Measurement and management of core activity performance in complex overhaul systems using a fuzzy hybrid model

[V. R. Milovanović]

III, 135 listova

Unapređenje uspešnosti osnovne delatnosti predstavlja najvažniji cilj svakog preduzeća. Pokazatelj ostvarenih rezultata se može dobiti merenjem performansi procesa. U ovoj doktorskoj disertaciji je razvijen model za merenje i upravljanje performansama osnovne delatnosti u složenim organizacijama koja se dominantno bave remontom tehničkih sistema. Model je razvijen u sedam faza. Prvu fazu čini dekomponovanje procesa remonta tehničkih sistema na potprocese koje je realizovano pomoću HIPO (eng. Hierarchy – Input – Output - Processing) metode i APQC (eng. American Productivity & Quality Center) okvira, kao i identifikovanje ključnih indikatora performansi (eng. Key Performance Indicators – skr. KPI). U drugoj fazi formirana su dva ekspertska tima koja su realizovala: ocenjivanje relativnih važnosti svakog KPI, ocenjivanje KPI na nivou svakog potprocesa i određivanje ciljnih vrednosti svakog KPI. Eksperti su ocene iskazali u vidu lingvističkih izraza koja su modelovana sa Fermatan fazi brojem (eng. Fermatean fuzzy sets – skr. FFS). U trećoj fazi određene su relativne važnosti potprocesa uz postignuti konsenzus pomoću Delfi metode proširene FFS. U četvrtoj fazi izračunate su vrednosti KPI na nivou svakog potprocesa. U petoj fazi izvršeno je rangiranje KPI pomoću višekriterijumskih metoda ELECTRE i VIKOR proširene sa FFS. Na osnovu metode sličnosti određeno je najbolje osam rangiranih KPI. U šestoj fazi mereni su KPI pomoću višekriterijumske metode SAW proširene sa FFS, a na osnovu dobijenih rezultata je donet zaključak o efikasnosti procesa remonta. U sedmoj fazi realizovano je unapređenje KPI pomoću genetskog algoritma na osnovu izmerenih i ciljnih vrednosti svakog KPI. Model je testiran na formalnim podacima iz remontne organizacije. Istraživanje je pokazalo da je moguće identifikovati KPI koji najviše utiču na proces remonta, i da se na osnovu izmerenih vrednosti KPI može utvrditi uspešnost procesa remonta. Takođe, korišćenjem genetskog algoritma je moguće pronaći najoptimalnije rešenje za unapređenje KPI. Provera osetljivosti modela je realizovano raznim pristupima kao što je: promena težine kriterijuma maksimalizovanog grupnog obeležja kod metode VIKOR; promena načina modelovanja neizvesnosti kod metode ELECTRE; promena načina određivanja distanci između FFS. Primenom metode sličnosti rezultati su pokazali da je predloženi model stabilan i da se kao takav može koristiti za merenje i upravljanje performansama procesa remonta tehničkih sistema.

Enhancing the efficiency of core business activities represents the primary objective of any enterprise. Performance measurement serves as a key indicator of achieved results. This doctoral dissertation develops a model for measuring and managing the performance of core activities in complex organizations predominantly engaged in the overhaul of technical systems. The model is developed through seven phases. The first phase involves the decomposition of the technical system overhaul process into subprocesses, using the Hierarchy-Input-Output-Processing (HIPO) method and the American Productivity & Quality Center (APQC) framework, along with the identification of Key Performance Indicators (KPIs). In the second phase, two expert teams were formed to evaluate the relative importance of each KPI, assess KPIs at the subprocess level, and determine target values for each KPI. Experts expressed their assessments using linguistic expressions modeled with Fermatean fuzzy sets (FFS). The third phase determines the relative importance of subprocesses through consensus achieved via the Delphi method extended with FFS. In the fourth phase, KPI values were calculated at the level of each subprocess. The fifth phase involved ranking KPIs using the multi-criteria methods ELECTRE and VIKOR, extended with FFS. Based on the similarity method, the top eight ranked KPIs were identified. In the sixth phase, KPIs were measured using the SAW method extended with FFS, and based on the obtained results, conclusions were drawn regarding the efficiency of the overhaul process. The seventh phase involved improving KPIs using a genetic algorithm (GA) based on measured and target values for each KPI. The model was tested on formal data from a maintenance organization. The research demonstrated that it is possible to identify KPIs with the greatest impact on the overhaul process and that the measured KPI values can determine the success of the overhaul process. Additionally, GA can be used to identify the most optimal solutions for KPI improvement. Model sensitivity was verified using various approaches, including changing the weight of the criteria for maximized group characteristics in the VIKOR method, modifying uncertainty modeling in the ELECTRE method, and altering the distance determination approach for FFS. The application of the similarity method confirmed that the proposed model is stable and can be effectively used for measuring and managing the performance of the technical system overhaul process.

performanse, KPI, FFS, proces remonta, GA, HIPO metod, ELECTRE, VIKOR, SAW metod, analiza osetljivosti

performance, KPI, FFS, overhaul process, GA, HIPO method, ELECTRE, VIKOR, SAW method, sensitivity analysis

621-7:004.421(043)

Tekst