Intelligent tolerancing of assemblies regarding technological capabilities of manufacturing processes

Sodobna proizvodnja je podvržena najrazličnejšim zahtevam, ki jih povezuje zahteva po učinkovitosti. Da zadostimo tej zahtevi, je treba tolerance sestavnih delov sklopa načrtovati s premislekom. Z ustreznim načrtovanjem toleranc lahko namreč zelo vplivamo na zmanjšanje proizvodnih stroškov. V ta namen je v doktorski disertaciji za reševanje kompleksnega problema načrtovanja toleranc razvit in predstavljen inteligentni sistem toleriranja, ki s pomočjo metod umetne inteligence, na podlagi vhodnih podatkov porazdeli tolerance sestavnih delov, tako da so stroški izdelave minimalni. Razvita sta dva različna modula za načrtovanje tolerancmodul z optimizacijo z rojem delcev in modul, ki temelji na gravitacijskem iskalnem algoritmu. Uporaba razvitega sistema je prikazana na dveh realnih primerih. Primerjane so vrednosti najnižjih doseženih stroškov s posamezno optimizacijsko metodo, vrednosti predlaganih toleranc in izbrani proizvodni procesi – stroji, ki jih je predlagala inteligenca. Skozi opravljena testiranja pri zasnovi inteligentnega sistema toleriranja se je optimizacija z rojem delcev izkazala za najučinkovitejšo metodo. Razvit je tudi uporabniški vmesnik, ki omogoča enostavno načrtovanje toleranc. V zaključku raziskave je tudi potrjena teza doktorske disertacije, hkrati pa so tudi podane smernice za nadaljnji razvoj in raziskave.Modern production is subjected to manifold requirements which all have one common link – effectiveness. To meet this requirement, tolerances of the assembly components need to be planned wisely. An appropriate tolerance allocation can lead to significantly reduced production costs. To solve the complex problem of tolerance allocation, an intelligent tolerancing system that uses artificial intelligence methods which, on the basis of input data, distribute tolerances of the components in a way to keep the manufacturing costs to a minimum, is developed and presented in this dissertation. For the tolerance allocation, two different modules are developedthe module with particle swarm optimization and the module based on gravitational search algorithm. The use of the developed system is shown in two real cases. The values of the lowest costs achieved by an individual optimization method, the values of suggested tolerances and selected manufacturing processes – machines suggested by intelligence are compared. In the course of testings performed when developing the intelligent tolerancing system, the particle swarm optimization proved to be the most efficient method. User interface that enables an easy tolerance allocation is developed as well. In the conclusion of the research the dissertation’s thesis is confirmed. At the same time, guidelines for the further development and researches are given