Base of The Fourth Industrial Revolution Modularization · PDF file ·...
Transcript of Base of The Fourth Industrial Revolution Modularization · PDF file ·...
SV.1
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Base of The Fourth Industrial Revolution
Modularization and the Things
Adaptive Production Technology
Example Applications
SV.2
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
The Fourth Industrial Revolution
Data relationships needs Models
The Internet of Things
Big Data, Smart …
IPV4 232 10 9 Addresses
IPV6 2128 1038 Addresses
Lot Size One
Individual Production
Tolerance and Unique
The Measurement of the World is not new Humboldt and Gauss
What’s it?
SV.3
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Software
Hardware
Intel Button PC (20nm Chip Design)
World Wide Web
W-LAN
Model Kit
The Internet of Things
Multidimensional
Input Output Signals create Big Data
1000 cores
and more
SV.4
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Information, Signal and Model
MaxN 2
Noise
N log 1 [bit/s]S
S
frequency
Information
signalvalue
timenoise
f
t
S
Max
Min
p
2
( )N d [bit]
log ( )
S
S
p ss
p s
Signal Entropy
Entropy per Seconds
Channel Capacity
MaxMax 2
Noise
C 2 log 1 [bit]S
fS
Deterministic and constant multidimensional signals or
models do not contain Information. We must identify these
only once.
Only random multidimensional signals or models contain
information. Signals must measured online and models must
be identified online.
Average values and statistical spreads or the covariance
matrixes are only first order approximations. These lead in
practice to relatively big errors between model and reality.
SV.5
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
The multi-dimensional Fourier transformation
of probability density function is equal to the
multi-dimensional empirical statistical central
moments. These can be measured!
We need Big Data, Advanced Mathematics
and at the End
Software for this.
SV.6
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
The humans must design the
Software. The computer is not
creative and not intelligent in
the sense of humans.
The software is not intelligent.
The software designer is it.
The Rule of the Humans
We must design a deterministic
and stochastic static and
dynamical multi dimensional
process and product model in
the software.
SV.8
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
FANUC
ABB
FANUC
Modularization and Things I
Drives,
Actuators
and Robots
External Sensors 6D Pose and
6D Force
Controls and
Feedback
Controls Beckhoff Siemens
KUKA FANUC
ABB
KUKA
Leica Nikon
Aikon
Linear
Drive
Rotational
Drive
Parallel
Kinematics
Serial
Kinematics
Hybrid-
Kinematics
Indoor-GPS Laser Measurement Technique
Video Metric Measurement
SV.9
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Gripper
Tool Change
Spindle
6D-Calibration
6D-Gripper and
Tool Calibration
Auxiliaries
3D/6D Marker external or/and
product integrated LED-
Points Edge
Elements Sphere
rR
Triple Mirrors Retro Reflective Marker
(wavelength selective)
Tool Change Systems Gripper
Milling
Spindle
Calibration Bodies and -auxiliary
Modularization and Things II
SV.11
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Adaptive Production Technology
Position Error
Components
(5 – 10 mm)
Position Error
Linear Axis
(5 – 20 mm at the
Process)
Position Error
Robot
(1 – 5 mm)
Position Error
Process to TCP
(0,5 – 1 mm)
Process Force Position
Influence
(0,5 – 1 mm)
Accuracy
Requirements
(0,1mm Range
from 10 to 20m
5µm/m)
SV.14
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Automated measurement, placement,
calibration of process,
part and machine models
• Adjustment, machining and assembly of CFRP large
component after technologically weighted best
fit criteria
• Tolerance management by high precision calibration,
adaptive path planning as well as actuator controlled
shape and position correction
• Optimized hierarchical fully automated calibration of
control models of serial and parallel machines as
well as robot systems
• Sensor guided adaptive CFRP production
technologies
Laser-Tracker
3D-Videometrie
kooperierende Multi-Robotersysteme
steiferBauteilbeschlag
elastischerBauteilbeschlag
3D-Messmarke
Montagebauteil
Master-Roboter
Slave-Roboter
Tool-Grabber-Sensor
STCP
SM
SG
SR
Walking-Rivet-Robot
Roboter
CAS CAD CAM Data Link I
CA := Computer aided, S := “sensoring”, D := Design, M := manufacturing
The Fourth Industrial
Revolution
We are already
started!!!!
