Post on 14-Aug-2020
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
令和元年(第19回)海上技術安全研究所研究発表会
o ボンダレンコ オレクシー環境・動力系
船舶推進プラントとして主機デジタルツインの開発
福田 哲吾環境・動力系
北川 泰士流体性能評価系
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
2Contents
1. Introduction
2. Digital Twin for Anomaly Detection
3. Digital Twin for Optimization
4. Engine Modelling for Digital Twin
5. Digital Twin Model Performance
6. Tracking Filter for Parameters Adaptation
7. Conclusions
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
3Introduction: 4th Industrial Revolution
1st Industrial Revolution
Mechanization, SteamPower
~1800 ~1900 ~1980 ~20..
2nd Industrial Revolution
Electricity, Diesel Engine
3rd Industrial Revolution
Automation, Computers& Electronics
4th Industrial Revolution
Cyber-physical systems,Artificial Intelligence
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
4Introduction: Digital Twin Concept
Digital Twin is the combination of real-time data
and a mapping of system behavior which
continuously coexists with its physical counterpart
Global Digital Mapping
Model
States
Predicted States
Actual States
Set of Models
Future
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
5Introduction: Digital Twin Ecosystem
IoT, BigData
Smart Maintenance
(CBM)
Energy Management
Propulsion Optimization
Smart Sensors
Smart Navigation
Local Digital Mapping
Digital Twin
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
6Introduction: Digital Twin Focus
Anomaly Detection
Transient Response
EfficiencyOptimization
AgingPrediction
In-DepthMonitoring
*Details are presented on PS-6 at the Poster Session
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
7Anomaly Detection
A
I, M
ach
ine
Lea
rnin
g
�′��, �′��, �′��,⋯
���, ���, ���, ⋯
{ Actual State }
��,� ��,� ��,� ��,���,���,���,���,�
1 0 0 10 0 1 01 1 0 01 0 1 0
Features Map
Anomaly Classification
Classification
{ Diagnostic Parameters }
Cyber Space - Engine Model
{ Reference State }
Real Space - Engine Monitoring
Digital Twin
(Aging Trend)
Intelligent Monitoring
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
8Optimization: Operation Efficiency
Digital Twin Instance
Performance Indicators
Manual Parameters Adjustment
Simulation
Auto-Tune/Optimization
Digital Twin
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
9Optimization: Cyber-Physical System
Design / Operation Update and Optimization
MarineDieselEngineSimulator
Propulsion System Responses
Running model ship
Engine ModelDigital Twin Instance
Digital Twin
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
10Engine Modelling for Digital Twin
Cycle-Mean Value (CMV)
Engine Model
Filling-Emptying (Phenomenological)
Engine Model
The engine is a series connection of throttles
Continuous flow of air and exh. gases through the throttles
Combustion is considered as a cycle-averaged enthalpy increment
The energy and mass conservation laws areresolved at every degree of the crank-angle
Combustion is represented by thephenomenological model of a heat releaserate (Wiebe function)
The engine is a set of intermittentlyconnected control volumes.
Fast Execution and Limited Information
Slow Execution and In-depth Information
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
11
Combined-CMV Model is a Good Candidate for the Digital Twin
Fast Execution of CMV Model
In-depth Detailingof Combustion Model
Continuous flow of air andexh. gases through thethrottles
Precise calculation of combustionfor closed part of the cycle (fromEVC to EVO)
Combined-CMV Model
Engine Modelling for Digital Twin
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
12Digital Twin Model Performance
2.0
2.5
3.0
3.5
4.0
4.5
5.0
4.8
6.1
7.4
8.7
10
11.3
12.6
30 50 70 90 110S
cav.
Air
Pre
ss.,
bar
A
Air
Mas
s F
low
, k
g/s
Load, %
Experiment
Simulation
100
200
300
400
500
600
30 50 70 90 110
Exh
. G
as T
emp
, deg
C
Load, %
0.94
0.96
0.98
1
1.02
1.04
1.06
30 50 70 90 110
Sp
ecif
ic F
uel
Con
sum
pti
on (
sfc)
, r.u
Load, %
0
0.2
0.4
0.6
0.8
1
1.2
320 340 360 380 400 420
In c
ylin
der
pre
ssu
re
Crank Angle, deg
ExperimentSimulation
Steady-state performance of
the prototype engine is
mapped fairly-well.100%
75%
50%
40%
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
13Digital Twin Model Performance
Transient performance of the
prototype engine is mapped
reasonably well, however there
is still a room for improvement.
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
14Engine Modelling for Digital Twin
Engine Decomposition
Unit Model
Parameters
In Out
Unit Model
Every unit model is composed of aset of algebraic equations and a setof constant parameters. The lasthas to be adapted to a particularengine and condition.
System Analysis
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
15Concept of the Tracking Filter
Monitored Data from Real Space
http://www.professionalmariner.com/April-2013/
Predicted Data from Cyber Space
{ Parameters Update }
Tracking Filter
The tracking filter is used to match continuously
the digital twin model with measured sensor
data from the real propulsion plant.
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
16Example of the Tracking Filter
�� = �� + ��� �� − ��
Air Cooler Unit Model:
��� = � � − Function of mass flow
The Kalman Filter for Coefficients adaptation:
��� = �� + ������� , ∴ ���� ≡ ���
������
= 0, ∴ �� �� = ��
��� = ����� + � ��� − ��
�, ∴ �� = ��, ��
�
H – is the Kalman Filter Gain
��� − measured temperature
���− predicted temperature
��, �� ��
���
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
17Conclusions
The Digital Twin is the emerging technology in demand for the 4th Industrial
Revolution. The concept of Digital Twin provides in-depth information for better
decision making and also bring broad possibilities to the ship operation enhancement.
Concept of Digital Twin Engine Model Cyber-Physical System
Collaborative Research
Shipping Company
• Digital Twin
• Operation Efficiency
Engine Maker
• Engine Model
• Intelligent Monitoring
Control System Maker
• Intelligent Monitoring
• Operation Efficiency
国立研究開発法人 海上・港湾・航空技術研究所 海上技術安全研究所
18