSFOC [g/kWh] 200
FUEL TECH DATAManual Inputs
OK
ENGINE PERFORMANCE DIAGRAMS
SISTER SHIPS COMPARISON
SHIP ACTIVE ALERTS12
SIMILAR EVENTS on SISTER SHIPS
FEEDBACK AND ROOT CAUSE
ENGINE PARAMETERS ANALISYS
![Page 55: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/55.jpg)
PREVIOUS JOB DONE on 15/09/2018
NEXT PLANNED JOB on15/11/2018
RUNNING HOURS5671
47% TO OVERHAUL
DIAGNOSIS HYPOTHESISLUBE OIL ANALISYS
OK
WATER ANALISYSOK
TRENDPOWER [kW] 18156
CONS [t/h] 3.62
SCATTERSFOC [g/kWh] 200
FUEL TECH DATAManual Inputs
OK
ENGINE PERFORMANCE DIAGRAMS
SISTER SHIPS COMPARISON
SHIP ACTIVE ALERTS12
SIMILAR EVENTS on SISTER SHIPS
FEEDBACK AND ROOT CAUSE
ENGINE PARAMETERS ANALISYS
![Page 56: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/56.jpg)
![Page 57: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/57.jpg)
/SHIP 1 SFOC ALERT
PREVIOUS JOB DONE on 15/09/2018
NEXT PLANNED JOB on15/11/2018
RUNNING HOURS5671
47% TO OVERHAUL
SIMILAR EVENTS on SISTER SHIPS
LUBE OIL ANALISYS OK
WATER ANALISYSOK
TRENDPOWER [kW] 18156
CONS [t/h] 3.62
SCATTERSFOC [g/kWh] 200
FUEL TECH DATAManual Inputs
OK
ENGINE PERFORMANCE DIAGRAMS
SISTER SHIPS COMPARISON
SHIP ACTIVE ALERTS12
DIAGNOSIS HYPOTHESIS
FEEDBACK AND ROOT CAUSE
ENGINE PARAMETERS ANALISYS
![Page 58: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/58.jpg)
![Page 59: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/59.jpg)
/SHIP 1 SFOC ALERT
PREVIOUS JOB DONE on 15/09/2018
NEXT PLANNED JOB on15/11/2018
RUNNING HOURS5671
47% TO OVERHAUL
DIAGNOSIS HYPOTHESISLUBE OIL ANALISYS
OK
WATER ANALISYSOK
TRENDPOWER [kW] 18156
CONS [t/h] 3.62
SCATTERSFOC [g/kWh] 200
FUEL TECH DATAManual Inputs
OK
ENGINE PERFORMANCE DIAGRAMS
SISTER SHIPS COMPARISON
SHIP ACTIVE ALERTS12
SIMILAR EVENTS on SISTER SHIPS
FEEDBACK AND ROOT CAUSE
ENGINE PARAMETERS ANALISYS
![Page 60: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/60.jpg)
![Page 61: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/61.jpg)
/SHIP 1 SFOC ALERT
PREVIOUS JOB DONE on 15/09/2018
NEXT PLANNED JOB on15/11/2018
RUNNING HOURS5671
47% TO OVERHAUL
DIAGNOSIS HYPOTHESISLUBE OIL ANALISYS
OK
WATER ANALISYSOK
TRENDPOWER [kW] 18156
CONS [t/h] 3.62
SCATTERSFOC [g/kWh] 200
FUEL TECH DATAManual Inputs
OK
ENGINE PERFORMANCE DIAGRAMS
SISTER SHIPS COMPARISON
SHIP ACTIVE ALERTS12
SIMILAR EVENTS on SISTER SHIPS
FEEDBACK AND ROOT CAUSE
ENGINE PARAMETERS ANALISYS
![Page 62: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/62.jpg)
![Page 63: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/63.jpg)
/SHIP 1 SFOC ALERT
PREVIOUS JOB DONE on 15/09/2018
NEXT PLANNED JOB on15/11/2018
RUNNING HOURS5671
47% TO OVERHAUL
DIAGNOSIS HYPOTHESISLUBE OIL ANALISYS
OK
WATER ANALISYSOK
TRENDPOWER [kW] 18156
CONS [t/h] 3.62
SCATTERSFOC [g/kWh] 200
FUEL TECH DATAManual Inputs
OK
ENGINE PERFORMANCE DIAGRAMS
SISTER SHIPS COMPARISON
SHIP ACTIVE ALERTS12
SIMILAR EVENTS on SISTER SHIPS
FEEDBACK AND ROOT CAUSE
ENGINE PARAMETERS ANALISYS
![Page 64: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/64.jpg)
• Maintenance issue
• Equipment issue
• Procedure issue
CHECK1st Level 2nd Level• Check current ship position
and activity• Fuel manual inputs• Check source data signals• Check sister ships for
common issue
PRELIMINARY CHECK• Maintenance last jobs • Maintenance planned
