OPTIMIZATION OF PROCESS PARAMETERS THROUGH INJECTION MOLDING CONTROL USING INTELLIGENT METHODS

V doktorski disertaciji je predstavljena metoda za optimiziranje procesnih parametrov v procesu injekcijskega brizganja termoplastičnih materialov, ki je implementirana v razvitem inteligentnem sistemu ΔC-3. Predlagana metoda optimizacije procesnih parametrov temelji na kombiniranem pristopu treh metod umetne inteligence, kjer so vsa pravila in dejstva, ki so bila osvojena med množičnimi eksperimentalnimi testiranji na dveh brizgalnih strojih, implementirana tako v bazi znanja kot tudi v podatkovni bazi in so tudi prenosljiva. Ustrezno vrednotenje robustnosti tega proizvodnega procesa je prikazano na primeru nelinearnega matematičnega modela, kjer je izvedena robustna sinteza vodenja hitrosti brizgalnega polža za načrtan robustni regulator po Glover-McFarlanovi metodi ob upoštevanju vpliva motenj in aditivnega modela odstopanj v okolju Matlab/Simulink. V nadaljevanju smo izvedli obsežne eksperimentalne študije na treh izbranih termoplastičnih materialih, kjer smo analizirali vpliv procesnih parametrov na prečni in vzdolžni skrček kot tudi na kot zvijanja testnih ploščic skupaj z meritvijo signalov akustične emisije. Z neporušitveno metodo smo dokazali njeno praktično uporabnost v tem proizvodnem procesu na primeru iskanja razpok na gravurnih orodnih vložkih in detekciji vlažnosti termoplastičnih materialov. Za podrobnejšo analizo zajetih signalov AE v časovno frekvenčnem prostoru smo uporabili Gaborjevo valčno transformacijo. Prav tako so bile izvedene tudi študije zapolnjevanja testnih ploščic v programskem paketu Moldflow, ki so bile osnova za izvedbo kasnejših morfoloških preiskav z vrstičnim elektronskim mikroskopom na površini testnih ploščic v področju slabe orientacije steklenih vlaken.This doctoral dissertation presents a method for optimizing process parameters in the injection molding process of thermoplastic materials, which is implemented in the developed intelligent system ΔC-3. The proposed method is based on a combined approach containing three methods of artificial intelligence where all the rules and facts, conquered during massive experimental tests on two injection molding machines, are implemented within the knowledge base as well as in the database, and are also transferable. Proper evaluation of the robustness of this production process is shown using the case of a non-linear mathematical model, where a robust control synthesis of injection screw speed for a designed robust controller according to the Glover-McFarlane method, was taken into account regarding the impact of interferences and additive model variations within the Matlab/Simulink environment. During the following, we carried out extensive experimental studies on three selected thermoplastic materials, where we analyzed the effect of process parameters on transverse and longitudinal shrinkages as well on the warping angles of test specimens, together with the measurements of acoustic emission signals. We proved that this non-destructive method would provide practical usages during the production processes in cases of searching for cracks on engraving tools\u27 inserts and humidity detection in thermoplastic materials. We used Gabor wavelet transformation for a more detailed analysis of the captured AE signals within the time-frequency space. In addition the filling studies of the test specimens were carried out using the Moldflow software package, and were the bases for the subsequent morphological investigations using a scanning electron microscope on the surfaces of the test specimens within the areas of poor orientation regarding the glass fibers