Algoritem za napovedovanje toplotne obremenitve stavb na večjem geografskem območju : doktorska disertacija

Bizjak, Marko

| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

First page > Show document

Show document

Title:	Algoritem za napovedovanje toplotne obremenitve stavb na večjem geografskem območju : doktorska disertacija
Authors:	ID Bizjak, Marko (Author) ID Lukač, Niko (Mentor) More about this mentor... ID Štumberger, Gorazd (Comentor)
Files:	DOK_Bizjak_Marko_2019.pdf (19,98 MB) MD5: 3AA86A6F4676B1E00734A3C3B2E1244C PID: 20.500.12556/dkum/a8aff1d1-7663-4330-bb8d-b18593ec87cf
Language:	Slovenian
Work type:	Doctoral dissertation
Typology:	2.08 - Doctoral Dissertation
Organization:	FERI - Faculty of Electrical Engineering and Computer Science
Abstract:	V doktorski disertaciji predstavimo nov algoritem za napovedovanje toplotne obremenitve stavb na večjem geografskem območju. Pri tem geografsko območje opisujejo visokoločljivostni podatki laserskega zajema LiDAR. Najprej ob vključitvi podatkov iz javnih prostorskih baz z algoritmom rekonstrukcije na osnovi 3D Boolovih operacij nad polprostori generiramo 3D trikotniške modele stavb. Pri tem vsakemu trikotniku določimo pripadajoč material. Trikotniki posameznega modela določajo zunanji ovoj stavbe. Modele stavb združimo z modeloma reliefa in vegetacije, ki sta prav tako generirana iz podatkov LiDAR, s čimer določimo geometrijske lastnosti okolja. Sledi izračun toplotne obremenitve stavb, ki ga izvedemo za vsak časovni korak v izbranem obdobju. Za posamezen trikotnik izračunamo vidno nebo, ovrednotimo sončno obsevanje ter na podlagi temperaturne razlike in fizikalnih lastnosti materialov izračunamo prenos toplote v danem časovnem trenutku. Izračun toplotne obremenitve stavb nato paraleliziramo na grafični procesni enoti s tehnologijo CUDA. V eksperimentalnem delu pokažemo uporabnost predlaganega algoritma za napovedovanje toplotne obremenitve stavb na večjem geografskem območju, pri čemer izvedemo tako dolgoročno kot kratkoročno napovedovanje na podlagi meteorološke napovedi. V primerjavi s sorodnimi algoritmi lahko dosežemo vsaj za 10 % bolj točne rezultate, pri čemer je predlagan algoritem tudi manj občutljiv na nižjo gostoto podatkov LiDAR. Na GPE je v primerjavi s CPE možno več kot 60-krat hitrejše izvajanje, kar predstavimo na koncu eksperimentalnega dela.
Keywords:	paralelno računanje, GPGPU, CUDA, LiDAR, modeli stavb, toplotna obremenitev, prenos toplote, računalniške simulacije
Place of publishing:	Maribor
Place of performance:	Maribor
Publisher:	[M. Bizjak]
Year of publishing:	2019
Number of pages:	X, 96 str.
PID:	20.500.12556/DKUM-73344
UDC:	004.9:697.329(043.3)
COBISS.SI-ID:	22496278
NUK URN:	URN:SI:UM:DK:HWUJMCDU
Publication date in DKUM:	11.07.2019
Views:	2165
Downloads:	419
Metadata:
Categories:	KTFMB - FERI
:	Copy citation

Average score:	(0 votes)
Your score:	Voting is allowed only for logged in users.
Share:

Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

Licences

License:	CC BY-NC-ND 4.0, Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International

Link:	http://creativecommons.org/licenses/by-nc-nd/4.0/
Description:	The most restrictive Creative Commons license. This only allows people to download and share the work for no commercial gain and for no other purposes.
Licensing start date:	27.03.2019

Secondary language

Language:	English
Title:	An algorithm for estimating thermal load of buildings on a large geographic area
Abstract:	A new algorithm for estimating the thermal load of buildings on a large geographic area is introduced in this Doctoral thesis. The large geographic area is described by high-resolution LiDAR data. By incorporating data from public spatial databases, we first generate 3D building models comprised of triangles, using an algorithm for building reconstruction on the basis of 3D Boolean operations over half-spaces, where each triangle's corresponding material is set. Triangles of each model define the building's envelope. Geometric features of the environment are described by combining building models with terrain and vegetation models, generated from LiDAR data as well. Then we calculate the thermal load, which is performed for each time-step within the selected period. For each triangle we calculate the visible sky, estimate solar irradiation and, based on the temperature difference and physical properties of materials, we calculate the heat transfer for each time-step. The calculation of thermal load is then parallelised on a Graphics Processing Unit using CUDA technology. In experiments we demonstrate the application of the proposed algorithm for estimating the thermal load of buildings on a large geographic area, where we perform long-term, as well as short-term, thermal load estimation on the basis of weather forecast. In comparison with related work, the proposed algorithm can yield at least 10 \% more accurate results, while being less sensitive to lower density LiDAR data. On a GPU the execution can be over 60-times faster than on a CPU, which is presented at the end of the experiments.
Keywords:	parallel computing, GPGPU, CUDA, LiDAR, building models, thermal load, heat transfer, computer simulations

Comments

Leave comment

You must log in to leave a comment.

Comments (0)

0 - 0 / 0

There are no comments!

Back