• Lubrication
• Turbine
• Compressor
• Cylinder head
• Missing or wrong data…
• Maintenance done on engine parts• Planned maintenance on engine
parts
• Lubrication system setup
• Contaminated lube oil
• Dirty turbine blades
• Exhaust gas leaking
• Waste gate setup
• Compressor filter cleaning
• Valve leaking• Gasket issue
• Fuel data issue
n Level• Shore: Check Last Maintenance
Reports• Shore: Check Planned Maintenance
• Ship: Check Lube System Setup• Shore: Check Lube Oil Analisys• Ship: Check Lube Oil Filters
• Shore: Check Performance• Ship: Clean Procedure
• Ship: Visual Check
• Ship: Check Parameters
• Shore: Check Last Done• Ship: Check Filter Status
• Shore: Check Performance• Ship: Visual Check
• Shore: Check Fuel Analisys
CHECK THE ROOT CAUSE ANALYSIS / FAULT TREE TO DETECT POSSIBLE INSIGHTS
![Page 65: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/65.jpg)
/SHIP 1 SFOC ALERT
PREVIOUS JOB DONE on 15/09/2018
NEXT PLANNED JOB on15/11/2018
RUNNING HOURS5671
47% TO OVERHAUL
DIAGNOSIS HYPOTHESISLUBE OIL ANALISYS
OK
WATER ANALISYSOK
TRENDPOWER [kW] 18156
CONS [t/h] 3.62
SCATTERSFOC [g/kWh] 200
FUEL TECH DATAManual Inputs
OK
ENGINE PERFORMANCE DIAGRAMS
SISTER SHIPS COMPARISON
SHIP ACTIVE ALERTS12
SIMILAR EVENTS on SISTER SHIPS
FEEDBACK AND ROOT CAUSE
ENGINE PARAMETERS ANALISYS
![Page 66: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/66.jpg)
/SHIP 1 SFOC ALERT
OK
![Page 67: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/67.jpg)
/SHIP 1 SFOC ALERT
HIGH PRESSURE
![Page 68: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/68.jpg)
/SHIP 1 SFOC ALERT
LATE COMBUSTION
(WITHIN PARAMETERS)
![Page 69: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/69.jpg)
/SHIP 1 SFOC ALERT
MULTIPLE INJECTION
ISSUE
![Page 70: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/70.jpg)
/SHIP 1 SFOC ALERT
PREVIOUS JOB DONE on 15/09/2018
NEXT PLANNED JOB on15/11/2018
RUNNING HOURS5671
47% TO OVERHAUL
DIAGNOSIS HYPOTHESISLUBE OIL ANALISYS
OK
WATER ANALISYSOK
TRENDPOWER [kW] 18156
CONS [t/h] 3.62
SCATTERSFOC [g/kWh] 200
FUEL TECH DATAManual Inputs
OK
ENGINE PERFORMANCE DIAGRAMS
SISTER SHIPS COMPARISON
SHIP ACTIVE ALERTS12
SIMILAR EVENTS on SISTER SHIPS
FEEDBACK AND ROOT CAUSE
ENGINE PARAMETERS ANALISYS
![Page 71: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/71.jpg)
1. The office defines the checks needed by the
crew on board
2. Streamlined activation of check internal
procedures
3. Integrated mobile tool available to crews
4. The crew perform the checks and gives
feedback to the office
5. The root cause is identified, recorded into the
system and available to detect similar events
OFFICE SHIP
ASK for CHECKS
FEEDBACK
![Page 72: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/72.jpg)
Approach shall be multi skill
DOMAIN EXPERTSENGINEERINGPHISICAL MODELS
MACHINE LEARNING DATA SCIENTISTS
ON FIELD FEEDBACKSHUMAN EXPERTISE
![Page 73: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/73.jpg)
Key words which have influenced our development road map
• Independency from equipment manufacturers, class societies, in general big operators which technology represents not a core business
• Cloud vs on premises. No marriages that force customers to be linked to 3rd parties.
