Agentno modeliranje dinamike tropov volkov v okolju NetLogo

KLINEC, MIMI (avtor), Mraz, Miha (mentor), Moškon, Miha (komentor)

V diplomskem delu smo zgradili in ovrednotili simulacijski model dinamike tropov volkov. Model je bil razvit v okolju za agentno modeliranje NetLogo. Pri agentnem modeliranju je potrebno definirati lastnosti okolja ter začetna notranja stanja in pravila dinamike agentov. V našem modelu je osnovna entiteta oziroma agent volčji trop. Ti agenti so postavljeni v okolje, ki predstavlja območje Slovenije s sosednjim delom Hrvaške. Okolje NetLogo in agentno modeliranje smo v delu ustrezno predstavili. Predstavili smo tudi vsebinsko ozadje modela, ki temelji na projektih spremljanja dinamike volkov na obravnavanem območju. Opisali smo tudi oblikovno zgradbo modela, tako da smo razložili delovanje posameznih procedur. Dinamiko modela smo analizirali z Morrisovo in Sobolovo metodo globalne občutljivostne analize in rezultate ustrezno ovrednotili. Za vsako metodo globalne analize smo izvedli 3072 simulacij. Vzorce smo generirali z metodama knjižnice SALib. Simulacije smo izvedli s pomočjo paralelizacije s paketom ipyparallel. Zagon simulacij modela v NetLogo okolju smo iz datoteke Python omogočili z vmesnikom pyNetLogo. Za izvedbo analize na podatkih simulacij smo uporabili metodi knjižnice SALib. Po analizi smo interpretirali pomembnejše rezultate in jih primerjali z realno dinamiko volčjih tropov.

• Jezik: Slovenski jezik
• Ključne besede: agentno modeliranje, okolje NetLogo, model populacije volkov, občutljivostna analiza, Morrisova metoda, Sobolova metoda, simulacija, dinamični model, paralelizacija
• Vrsta gradiva: Diplomsko delo/naloga
• Tipologija: 2.11 - Diplomsko delo
• Organizacija: FRI - Fakulteta za računalništvo in informatiko
• Leto izida: 2023
• PID: 20.500.12556/RUL-149500
• COBISS.SI-ID: 164049155
• Datum objave v RUL: 07.09.2023

Sekundarni jezik

• Jezik: Angleški jezik
• Naslov: Agent-based modelling of wolf pack dynamics in NetLogo

Izvleček

In the thesis, we built and evaluated a simulation model of the wolf packs dynamics. The model was developed in the NetLogo agent-based modelling environment. In agent-based modelling, it is necessary to define the properties of the environment, as well as the agents' initial internal states and rules of agent dynamics. In our model, the main entity or agent is the wolf pack. These agents are placed in an environment that represents the area of Slovenia with the neighboring part of Croatia. In this work we presented the NetLogo environment and agent-based modelling. We also presented the substantive background of the model, which is based on projects monitoring the dynamics of wolves in the area under consideration. Furthermore we described the design structure of the model by explaining the operation of individual procedures. We analyzed the model dynamics using the Morris and Sobol method of global sensitivity analysis and evaluated the obtained results accordingly. We performed 3072 simulations for each global analysis method. The samples were generated using the methods of the SALib library. The simulations were performed using parallelization with the ipyparallel package. The launch of model simulations in the NetLogo environment was enabled from a Python file using the pyNetLogo interface. To perform the analysis on the simulation data, we used the methods of the SALib library. After the analysis, we interpreted the more important results and compared them with the real dynamics of wolf packs.

• Ključne besede: agent-based modelling, NetLogo environment, wolf population model, sensitivity analysis, Morris method, Sobol method, simulation, dynamic model, parallelization