tomoya yamauchi munehiko minoura
TRANSCRIPT
2013 International Research Exchange Meeting of Ship and Ocean Engineering (SOE) in Osaka
December, 20th, 2013
Osaka University Division of Global Architecture Dept. of Naval Architecture & Ocean Engineering Kashiwagi Laboratory Second year master's student
Tomoya Yamauchi Munehiko Minoura
Background (1)
In the recent international marine traffic... the concerns to the environmental issues and the economic efficiency have been raised.
To acquire the sea conditions ships encounter is important.
・ Sailing safety and effectively
・ Estimating and assessing performances of the ship
■ Using onboard monitoring data 【Stochastic and statistical method】
Estimating and Assessing methods of the sea condition in the real sea
Kinematical inverse analysis method
Radar image analysis method
Wave prediction and hindcasting
Each method has each defect,
hence these methods are combined
to use practically as possible.
Try to estimate the sea conditions using onboard monitoring data measured on the ship. ship speed, motion, structure response...
■ Previous research (Minoura, Koike)
■ The proposal of the stochastic and statistical model for onboard
monitoring analysis and practical use for the performance estimation.
■ The correlation between sea conditions and ship responses are indicated
using a conditional probability density function(CPDF), and the density
function is identified utilizing the independent component analysis(ICA).
■ The estimations of the power curve and sensitivity of performances in
the real sea using onboard monitoring data (sea conditions and ship
responses).
■ The new proposal ■ The method of estimating sea conditions from onboard monitoring data measured on a ship. (only ship response data)
Background (2)
(1) The correlation between ship responses and sea conditions are acquired statistically from onboard monitoring data.
(2) Utilizing the acquired correlation, the sea conditions are estimated from only ship response onboard monitoring data.
(3) We need some data sets of ship responses and sea conditions to identify the correlation. These data sets are given through some different ways.
【 Analysis outline 】
Sea conditions are estimated statistically from monitoring data. We do not use kinematical calculations.
All measured data are utilized except outliers due to troubles of monitoring systems. We do not choose specific conditions.
Condition 1 U = 25kts, 𝑁𝑁 = 98rpm, 𝑑𝑑 = 11𝑚𝑚,𝑉𝑉𝑎𝑎 = 0,𝑉𝑉𝑐𝑐 = 0
Wave height = 𝐸𝐸 𝐻𝐻1/3|𝑈𝑈,𝑁𝑁, 𝑑𝑑,𝑉𝑉𝑎𝑎 ,𝑉𝑉𝑐𝑐 = 1𝑚𝑚
Identification of CPDF
CPDF of a wave height:
Ship speed, Engine motion, Sea condition...
Onboard monitoring
data
【Analysis flow】
Statistical model for analysis
Analysis flow
),,,,( 3/1 cVVdNUHp a
Condition 2 U = 15kts, 𝑁𝑁 = 60rpm, 𝑑𝑑 = 11𝑚𝑚,𝑉𝑉𝑎𝑎 = 10𝑚𝑚/𝑠𝑠,𝑉𝑉𝑐𝑐 = 5𝑚𝑚/𝑠𝑠
Wave height = 𝐸𝐸 𝐻𝐻1/3|𝑈𝑈,𝑁𝑁, 𝑑𝑑,𝑉𝑉𝑎𝑎 ,𝑉𝑉𝑐𝑐 = 4𝑚𝑚
Condition 3 U = 5kts, 𝑁𝑁 = 40rpm, 𝑑𝑑 = 11𝑚𝑚,𝑉𝑉𝑎𝑎 = 15𝑚𝑚/𝑠𝑠,𝑉𝑉𝑐𝑐 = 10𝑚𝑚/𝑠𝑠
Wave height = 𝐸𝐸 𝐻𝐻1/3|𝑈𝑈,𝑁𝑁, 𝑑𝑑,𝑉𝑉𝑎𝑎 ,𝑉𝑉𝑐𝑐 = 7𝑚𝑚
Statistical model for analysis
Short term ship response data ■ Measured onboard monitoring data
【Wave height (m)】
【Ship speed (knot)】
【Sig. of pitch motion (degree)】
【Sig. of heave motion (m)】
Identify a conditional probability density function of the sea condition under some ship response conditions.
Data have correlations each other, hence we can’t identify it directly.
