Inauguration Is mechatronics still relevant after 45 years? Oct 2016

Transcription

1 Inauguratin Is mechatrnics still relevant after 45 years? Oct 2016

2 Mechatrnics still relevant after 45 years? Integratin f electrnics, cntrl engineering and mechanical engineering" [TetsurM 69] Science f mtin cntrl [VanBrusselH 96] The synergistic cmbinatin f mechanical engineering, electrnics, cntrl systems and cmputers, all integrated thrugh the design prcess [Craig99] Cntain elements f technlgies (mechanics&cntrl, electrnics, cmputing) and hlistic aspect f synthesis Evlutin : base technlgies cmplemented with ther sciences as they develp hlistic aspect remains key

3 Evlutin Mechatrnics 1995

4 Evlutin Mechatrnics 2016

5 Evlutin Mechatrnics Change past 20 years Observatins Still largely abut cntrl, systems, mdelling Cnstituting technlgies are mechanics & cntrl electrnics cmputing Same hlistic apprach t synthesis (and design ptimizatin) Leaves classical mechtrnics : ther technlgies dmains kick in (eg AI/ machine learning/new sensing) Cntrls refine / advance Increased fcus n systems

6 Evlutin Mechatrnics Mechatrnics changed - maybe the engineering jb as well? Deal with mre cmplex (mechatrnic) systems Apply mre science dmains at same time Lw level wrk mstly autmated by CAE (eg FEA, detail drawing, aut cding)

7 Evlutin Mechatrnics new fcus engineering jb Creative aspects : synthesis as art : Cmbining and integrating technlgy - Multiphysics hlistic perspective (helicpter view) Discvered system engineering as a tl t deal with cmplexity f ur mechatrnic systems => Mechatrnics methdlgy (gverned by system engineering) became relevant t ur new engineering jb

8 Future Mechatrnics What will it bring? Value thinking! Mechatrnics must satisfy unmet need. Bringing tgether different technlgies : Grwth happens n crssrads f technlgy [Ppper 63] Hlistic synthesis apprach

9 Future Mechatrnics What will it bring? Mechatrnics lses smewhat its histrical affinity with mechanics/ electrnics/ classical cntrl but will gain affinity with ther disciplines (as new science becmes available t technlgy) Further refinement cntrls and paradigm shift t Self-X New mechatrnic tplgies as cyber physical systems spawned by industry 4.0 need System engineering as tl managing cmplexity becmes necessity

10 Future Mechatrnics grwing affinity with ther technlgy dmains Materials - plymeres Cntstructin Aer Actuatrs Agrnmy Micrtechnlgy - miniaturizatin Chemical (prcesses) Engineering technlgy Bi and bimed Pwerdensity & perfrmance Sensrs Health Tehnlgy dmains CW AI (machine learning) NVH Energy Envirnmental Management

11 Grwing affinity with ther technlgy dmains micr and precisin Great future fr miniaturizatin f mechatrnic systems Imprtant driver - must dminate yur thinking [ReynaertsD] Mre functins per vlume > enabler Prduct cst appraches material cst fr high vlumes Miniaturizatin f mechatrnic systems hindered by lack f actuatin and sensing (nt cmputing pwer) Physics change [FeynmanR 59]: There is plenty f rm at the bttm : an invitatin t enter new physics

12 Grwing affinity with ther technlgy dmains Machine learning & AI The human brain excels in certain tasks like pattern recgnitin & cmplex mtin cntrl Recent develpments in the dmain f Machine learning and AI pen up applicatins t machines Future mechatrnic systems will lean n Pattern recgnitin fr CM/ RUL Infer mdels frm data (eg user intentins, sft mdels, SLAM) Learn and respresent knwledge New user interface

13 Grwing affinity with ther technlgy dmains New sensing and cmputing Utilize massive sensrs with data techniques t extract infrmatin n many rather than utilize multiple dedicated sensrs Visin [MIT 10] Thermgraphic Ultrasund

14 Refinement Cntrls mdel based Imprtant ingredient t mechatrnic systems Mdel based: ffline expert hard mdels & data inferred sft mdels Cntrls will deal with uncertainty Eg minimal perfrmance based n hard classical mdel based cntrllers, imprved perfrmance n sft mdel based cntrllers Learning (ILC, reinfrcement,.) Mixed/ Weighted with a priri (mdel) infrmatin >All said and dne? N, is abut infrmatin and hw yu utilize r why the shwer in yur htel rm is always mre hard t cntrl

