Computer-Aided Problem Solving: A Dialogue-based System to Support the Analysis of Inventive Problems

Transcription

1 Knwledge Aided Engineering Manufacturing and Related Technlgies Cmputer-Aided Prblem Slving: A Dialgue-based System t Supprt the Analysis f Inventive Prblems Nicclò Becattini Yuri Brgianni Gaetan Cascini - gaetan.cascini@plimi.it Federic Rtini

2 Cntext 2 TRIZ imprves individuals and rganizatins innvatin capabilities Surce: Presentatin at Japanese TRIZ Sympsium 2009 by Panasnic

3 Cntext 3 On the ther hand assimilating the TRIZ lgic requires nt negligible learning effrts; A typical curse t gain sufficient prblem slving skills requires: At least 40 hurs frnt lessns Several mnths f practice with real case studies As well TRIZ based CAI (Cmputer-Aided Innvatin) tls dn t result t be effective with fellws wh have nt attended vcatinal curses

4 Cntext 4 Creativity and inventiveness, which are crucial fr innvatin, are nt supprted by PLM systems [1]. It is required t extend the supprt t the whle design prcess Level f Abstractin CAI PLM CADCAE CAM Cnceptual Design Embdiment Design Detail Design Manufacturing Prduct Develpment Cycle [1] Van Elsas A., Vergeest J.S.M., New Functinality fr cmputer-aided cnceptual design: the displacement feature, Design Studies. 19 (1998)

5 Gal and utline 5 The authrs prpse a prttype algrithm fr prblem slving viable t be implemented in a sftware applicatin, taking in cnsideratin the recalled lacks f CAI systems. The framewrk f OTSM- TRIZ cnstitutes the cre f the theritical backgrund fr the develpment f such dialgue-based system

6 Gal and utline 6 Step-by-step algrithm fr the analysis f the prblem and the rganizatin f the related infrmatin accrding t the TRIZ knwledge-base, in rder t supprt the prblem slving prcess. Outline System Requirements Related art: Prblems characterizatin Prblem Slving appraches Cmputer-Aided tls fr Prblem Slving Reference mdels A dialgue-based system fr the analysis f inventive prblems Detailed structure f the algrithm Validatin Preliminary tests with students f Engineering Design Tests in industry Cnclusins and future wrks

7 System Requirements 7 Step-by-step algrithm fr the analysis f the prblem and rganizatin f the related infrmatin accrding t the TRIZ knwledge-base, in rder t supprt the prblem slving prcess. Capability t face Difficult prblems by giving pririties t the bjctives Nn-typical prblems by highlighting cnflicting requirements and depicting the main characteristics f the desired slutin Ill-structured prblems, turning them int well-structured by means f their prper definitin Inventive prblems by helping the user in retrieving useful infrmatin and data frm available resurces in varius dmains Prvide useful stimuli t the prblem slver in rder t leverage his/her knwledge and creative skills Drive the prblem slver in useful infrmatin searches that may widen its design space and then the space f slutin Definitin f search criteria that allws autmatizatin in rder t require minimum effrts t the designer

8 System Requirements 8 Main characteristics f the dialgue-based system fr Prblem Analysis: limited amunt f training natural language speed up the infrmatin search n patent-mining cmpetences required

9 9 Related art: prblems characterizatin Cmmn prblems t be faced during inventive design activity: Difficult prblems / Easy prblems [2] Intransparency Cmplexity Dynamics Plytely Typical prblems / Nn-typical prblems [3] Nn-typical prblems require t leverage persnal knwledge and wisdm Inventive prblems / Nn-inventive prblems [4] Whatever lead t an useful, nvel and unbvius slutin Well-structured prblems / Ill-structured prblems [5] Six main characteristics define WSP, ISP cnstitutes a class f residuals. [2] Funke J., Frensch P.A., Cmplex prblem slving: the Eurpean perspective in Jnassen D.H. (Ed.) Learning t slve cmplex scientific prblems Lawrence Erilbaum, New Yrk, 2007, [3] Altshuller G.S., Creativity as an exact science, Grdn and Breach Science, New Yrk, [4] Cavallucci D., Russelt F., Zanni C., Assisting R&D activities thrugh definitin f prblem mapping. CIRP Jurnal f Manufacturing Science and Technlgy. 1 (2009) [5] Simn H.A., The structure f ill-structured prblems, Artificial Intelligence, 4, (1973),

