Izrada hidrološkog modela rijeke Bednje baziranog na povratnoj neuronskoj mreži

master's thesis

• Title (english): The development of a hydrological model of the Bednja river based on a recurrent neural network
• Author: Dora Varjačić
• Mentor: Dijana Oskoruš (mentor)
• Committee member: Sanja Kovač (predsjednik povjerenstva)
• Committee member: Dijana Oskoruš (član povjerenstva)
• Committee member: Karlo Leskovar (član povjerenstva)
• Committee member: Ranko Biondić (član povjerenstva)
• Committee member: Jelena Loborec (član povjerenstva)
• Granter: University of Zagreb; Faculty of Geotechnical Engineering; Varaždin
• Defense date and country: 2023-09-25, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Interdisciplinary Technical Sciences; Environmental Engineering

Abstract

Hidrologija je znanost koja proučava distribuciju i kretanje vode na površini i ispod
površine Zemlje te u njezinoj atmosferi. Informacije dobivene analizom ponašanja i
kretanja vode kroz hidrološki ciklus od velike su važnosti u područjima kao što su
upravljanje vodnim resursima, upravljanje rizicima od opasnosti povezanih s vodom,
projektiranje infrastrukture, proizvodnja električne energije i zaštita okoliša.
Hidrološko modeliranje moćan je alat koji omogućava simulaciju i razumijevanje
interakcija između različitih komponenti hidrološkog ciklusa, a odabir vrste modela ovisi
o ciljevima istraživanja, dostupnim podacima, računalnim resursima i karakteristikama
hidrološkog sustava koji se modelira.
U hidrološko modeliranje sve se više integrira umjetna inteligencija, kako bi se
prevladala ograničenja klasičnih modela te poboljšala njihova točnost i mogućnosti
predviđanja. Jedno od potpolja umjetne inteligencije su umjetne neuronske mreže, koje
se koriste za određivanje nelinearnih odnosa i vremenskih ovisnosti u hidrološkim
podacima.
U ovom radu korišten je programski jezik Python za razvoj i implementaciju
hidrološkog modela baziranog na povratnoj neuronskoj mreži. Ova specifična arhitektura
neuronske mreže odabrana je zbog svoje sposobnosti hvatanja uzoraka i vremenskih
ovisnosti u nizovima podataka.
Izrađen je model sliva Bednje, koji ima bujični režim ovisan o količini oborina. Model
predviđa protoke na hidrološkoj postaji Ludbreg, na temelju prethodnih podataka o
oborinama i protocima s uzvodnih postaja.

Abstract (english)

The development of a hydrological model of the Bednja River based on a
recurrent neural network
Hydrology is a science that studies distribution and movement of water below and
above the Earth's surface and in its atmosphere. The information obtained from the
analysis of the behaviour and movement of water through the hydrological cycle is of
great importance in areas such as water resources management, risk management of
water-related hazards, infrastructure design, electricity generation and environmental
protection.
Hydrological modelling is a powerful tool that enables simulation and understanding
of interactions between different components of the hydrological cycle, and the choice of
a model type depends on the research objectives, available data, computer resources and
characteristics of the hydrological system that is being modelled.
Artificial intelligence has been increasingly integrated into hydrological modelling, in
order to overcome the limitations of classic models and enhance their accuracy and
predictive capabilities. One of the subfields of artificial intelligence are artificial neural
networks, which are used to determine nonlinear relationships and time dependencies in
hydrological data.
In this paper, the Python programming language was used for the development and
implementation of a hydrological model based on a recurrent neural network. This
specific neural network architecture was chosen for its ability to capture patterns and
temporal dependencies in data sets.
A model of the Bednja basin was created, which has a flood regime dependent on the
amount of precipitation. The model predicts flows at the Ludbreg hydrological station,
based on previous data on precipitation and flows from upstream stations.

• Keywords: hidrologija; sliv Bednje; hidrološko modeliranje; povratna neuronska mreža
• Keywords (english): hydrology; hydrological modelling; Bednja basin; recurrent neural network
• Language: croatian
• URN:NBN: urn:nbn:hr:130:238620
• Study programme: Title: Geoenvironmental Engineering; specializations in: Geoenvironmental Engineering, Environmental Management, Water Management; Course: Water Management; Study programme type: university; Study level: graduate; Academic / professional title: sveučilišni/a magistar/magistra inženjer/inženjerka inženjerstva okoliša (sveučilišni/a magistar/magistra inženjer/inženjerka inženjerstva okoliša)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2024-02-13 10:39:58