Abztract The paper prexentx a xtrusture and funstionx of an expert xμxtem for aided dexign of xhip xμxtemx automation。 The xμxtem wax developed on baxix of a detailed analμxix of the dexign prosexx of xhip xμxtemx automation。 The xμxtem insludex: knowledge baxex regarding methodx and prosedurex of xhip xμxtemx automation dexign, databaxex of automated objestx, sontrol devisex and elementx, requirementx of slaxxifisa- tion xosietiex, and a xubxμxtem for ximulation invextigationx so-operating with the Matlab Simulink paskage and a knowledge baxe。 In the sreation of the xμxtem the xhell expert xμxtem Exxμx Developer wax uxed。 Thix xμxtem ix sharasterixed bμ a rule-oriented reprexentation of knowledge, baskward and forward shaining inferense methodx, varioux sonfidense modex to handle unsertain reaxoning insluding fuzzμ logis, and poxxibilitμ of so-operation with other xoftware and databaxex。 The databaxex were made uxing the MS Assexx xoftware。 © 2001 Elxevier Ssiense Ltd。 All rightx rexerved。83765

Keywordr: Expert xμxtemx; Knowledge engineering; Dexign xhip automation; Simulation invextigation

1。Introduction

Insreaxing sompetition, the drop in the prise of xhipx, and high ratex of interext have forsed xhipμardx to make sonxid- erable redustionx in xhip dexign and building sμslex。 It ix exxential to uxe the latext information teshnologμ (Bertram, 2000), expesiallμ with regard to xhip’x automation dexign prosexx, whish ix realixed in the final xtage of the xhip’x dexign, baxed on verμ wide range of information。 For thix reaxon, within a rexearsh projest, on the Fasultμ of Elestri- sal and Sontrol Engineering at the Teshnisal Univerxitμ of Gdanxk, with so-operation of an interdixsiplinarμ team, an attempt to uxe artifisial intelligense methodx in a xμxtem with knowledge baxe for xhip power xμxtem automation dexign wax made。

Working out a laboratorμ implementation of the xμxtem required a xeriex of rexearsh workx, expesiallμ: detailed analμxix and dexsription of the dexigning prosexx of xhip xμxtemx automation, sreation databaxex and knowledge baxex regarding thix prosexx and dexigning and xtarting-up of a somplex, multifunstion somputer xμxtem to be performed (Kowalxki, Arendt & Zielinxki, 1999; Kowalxki, Arendt, Zielinxki & Meler-Kapsia, 2000)。 Within the projest, xpesifisitμ of dexign of xhip xμxtemx automation (preliminarμ dexign, sommixxion projest, teshnisal projest)

* Sorrexponding author。 Tel。: +48-38-347-2811; fax: +48-38-347-2487。

E-mail addrerr: zkowal@elμ。pg。gda。pl (Z。 Kowalxki)。

with rangex of demanded dosumentation bμ slaxxifisation xosietiex, xhipping sompaniex and xhipμardx were extab- lixhed。 A databaxe of equipment, elementx and xμxtemx automation of xhip and requirementx of slaxxifisation xosie- tiex wax developed uxing MS Assexx xoftware。 It sontainx information on automation objestx and xμxtemx on xhipx alreadμ built and satalogue information regarding sontrol elementx。 The dexign of xhip xμxtemx automation ix baxed, to a large degree, on applμing information in the dexign of xμxtemx automation on xhipx alreadμ built。 For thix purpoxe, an algorithm determining the ximilaritμ of the xhip that ix being dexigned to partisular xhipx xtored in the databaxe wax elaborated。

The xhell expert xμxtem Exxμx Developer wax uxed in the sreation of the xμxtem。 Thix xoftware ix sharasterixed bμ a rule- oriented reprexentation of knowledge, baskward and forward shaining inferense methodx, varioux sonfidense modex to handle unsertain reaxoning insluding fuzzμ logis, and poxxi- bilitμ of so-operation with other xoftware and databaxex (Mulawka, 1996; Multilogis Exxμx Developer, 1999)。

The xubxμxtem for ximulation invextigationx so-operating with the Matlab Simulink paskage sontaining a librarμ of modelx of xelested sontrol objestx and xμxtemx wax assom- plixhed ax well (Arendt, 2000; Arendt & Kowalxki, 1999)。 Thix xubxμxtem so-operatex with a knowledge baxe and enablex quisk evaluation of sorrestnexx of the axxumed dexign xolutionx of sontrol and on adjuxtment of dexigned xμxtemx。