Steven Y. Liang (梁越昇), Ph.D. Morris M. Bryan Professor of Advanced Manufacturing Systems George W. Woodruff School of Mechanical Engineering Georgia Institute of Technology U.S.A. (美國喬治亞理工學院 機械工程學院 先進制造系统研究講座教授) Phone: +1-404-894-8164, Fax: +1-404-894-9342 <[email protected]> Intelligent Machinery and Equipment Technology (智慧化機械設備技術)
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Microsoft PowerPoint - LiangSlidesMETS2016 [Compatibility
Mode]Steven Y. Liang (), Ph.D. Morris M. Bryan Professor of
Advanced Manufacturing Systems
George W. Woodruff School of Mechanical Engineering Georgia
Institute of Technology
U.S.A. (
)
()
(Woodruff Faculty Fellow, Georgia Tech)
(1605)
(Manufacturing Engineering Division, ASME)
(Robert B. Douglas Young Manufacturing Engineer Award, SME)
(Ralph R. Teetor Educational Award, SAE)
(Blackall Machine Tool and Gage Award, ASME)
(Milton C. Shaw Manufacturing Research Medal, ASME)
(ASME)
(SME)
• Initiative in 1989 by Hiroyuki Yoshikawa – “The intelligent
manufacturing system takes
intellectual activities in manufacturing and uses them to better
harmonize human beings and intelligent machines. Integrating the
entire corporation, from marketing through design, production and
distribution, in a flexible manner, which improves
productivity.“
(“ , ”-- 1989)
Intelligent Manufacturing Systems (IMS) ()
4
• James S. Albus defined intelligence as the ability of a system to
act appropriately in an uncertain environment to increase the
probability of success given the criteria of success. ( )
• Nam P. Suh recognized intelligence to be the ability of
manufacturing system to reconfigure the production system,
including machines, purchasing, inventory control, and factory
layout, in response to changing market demands for various types of
products. (
)
The use of computational data analysis, machine learning, scenario
simulation, future prediction, and system optimization (
) in support of decision making and manufacturing execution so as
to extend human capabilities in:
• coping with complexities ()
• coping with uncertainties ()
• coping with limited corporate memory ()
• coping with urgencies ()
Up-to-date Connotation of IMS
– To cope with complexities ()
– To derive information from big data ()
– To accumulate corporate experience ()
– To achieve rapid decisions ()
Strategies to Win ()
IMS works for continuous production, but not discrete. (
,) No. The higher complexity of discrete type production can
benefit more from IMS.(,IMS)
IMS can only follow human coding but it cannot outperform human.
(IMS ,) No. IMS often performs human by virtue of
self-leaning.(IMS
,)
IMS mostly refers to the enhancement of automation hardware. (IMS )
No. IMS improves the utilization of hardware via intelligent
software. (IMS ,)
IMS works once it is set up, but that takes forever. IMS) No. IMS
builds on self-learning that takes little effort to set
up.(IMS
,)
B’
Predictive Modeling Module
Parameters
12
routines
(Quality, Cost, Delivery, Other KPI’s)
(Raw Materials, Machines, Scheduling, Process Parameters)
13
14
-
-
-
– (Gurney, 1997 and Mehrotra et al. 1997) • Fuzzy logic()
– (Klir and Yuan 1995) • Genetic algorithms()
– (Goldberg, 1989 and Nissen and Biethahn, 1995) • Case-based
reasoning()
– (Watson and Marir, 1994, Kolodner, 1993)
Scientific Pillars for Intelligent Systems
16
-
-
-
Tool wear, chatter, runout, deflection, thermal expansion,
…...