SV.15
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Siemens
Control Level
Leading Level
Cell Level
Field
Level
Actuator
Sensor
Level
µs to ms
ms to s
seconds to minutes
minutes to days/months
CAS CAD CAM Data Link II
Shannon Sampling Theorem (high dynamic result in high sampling rates;
no lost information)
The
Internet
of
Things
High Level
Language
C++ or C#
S B2 , 5 to 10f q f q
f := frequency
B := band width
S := sampling
Hard Real
Time
SV.16
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Adaption and 0.5 to 1.5 Bill. €
costs reduction potentials (Robot production cells on linear axes versus construction of special purpose
machines require flexible problem adapted optimal linear axis and robot calibrations
or/and sensor guided robots; Otherwise special machine necessary)
Teach costs per Product 1,000€ to 30,000€ (dependent on the number of teach poses;
various processes till now not possible)
Teach costs per Robot 10,000€ to 300,000€ (10 Products per Robot)
Robot change in production line 100,000€ to 400,000€ (Reapplication of teaching Production holdup)
Amortization Potential
SV.17
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Rough Inter-polation
Inverse kin.Transform
Look-UpTable
Fine Inter-polation
PositionRegulator
MachineCoordinates
IdentifiedMachine/Robot
Parameter
Time Sequence
{ }tk
xG( )tk
xF( )tm
pRI
pRI pEI
pRI
xTCP
A( )t
Motion Model geomtric Parameter
AxisSensor (x)
Velocityand Acceleration
Definitions
pTCP
A( )t vTCP
A aTCP
A
< 10ms
Time Sequence
{ }tm
< 1ms
PowerAmplifier
Sensor(x,x,x)
Comp. AxisResidual Moments
x( )tm
x( )tm
x( )tm
FrictionCompensation
x( )tm
+ -
Interpreter
ServoControl
ElasticityModel +
-
Target Pose
pTCP
AS( )t
xS( )tm
xS( )tm
xS( )tm
Fast iterative-inverse
under consideration of
parallelism and orthogonality
errors of the axes as well as
drive elasticity. (500 high-speed compared with
classic iterative approaches Makes
10 ms possible for precision
calibration applications and realize
real time abilities !!!!)
Hierarchical iterative model
calbration orientated at the
numeric rank defects and the
sensitivity of the control model
parameters (Realize fully automatic high speed
precision calibration)
6D position invariant modeling
of the elasticity behavior
(Realize a real time compensation
of process force induced 6D
position variant positioning errors.)
Analytical identification of 6 D
transform measuring to TCP
frame (Identification 6 D-measuring
characteristics, Identification 6 D tool
correction,
Identification of cooperating multi-
robot systems, Precision model
calibration etc.)
Automation and System Integration
These Principles are General Applicable
Modeling Structure and Parameter Identification
Inverse Models / Controller Loops
Information Gain by Measuring
Integration of the procedures
to the cell level
or under
real time requirements into the field level
SV.18
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
F-Aromon
Adaptive milling
processing
Adaptive component
and section assembly
Adaptive component shape and position correction
CFK AFMO FFM
Examples
ProsihP FFM
SV.19
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Component Shape and Positional Correction
• Flexibly adjustable contour
(for several aircraft types usable)
• Economical vacuum grab systems
• 6 D force and moment controlled process
• Wrinkling could reduced
• Temperature and range of the tide compensation
• Adaption in the shape and positional correction
• Economical foundations
• Shape and position precision converges towards the machines repeatability to 2 - 4
iterations in the area of 50 to 100 µm (for the hole fuselage)
SV.20
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Hand Eye Calibration
Movement
Measure
Iterative Model Calibration
SV.22
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Joint-Controll
Network-Card
Joint-Regulator
Grob-Inter-polation{ }it
{ }kt
Bewegungs-variante G
Maschinenkoordinaten
Geschwindigkeits-und Beschleunigungs-
vorgaben
Zeitfolge
Zeitfolge
AnalytischeInverse
Fein-Inter-polation
Lage-regler
Trans-formation
TCP
A ( )itp
( )itx
F ( )ktx
I ( )tx
Steuerrechner
I/O-DA
Sensor(x,x,x)
Power-Amplifier
ExternalSensor
Realtime Gigabit Ethernet
EEWEEW
Software
Werkzeugmaschine
Maschinen-steuerung
T-Mac
Wasser-schneid-
ToolLeica-Laser-
Tracker
Fräs-Tool
T-Scan
T-Probe
6D Pose Measurement “On the Fly”
• Reduction of unproductive
measuring and auxiliary
progress times by 80%
• Standardized interfaces
• Modular hardware and
software architecture
• Adaptive flexible production
and assembly cells
• Adaptive path control
• 0.5 to 1 Bill. € per year
SV.23
PD Dr.-Ing. habil. Jörg Wollnack 18.12.2015
Calibrated Model
Non calibrated
Model
/ 100 µm
/ 1 mm
RMS x / m y / m
x / m y / m
Model Based Precision Increase
Machine
Application
C, C++, C#
Programmers Interface
ModelingSensor Vision
Motion & I/O - Interface
MotionPlanning
I / O
BlumeBlume
Sensors
OfflineProgramming
and Simulation
Sensor-Int.
I/O
Motion
Robots
OperatingSystem
KernelMode
User Mode(protected)
Hardware Interface Layer
Hardware
Filesystem
OtherDeviceDriver
NetworkDrivers
GraphicsDrivers
GraphicsSubsystem
I/O Manager
Real Time
OO Code Classes API Vision
• Structured model based instead of
unstructured (look up table) volume metric
6 D position error compensation
• Application for assembly and production
facilities, large volume machine tool as well
as industrial robots
• Significant reduction of the measuring
time by 95%
• Automatic causal detection of wearout,
defects etc..
• Accuracy gain (factor 5 to 10)
• Recorded Quality
• 0.5 to 1.0 Bill. € per year