• 1 remote control for multiple components (performance, energy, procurement, logistics, maintenance, compliance, operations, etc.)
• Smooth adoption - Step by step introduction (no all in)
• Usability both on new and existing ships
• Scalability according to client size in terms of functionalities and hardware architecture. Let everyone access new technologies
• Enhance the analysis with the utilization
• Data collector designed to be open to acquire any kind of signal (digital, analog) with plug & play connectors for the market standards
(NEMEA, Modbus, OPC, etc.) and/or proprietary standards (Kongsberg, ABB, etc.)
![Page 74: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/74.jpg)
Stock
Logistics
Procurement
Budget & Cost Control
Event & Incident mngt
Risk Assessment Control
Documentation (designs, 3d Models, Monographies, etc.)
Analytics
Route / Speed Optimization
Predictive Maintenance
Oil Record Books (Logbooks)
Bunker
Consumptions
Vessel Positioning (past & forecast routes)
Trim Optim
Meteo
Voyage Reports
PMS
Fleet Control Room
![Page 75: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/75.jpg)
ALARMS / ALERTS
OPTIMIZATION TOOLS (ON BOARD & OFFICE)
ANALYTICS
REPORTING
WHAT IF ANALYSIS
ANALYSIS OF EFFECTIVENESS
ELECTRONIC OFFICIAL DOCs
MONITORING & PROACTIVE DECISION
SUPPORT SYSTEM
![Page 76: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/76.jpg)
Enable customers to build their data connections and rules
“Data connections and the added
value they bring represent the future
client treasury”
Expertise + new technologies
• Raw or elaborated data acquisition
• Data processing
• Machine learning
• Data mining
• Artificial intelligence
![Page 77: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/77.jpg)
Conclusions –Attention points
• Holistic approach – no data silos but data correlations -differential diagnosis
• Mixed utilization of data driven and experience driven approach
• Adoption of robust systems – all data sources concur to knowledge management and decision making with harmonized and meaningful data
• Artificial intelligence / machine learning smart introduction
• Attention to provide the right tools to various operators / decision makers (crew, tech ops, management)
![Page 78: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/78.jpg)
Why IB
Wide coverage of data domain: from performance data to maintenance / administrative data
Source of data: from automatic collection (Modbus, opc, nemea, etc.) to transactional / manual inputs (database) info
INFOSHIP Software Suite: Availability of «ready to use» specific software modules (PMS, Procurement, Energy, etc.)
Alarm rule definition tool and fleet control room
No limitations for its applicability to some specific equipment / manufacturer, but access to all plants/vendors
Super powerfull Analytics tool (big data!!)
Hardware and software independency
![Page 79: PREDICTIVE MAINTENANCE & MACHINE LEARNING](https://reader033.vdocuments.site/reader033/viewer/2022061100/629aa9f2b02be226bc6d4f0b/html5/thumbnails/79.jpg)
“The best way to predict your future is to create it.”
― Abraham Lincoln