10 10 Related art: Prblem Slving Appraches Cgnitive appraches: Characteristics [6] Analgy Abstractin Assciatin f ideas Crss-fertilizatin thinking prcesses Leverage n tacit knwledge Examplary Cgnitive Methds fr Prblem Slving Case-Based Reasning Brainstrming [6] Rss V.E., A mdel f inventive ideatin, Thinking skills and creativity. 1 (2006),

11 11 Related art: Prblem Slving Appraches Systematic appraches: Characteristics Linear Step-by-step prcedures Efficient fr just sme types f prblems (pr versatility) Desn t require specific creative capabilities Rely n explicit knwledge (Patents, Handbks, Scientific Literature, ) Examplary Systematic Methds fr Prblem Slving TRIZ Cnstraint Satisfactin Prblems and its evlutin

12 12 Related art: Prblem Slving Appraches TRIZ cannt be cnsidered as a purely systematic methds since ARIZ is a tl t aid thinking, but it cannt replace thught itself ( ) it is exactly the sparks f imaginatin that lead humans t prduce creative nn-typical ideas. [7] The plain differences between cgnitive and systematic appraches shuld be cnciliated in a step-by-step methd that leverages individual s tacit knwledge since: design methdlgy shuld therefre fster and guide the ability f designers, encurage creativity, and at the sam time drive hme the need fr bjective evaluatin f the results [8] [7] Khmenk N., De Gui R., Leilat L., Kaikv I., A framewrk fr OTSM-TRIZ Based Cmputer Supprt t be used in Cmplex Prblem Management, Internatinal Jurnal f Cmputer Applicatin in Technlgy, 30, (2007), [8] Pahl G., Beitz W., Engineering Desing A systematic apprach, 2 nd Editin, Springer, Lndn 1996.

13 13 Cmputer-Aided tls fr Prblem Slving Fur kinds f aiding the user within Cmputer systems fr prblem slving [9]: by facilitating the management f the wrking prcess, encuraging the perseverance f designer in the research f innvative slutins; by easing the cmmunicatin between design team members, since circulatin and integratin f ideas play a relevant rle in the creative prcess; by aiding the designer with a caching activity, acting as an expert system that guides the user thrughut cgnitive prcesses; by cperating in the creative prcess, thanks t the Artificial Intelligence systems that cntribute t ideas generatin. [9] T. Lubart, Hw can cmputers be partners in the creative prcess: Classificatin and cmmentary n the Special Issue, Internatinal Jurnal f Human-Cmputer Studies. 63 (2005)

14 Reference Mdels: OTSM-TRIZ 14 Hill mdel (abstractin-synthesis) [*] Tngs mdel (frm current situatin t ideality, barriers identificatin) [*] Level f abstractin General prblem (Cntradictin) Ideal slutin General ( Physical ) slutin Specific Prblem t be slved (Inventive Situatin) Specific ( Engineering ) slutin Instruments fr NOT Typical Prblem Instruments fr Typical Prblem Instruments fr NOT Typical Prblem Initial Situatin (Available Resurces) A Barrier (Cntradictin) preventing the achievement f the Mst Desirable Result Mst Desirable Result Cnceptual Slutin * surce: Niklai Khmenk

15 Reference Mdels: OTSM-TRIZ, TRIZ Funnel mdel (cnvergent prcess) [*] System Operatr (system thinking) 15 Past Future * surce: Niklai Khmenk

16 16 A dialgue-based system fr the analysis f inventive prblems The algrithm is structured in lgical blcks aimed at investigating different aspects f the system; diversified utputs are freseen upn the reached degree f frmalizatin and abstractin f the prblem

17 A dialgue-based system fr the analysis f inventive prblems Main features f the algrithm regarding knwledge search and prblem slving pprtunities: it helps at first the user in expliting his knw hw by suggesting prper prblem slving paths, that dn t require external expertise, thus facilitating the implementatin f the ideas generated; the search in external industrial dmains is addressed when the abstractin prcess has been cmpleted (i.e. definitin f a physical cntradictin); it prvides indicatins fr suitable slving alternatives thrugh different TRIZ tls, e.g. separating in time/space, trimming useless r lw-valued cmpnents, etc. 17 Level f abstractin General prblem (Cntradictin) Ideal slutin General ( Physical ) slutin Specific Prblem t be slved (Inventive Situatin) Specific ( Engineering ) slutin Instruments fr NOT Typical Prblem Instruments fr Typical Prblem Instruments fr NOT Typical Prblem