TablePart Program (Feeds, speeds, positions)
Interpolator Up/Down Counter
,,
22
Part Specification
Interpolator Up/Down Counter
Table
23
24
nominal DOC
machined surface
optimal trajectory
G-ratio
power
workpiece
grinding wheel, motor and tachometer
0 5 1 0 1 5 2 0 2 5 3 0 0
5 0 0
Po w
er (W
at ts
)
0 5 1 0 1 5 2 0 2 5 3 0 0
1
2
3
4
5
Ve lo
ci ty
(m m
/s ec
0 5 10 15 20 0
500
1000
1500
2000
1
2
3
4
5
dyno table
Feedrate = 350 mm/s Depth of cut = 200 mm Cutting angular speed =
22 rad/s
100m
10m
Grinding Gap Control ()
- Reduce cycle time () - Reduce cost () - Avoid crash ()
- Increase grinding speed () - Reduce wheel wear () - Monitor part
precision ()
grinding wheel ()
- Reducing dressing strokes () - Improve wheel quality () -
Minimize welts on parts ()
Intelligent High Performance Grinding ()
Adaptive Noncircular Grinding ()
- Increase part precision () - increase process agility () - Reduce
cycle time ()
- Shorten air grinding time () - Compensate wheel diameter () -
Ensure reliable contour tracing ()
Intelligent High Performance Grinding ()
-
-
-
Rough Grinding Cycle: • Workpiece speed (m/s) (Continuous) •
Crossfeed (mm) (Continuous) • Dressing depth (um) (Continuous) •
Depth of cut (mm) (Continuous) • The number of grinding passes
(Integer) • The number of workpieces between dressing (Integer)
Finish Grinding Cycle: • Workpiece speed (m/s) (Continuous) •
Crossfeed (mm) (Continuous) • Dressing depth (um) (Continuous) •
Depth of cut (mm) (Continuous) • The number of workpieces between
dressing (Integer)
wv
33
Population Initialization
Population
Cost per Part ($)
Generation
0 20 40 60 80 100
Generation
Generation
(m/s) 0.194 (m/s) 0.3 (mm) 2.3 (mm) 2.3
Roughing (um) 50.6 Finishing (um) 15.8 (mm) 0.033 (mm) 0.01 3
31
37
Intelligent Maintenance System ()
• Based on machinery data to accurately predict and prevent
potential failures in the future. ( ,)
• Difficult to find the balance between avoiding risks and
maximizing utilization, without intelligent maintenance system. (,
)
• Intelligent maintenance system supervises maintenance decisions
based on machine conditions. (,,)
de te
rio ra
tio n
damage tolerance
increase of utilization w/ intelligent maintenance
reduced risks w/ intelligent maintenance
time
diagnostic error
prognostic error
depreciation 36%
Signal Processing Module
Signal Representation Module
Signal Processing Module
Signal Representation Module
HFRT Processed Signal
error noise
-1
0
1
2
3
4
5
Def ect Widt h/ 100 Peak_cep*1000
Defect Width vs AE 1st Peak
-5 0 5
Be a r i ng Numbe r
Widt h/ 10 1st Peak
Training and Prediction Results
Bearing Number
C al
cu la
te d
an d
D es
ire d
R es
ul ts
(u m
Trend Analysis Intelligent Modeling
15 20 25 30 35 40 5.5
6
6.5
7
7.5
8
8.5
9
9.5
Measurement Prediction No. 1 Prediction No. 2 Prediction No. 3
Prediction No. 4
15 20 25 30 35 40 0
0.5
1
1.5
2
2.5
3
12
Measurement Prediction No. 1 Prediction No. 2 Prediction No. 3
Prediction No. 4
Running Cycle (Million)Running Cycle (Million)
D ef
Remaining Life Prediction ()
• 0.35-million cycle corresponds to 3.2 life hours • Early
prediction for making maintenance scheduling
2.6 2.8 3 3.2 3.4 3.6 3.4
3.6
3.8
4
4.2
4.4
m ill
io n
cy cl
e to
r ea
ch t
hr es
ho ld
R M
Running Time (million cycles)
• It uses computational data analysis, knowledge learning, future
prediction, and system optimization systematically, automatically,
and rapidly. ( )
• GT has successfully implemented IMS for process control, process
planning, quality/functionality control, maintenance, and
scheduling/acquisition for aerospace, automotive, machine tools,
automation, and critical component industries.
• It outperforms human engineers by offering fast, accurate, and
optimal plans otherwise hard to achieve by traditional
manufacturing paradigms. ( )
Summary of Intelligent Equipment Benefits ()
52
• IMS cannot perform well in the lack of prior production data.
(,
)
,)
• IMS cannot replace human engineer’s creativity and innovation.
(
)
53
54
Suggestions
,,;
;
;
,;
,.
55