06
12
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Wave height(m)
01530
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Ship speed(knot)
024
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Pitch motion(degree)
0
3
6
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Heave motion (m)
Onboard monitoring data z Wave height ( ) ・Ship speed( ) Engine speed ( ) ・Pitch ( ) Roll ( ) ・Yaw ( )
【Correlation】
ICA
z = Ws
【No correlation】
Independent component s
s1
s2
s3
s4 s5
s6
r1(s1)
r2(s2)
【 CPDF of the sea condition under the ship response conditions 】
∫∞
∞−
− =nnn
nnnnz
dzsrsrsrsrsrsrzzzzp
)()()()()()(),,,(
2211
2211121
1z 2z
2z
1z3z
3z
4z
4z
5z
5z
6z
6z
Statistical model for analysis
Independent component analysis (ICA)
Joint probability density function Conditional probability density function
𝑓𝑓𝑌𝑌|𝑋𝑋 𝑦𝑦 𝑎𝑎 =𝑓𝑓𝑌𝑌𝑋𝑋 𝑦𝑦,𝑎𝑎𝑓𝑓𝑋𝑋 𝑎𝑎
𝑓𝑓𝑋𝑋 𝑥𝑥
𝑓𝑓𝑋𝑋 𝑎𝑎
𝑥𝑥 = 𝑎𝑎
𝑓𝑓𝑌𝑌𝑋𝑋 𝑦𝑦, 𝑎𝑎
𝑓𝑓𝑌𝑌𝑋𝑋 𝑦𝑦, 𝑥𝑥
Area= 𝑓𝑓𝑋𝑋 𝑎𝑎
𝑥𝑥
𝑦𝑦
𝑓𝑓
𝑦𝑦
𝑓𝑓
Normalization of the area = 1
Statistical model for analysis
Conditional probability density function (CPDF)
1x2x3x4x5x
➣ Wave height ➣ Ship speed ➣ Pitch motion ➣ Roll motion ➣ Yaw motion...
z 【Variable transformations】
Onboard monitoring data
Z is decomposed into independent components s
z = Ws ICA
)()()(
221
nz
znz zzp
pzzzp
z
=
∫∞
∞−
=1)(
)(
dzp
p
z
z
z
z
∫∞
∞−
=1)(
)(
dzr
r
s
s
【 CPDF of the sea condition 】
⇒ ⇒ ⇒ ⇒ ⇒
Statistical model for analysis
Statistical model of independent components
-20 0 200
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
s1
Prob
. den
sity
-20 0 200
0.2
0.4
0.6
0.8
1
1.2
1.4
s2
Prob
. den
sity
-20 0 200
0.5
1
1.5
2
2.5
3
s3
Prob
. den
sity
【Joint probability density function
)()()()( 2211 nn srsrsrr =sof independent components】
Statistical estimations using onboard analysis
Estimation of sea conditions
■ Onboard monitoring data in the real sea (In this estimation, simulation data are substituted)
➣ Calculated by RIOS (The Research Initiative on Oceangoing Ships) ➣ The container ship sailing between Japan and North America (LPP×B×D=230.0m×32.2m×19.0m) ➣ Time series data measured every one hour for two years ➣ Consideration of conscious ship speed decreasing
120oE 160oE 160oW 120oW
0o
15oN
30oN
45oN
60oN
0
1
2
3
4
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
pitch (deg)
Time (h)
Using these data group for conditions
CPDF is identified by these data group
・ Wave height (m) ・ Wave period (s) ・ Wave direction (deg)
・ Ship speed (knot) ・ Surge motion (m) ・ Sway motion (m)
Subjects of estimation Condition data
・ Heave motion (m) ・ Pitch motion (deg) ・ Roll motion (deg) ・ Yaw motion (deg)
Wave height (m)
Ship speed (knot)
Surge (m)
Sway (m)
Heave (m)
Pitch (degree)
Roll (degree)
Yaw (degree)
Conditional components
■ The estimation of the wave height
1x
4x
5x
3x
6x
2x
8x7x
Comparison with simulation data
Simulation value (m)
Estimation value (m)
0
10
20
30
40
50
60
0 0.5 1 1.5 2 2.5 3 3.5
Error
(%)
Error distribution
00.5
11.5
22.5
33.5
0 0.5 1 1.5 2 2.5 3 3.5
Statistical estimations using onboard analysis
Estimation of wave height (1)
0 100 200 300 400 500 600 700 800 9000
1
2
3
4
Time (h)
Wave h
eig
ht
(m)
Estimation value (m)
Red dot : estimated data
Black line : simulation data
100×−
SimulationEstimationSimulation
■ The estimation of the wave height (add Current speed and direction)
