doktorska disertacija
Špela Brglez (Author), Bojan Dolšak (Mentor)

Abstract

V industrijski praksi in v študijskem okolju se izkazuje potreba po vodenju in usmerjanju manj izkušenih razvojnih inženirjev skozi fazo snovanja zahtevnika. Zahtevnik je zelo pomemben dokument, saj se inženirji ves čas razvoja izdelka nanj sklicujejo, na koncu pa se zadovoljivost izdelka meri prav glede na izpolnjevanje začetnih zahtev. Če v tej fazi niso vzpostavljene vse potrebne zahteve za uspešno opredelitev novega izdelka, so potrebne kasnejše neželene iteracije v procesu razvoja, da izpuščene zahteve vzpostavimo. Če je v tej fazi vzpostavljenih preveč zahtev, je izdelek preveč določen, kar lahko vodi k izbiri neoptimalnega koncepta ali celo v povsem neuspešen razvoj. Če so v zahtevniku vzpostavljene zahteve neskladne, to v nadaljnje faze razvojnega procesa vnese negotovost glede izbire, kateri izmed dveh neskladnih zahtev slediti. V doktorski disertaciji smo zato zbrali in uredili znanje o zahtevah in snovanju zahtevnika. Iz znanja smo izpeljali metodo za vzpostavljanje in klasifikacijo zahtev na osnovi potrebnosti. Ta razvojnega inženirja vodi skozi ves postopek snovanja zahtevnika in ga opozarja na morebitno prešibko ali premočno določenost izdelka. Zasnovali smo še verjetnostni model za prepoznavanje neskladnih in nasprotujočih si zahtev, ki uporabnika opozori, če je verjetnost, da sta zahtevi v neskladju, večja od določene meje. Metodo in model smo implementirali v obliki inteligentne podpore za pomoč pri snovanju zahtevnika. Inteligentno okolje, ki smo ga poimenovali GRACE, smo razvijali v dveh stopnjah – druga stopnja je bila nadgradnja prve z upoštevanjem mnenja treh skupin: manj izkušenih razvojnih inženirjev, ekspertov iz industrije in ekspertov s področja razvoja inteligentnih svetovalnih sistemov za podporo v procesu razvoja novih izdelkov. Rezultati različnih opravljenih testiranj so pokazali, da manj izkušeni razvojni inženirji resnično potrebujejo oporo pri odločanju glede števila in tematike definiranih zahtev. Dokazali smo tudi, da se rezultati metode dobro skladajo s primerom iz industrijske prakse in z mnenjem ekspertov. Izvedli smo občutljivostno analizo, ki je potrdila pravilno in ustrezno delovanje modela za prepoznavanje neskladnih zahtev.

Keywords

razvoj izdelka;zahtevnik;zgodnje faze razvoja;verjetnost;klasifikacija zahtev;doktorske disertacije;

Data

Language: Slovenian
Year of publishing:
Typology: 2.08 - Doctoral Dissertation
Organization: UM FS - Faculty of Mechanical Engineering
Publisher: [Š. Brglez]
UDC: 004.896:[005.41:658.512.2](043.3)
COBISS: 18824726 Link will open in a new window
Views: 1717
Downloads: 251
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: INTELLIGENT DECISION SUPPORT FOR REQUIREMENTS LIST ELICITATION IN PRODUCT DESIGN PROCESS
Secondary abstract: The need for guidance of less experienced design engineers through the phase or requirements list elicitation can be observed in industrial practice as well as in academic environment. Requirements list is a very important document, for it is referenced throughout the design process and at the end of desing, appropriateness of the product is measured against the initial requirements. If some necessary requirements are missing in this phase, some unwanted iterations of the design process are needed later on, in order to elicit such requirements. If there are too many requirements elicited, the product is over-constrained, which can lead to an unoptimal concept generation or to the unsuccessful design process in general. Inconsistent requirements of the requirements list bring some ambiguity into the process, about which requirement to follow in the next phases. In the doctoral dissertation we collected and ordered knowledge about requirements and requirements list elicitation. From this gathered knowledge we derived a method for requirements elicitation and classification on the basis of necessity. The method leads the design engineer through requirements list elicitation and it alerts him or her about possible under- or over-constraining of the product. We also produced a probability model for inconsistent and contradictory requirements recognition, which alerts the user, when the probability of inconsistency is higher than a defined limit. The method and the model were implemented in the form of intelligent support system for help with requirements list elicitation. The intelligent environment, which we named GRACE, was developed in two stages – the second stage was an upgrade of the first in which we took into account the opinions of: less experienced design engineers, experts from industry and experts of development of decision support systems for product design area. Results of different testings showed that less experienced design engineers ideed need some guidance for deciding about the number and topics of the elicited requirements. We also proved that the results of the method fit very closely with a real industry case and with expert opinions. We conducted a sensitivity analysis that confirmed the correctness and appropriateness of the probability model for inconsistency recognition.
Secondary keywords: product design;requirements list;early phases of design;requirements classification;
URN: URN:SI:UM:
Type (COBISS): Dissertation
Thesis comment: Univ. v Mariboru, Fak. za strojništvo
Pages: VIII, 128 str.
ID: 8753070