18 A dialgue-based system fr the analysis f inventive prblems Lgical blck Initial Situatin (IS): it prvides a preliminary descriptin f the system and the prblem under investigatin Intrduced variables/parameters: the 18

19 A dialgue-based system fr the analysis f inventive prblems Lgical blck Negative Effect (NE): it investigates the undesired effect that arises in the system, as well as the negative cnsequences and impact. 19

20 A dialgue-based system fr the analysis f inventive prblems Lgical blck ARIZ (AR): it is suppsed t allw the definitin f a physical cntradictin in TRIZ terms. 20

21 A dialgue-based system fr the analysis f inventive prblems Lgical blck Perfrmance (PE): it is meant t refrmulate the system t be analyzed r the negative effect. 21

22 A dialgue-based system fr the analysis f inventive prblems Lgical blck Resurces (RE): it deepens the investigatin f the system, in terms f the excessive amunt f resurces spent during its emplyment. 22

23 A dialgue-based system fr the analysis f inventive prblems Lgical blck Csts (CO): it investigates the reasns f incurring high csts. 23

24 A dialgue-based system fr the analysis f inventive prblems Lgical blck Prcess (PR): it investigates the criticalities f the manufacturing r delivering prcess. 24

25 Validatin: preliminary tests at University The dialgue-based algrithm have been preliminary tested n three case studies with students frm: Plitecnic di Milan University f Flrence Metrics fr results Evaluatin GOOD RESULT: Precise descriptin f the prblem N particular mistakes r misinterpretatin Definitin f useful variables in rder t perfrm a Patent Search by keywrds SATISFACTORY RESULT: Almst cmplete descriptin f the prblem Glbal cmprehensin f the prblem Variable capable t suggest just preliminary directins fr Patent Search by keywrds UNSATISFACORY RESULT: Pr descriptin f the prblem Diffused misinterpretatins Variable nt capable t enlarge the space f slutin by Patent Search 25

26 The dialgue-based algrithm have been preliminary tested n three case studies with students frm: Plitecnic di Milan University f Flrence Case Studies Validatin: preliminary tests at University A. Reducing energy waste frm an andizing tank, withut creating bstacles t the prcess; B. Extreme wear f the fringe curtain in a X-ray inspectin system fr fd industry; C. Cmplete transfer f xygen frm a large cylinder t several smaller nes, withut any cmpressr. Althugh each prblem structure depends n the user interpretatin, it is expected that peple mdel Case A as a typical TRIZ cntradictin r as a situatin where a given perfrmance is required; Case B clearly pints t a negative effect, but shuld be preferentially mdelled as a cntradictin; Case C shuld address the prblem slver twards the implementatin f a new perfrmance r the imprvement f an existing ne. 26

27 The dialgue-based algrithm have been preliminary tested n three case studies with students frm: Plitecnic di Milan University f Flrence Results Validatin: preliminary tests at University 27

28 The dialgue-based algrithm have been preliminary tested n three case studies with students frm: Plitecnic di Milan University f Flrence Results Validatin: preliminary tests at University 28

29 The dialgue-based algrithm have been preliminary tested n three case studies with students frm: Plitecnic di Milan University f Flrence Results Validatin: preliminary tests at University 29

30 The dialgue-based algrithm have been preliminary tested n three case studies with students frm: Plitecnic di Milan University f Flrence Results Validatin: preliminary tests at University 30

31 Tests n industrial case studies 31 The dialgue-based algrithm has been further tested n three case studies with cmpanies that wrk in different industrial fields: Ludspeakers prductin Specific Prblem: Manufacturing prcess fr the prductin f acustic membranes fr ludspeakers Arthrscpic device prductin Specific prblem: Design f an innvative arthrscpic device fr human jints Feet and legs prsthesis Reduce weight f prsthesis cmpnents in rder t increase user s cmfrt during sprt activities Preliminary results Situatin analysis clarifies main issues Definitin f at least ne Cntradictin accrding t TRIZ bdy f knwledge Definitin f the main characteristics f the desired slutin Further Develpments Full validatin activity nce the Patent search has been fully implemented

32 Cnclusins 32 The dialgue-based algrithm shws psitive results bth with students and with technicians in industry Such Dialgue-based system cnstitutes a systematic guide t enhance individual s cgnitive capabilities The supprt t the prblem slving prcess by means f external knwledge is under testing als thrugh integratin f custumized field thesauri A mre extensive validatin and imprvement prcess has been planned (industrial cperatins are welcme)