Comparison with simulation data
Simulation value (m)
Estimation value (m)
Error
(%)
Error distribution
Statistical estimations using onboard analysis
Estimation of wave height (2)
Estimation value (m)
00.5
11.5
22.5
33.5
0 0.5 1 1.5 2 2.5 3 3.50
10
20
30
40
50
60
0 0.5 1 1.5 2 2.5 3 3.5
Wane height (m)
Ship speed (knot)
Current speed (m/s)
Current direction (sin)
Current direction (cos)
Surge (m)
Sway (m)
Heave (m)
Pitch (degree)
Roll (degree)
Yaw (degree)
1x
4x5x
3x
6x
2x
8x7x
9x10x11x
0 100 200 300 400 500 600 700 800 9000
1
2
3
4
Time (h)
Wave h
eig
ht
(m)
In this situation, changing conditions do not influence the estimation
➣ Ship speed and ship motions are enough for the estimation of wave height ?
➣ Wave height and current speed and direction have less correlation hence current data are not needed for the estimation ?
Statistical estimations using onboard analysis
Estimation of wave height (3)
Time (h)
■ Comparison of estimation results Wave height (m)
0
0.5
1
1.5
2
2.5
3
3.5
0 100 200 300 400 500 600 700 800 900
SimulationEstimation(add Current)Estimation
Wave height (m) Wave period (s) Wave direction (degree)
Ship speed (knot) -0.40 -0.33 -0.16
Rudder angle (degree) 0.22 0.10 0.02
Current speed (m/s) -0.19 -0.33 -0.25
surge motion (m) 0.38 0.25 0.12
sway motion (m) 0.65 0.64 0.15
heave motion (m) 0.93 0.67 0.12
roll motion (degree) 0.70 0.67 0.15
pitch motion (degree) 0.94 0.64 0.18
yaw motion (degree) 0.58 0.54 0.15
■ Correlation coefficients
Wave period (s)
Ship speed (knot)
Surge (m)
Sway (m)
Heave (m)
Pitch (degree)
Roll (degree)
Yaw (degree)
Conditional components
■ The estimation of the wave period
1x
4x
5x
3x
6x
2x
8x7x
Comparison with simulation data
Simulation value (m)
Estimation value (m)
Error
(%)
Error distribution
Statistical estimations using onboard analysis
Estimation of wave period
Estimation value (m)
5
6
7
8
9
10
11
12
5 6 7 8 9 10 11 120
10
20
30
40
50
60
5 6 7 8 9 10 11 12
0 100 200 300 400 500 600 700 800 9005
10
15
Time (h)
Wav
e pe
riod
(s)
Wave direction
(degree)
Ship speed (knot)
Surge (m)
Sway (m)
Heave (m)
Pitch (degree)
Roll (degree)
Yaw (degree)
Conditional components
■ The estimation of the wave direction
1x
4x
5x
3x
6x
2x
8x7x
Comparison with simulation data
Simulation value (m)
Estimation value (m)
Error
(%)
Error distribution
Statistical estimations using onboard analysis
Estimation of wave direction
Estimation value (m)
0 100 200 300 400 500 600 700 800 9000
100
200
300
400
500
600
Time (h)
Wav
e di
rect
ion
(deg
ree)
0
100
200
300
400
500
600
0 100 200 300 400 500 6000
20
40
60
80
100
0 100 200 300 400 500
Conclusions
■ The statistical model for onboard monitoring analysis ➣ The statistical and analytical estimation method is proposed.
➣ The correlations of sea conditions and ship responses were indicated as CPDF and those functions were derived by ICA.
■ Estimations of sea conditions using ship response data ➣ The CPDF of sea conditions were identified by ship response time
series data and wave height, wave period, and wave direction were estimated by ship speed and ship response data.
As results, wave height and wave period were estimated quantitatively. However the estimation of wave direction should be improved. For future works, we will try to estimate sea conditions using real onboard monitoring data measured on a ship.