15 Refinement Cntrls paradigm shift t self-x Paradigm shift t self-x (-cnfiguratin, -ptimizatin, -calibratin, ) Paradigm shift = Optimizer makes decisins based n bservatins/mdels f itself and its envirnment and a user specified bjective Eg MPC utilized fr reference tracking t bjective tracking Sft Mdels can be acquired during peratin Uncertaintainty can be included Eg state machines transitins

16 Industry 4.0 paradigm leading up t new tplgies fr mechatrnics machines & prcesses Industry 4.0 prmises Flexibility and mass custmizatin Shrter setup times Less variatin and waste Prducts fr/in that industry Central is data/infrmatin Distributed architecture f flexible plug and prduce mdules Cheap sensr, cmmunicatin and cmputing

17 New tplgies tplgy 4.0 prductin ecsystem

18 The new industry 4.0 paradigm prductin architecture Mdular: prductin mdules are cnnected flexible tgether Legacy machines/cmpnents thrugh cnnectrs r middle ware Additin new prcesses and sensing: additive, cbt assembly, sensr feedback: tactile, frce, camera > new technlgy dmains Architecture requires re-cnfigurability Our mechatrnic systems must be mdular - recnfigurable supprt new technlgy dmains

19 The new industry 4.0 paradigm rbust scheduling and cntrl Requirements Deal with lw vlume and high variability Fault tlerant rbustness prductin system must deal with the unexpected [HVanBrussel 2015] Recnfigurable Plant ecsystem becmes t cmplex t be cntrlled by an expert written central prgram => Self X (self-cnfiguratin/rganizatin) Eg Agent based cntrl : Observatin f self rganizing behavir in nature [Kestler the ghst in the machine ] led t distributed multi-agent based cntrl systems

20 The new industry 4.0 paradigm data t infrmatin Industry 4.0 relies n infrmatin exchange prductin system and data driven ptimizatin f prcesses But dn t fld the bus with data Machines need t turn data int useable interpretable infrmatin [Lee] Cnnectrs: lwest level Data t infrmatin (lcal machine) fr self knwledge Cyber: between machines Cgnitin: deep knwledge and representatin t human peratrs, basis fr distributed cntrl => infrmatin is key t rganize ur mechatrnic systems arund

21 New tplgies Structure (mechatrnic) cyber-physical systems Industry 4.0 hlds sme prmises based n putting data central Leads t different accents in machine tplgy Tday machines rganized hierarchically n technlgy r functinal grups Trend new machine architecture = Distributed (redundant) mechatrnic bjects gruped arund infrmatin (mimics ind 4.0 tplgy)

22 New tplgies Structure (mechatrnic) cyber-physical systems Mechatrnic bject Actuatr Sensr Objects Cmputing unit Shared within and with ther machines Redundant Key is putting data central User/agent selects dynamically bjects like sensing, cmputatin, actuatin client/servers ut f a netwrk based n Cst functin Fitness fr the jb Cnditin Mechatrnic bject Mechatrnic bject Mechatrnic bject Mechatrnic bject Cmmunicatin Data Security - Encryptin Strage Machine learning AI Clud Mechatrnic bject bjective Mechatrnic bject (Virtual) Sensr Cmputatin Actuating [adapted frm Bradley 16]

23 New tplgies Structure (mechatrnic) cyber-physical systems Prmises Faster develpment: n dedicated develpment cmpnent per functin reuse f existing bjects Rbust varying levels f cnfidence pssible and built in Interchangeable n functinal spec / nt n physical spec Dream f multi vendr cmpatible bjects

24 Future Mechatrnics System engineering t deal with cmplex mechatrnic systems We live in expnential times - ur bilgy desn t (yet) There are limits t what level f cmplexity can be grasped by the human brain System engineering is tl t manage design f cmplex mechatrnic system Large rle fr system architect Strict methdlgy eg INCOSE V-mdel Mdel based : the mdel drives the prcess. The engineers frm different disciplines and levels use mdels t talk t each ther.

25 Mechatrnics Take away Mechatrnics: all said and dne? N: changed ver time and is still changing tday based n technlgies available The essence = hlistic multi-dmain apprach never has been mre relevant [Millbank] Challenges Cmplexity systems Mre technlgies & New tplgies (distributed nn-hierarchical) > System engineering: cmmunicatin between disciplines based n mdels > Self-X User experience becmes part f system: system and user bth adapt Reliability & Uptime Presence user reduced, expert t cmplex > cnditin mnitring embedded in design f the system Safety Security Pwer density actuatin needs t fllw µpwer and sensrs

26 Mechatrnics 45 Years, and the best is yet t cme Mark Versteyhe mark.versteyhe@kuleuven.be

27