Title Analiza multidiskretne daljinske detekcije slučajnog optičkog izvora Title (english) The Multidiscrete Analysis of Remote Detection of a Stochastical Optical Source Author Darko Kolarić Mentor Karolj Skala (mentor)Committee member Boris Kviz (član povjerenstva)Committee member Karolj Skala (član povjerenstva)Committee member Neven Mijat (član povjerenstva)Committee member Branko Mikac (član povjerenstva)Committee member Zvonimir Šipuš (član povjerenstva)Granter University of Zagreb Defense date and country 2002-10-03, Croatia Scientific / art field, TECHNICAL SCIENCES Universal decimal classificationUDC ) 621.3 - Electrical engineering Abstract U disertaciji su razmatrane mogućnosti daljinske detekcije plamena u svrhu pronalaženja postupka i tehničke realizacije sklopovske izvedbe za rano otkrivanje šumskih požara. Spoznaje o značajkama dostupne iz stručne literature, odnosno iz rezultata vlastitih istraživanja slučajnih optičkih izvora, razmatrane su u prvom dijelu ove analize, pri čemu je dano težište na postupak prijelaza energija od plamena na površinu okolnih objekata mehanizmom prijenosa, zasnovanog na elektromagnetskom zračenju plamena u infracrvenom spektralnom području valnih duljina =2-5 m. Postupak analize provodi se na području formiranja i ponašanja plamena, kod raspršenog turbulentnog gorenja krutog nehomogenog goriva. Na osnovi prezentiranih spoznaja o značajkama plamena razvijeno je instrumentalno i sklopovsko okruženje za provođenje niza mjerenja, koja su dala konačne smjernice za usvajanje postupka, koji će omogućiti detekciju optičkog izvora na daljini. U tu svrhu provedeno je 541 mjerenje spektralnih, prostornih i vremenskih značajki plamena. Na temelju dobivenih rezultata razrađen je algoritam za daljinsko prepoznavanje plamena, na osnovi promjene njegovih vremenskih značajki (vremenske serije intenziteta elektromagnetskog zračenja plamena), uslijed utjecaja meteoroloških uvjeta (vjetra) na svojstvene vrijednosti iznosa titranja plamena u pojasu frekvencija 1-15 Hz. Ekstrakcija informacije iz vremenske serije intenziteta plamena zasniva se na vremenskom lociranju promjene titranja plamena, pomoću multirezolucijske analize s wavelet transformacijom (osnovna funkcija transformacije Daubechies br.6). Mogućnost lociranja utjecaja vanjskog faktora (vjetra) u vremenskoj seriji intenziteta zračenja plamena, rezultirala je kvalitetnim prepoznavanjem plamena od ostalih toplinskih izvora, koji se nalaze u vidnom polju detekcijske naprave. Rezultati mjerenja provedenih u postupku analize svojstvenih značajki plamena, te programsko-sklopovsko okruženje koje je razvijeno za tu namjenu, omogućili su izvedbu razvojnog modela detekcijskog sustava. Eksperimentalna ispitivanja koja su obavljena u realnim uvjetima, u potpunosti su potvrdila pravovaljanost izbora postupka za prepoznavanje optičkog izvora elektromagnetskog zračenja, te tehničku izvedbu sklopovskog okruženja detekcijskog sustava.
Abstract (english) The meaning of researches is seen in new electromagnetic wave radiation non-stationary sources characterization with designations of non-linear dynamics. The research tries to gain knowledge to prove the thesis about the feasibility of optimum detection of non-stationary stochastic radiation sources, which have natural development in space and time. The starting point is the state of science, with the use of the existing scientific theory and method from the field of the theory of statistic detection, algorithmic signal processing and the method for information extraction. The application of above modern methods in the multidisciplinary science field on fire and burning, scientifically developed and treated since 1984, is important. The processes of burning various fuels (interactive multidisciplinary process) are analyzed and they show an ever more chaotic picture, unlike usual views about burning as an accidental process. In that field, there is an increasing interest and a noticeable improvement due to new knowledge and technologies in the field of algorithmic computer information processing, and an ever greater need for creation of efficient systems of remote fire detection for natural resources (forests) protection and environment preservation. Due to that, the proposed research is important because, beside the scientifically technological use, it has an exceptional role in creating preconditions for building of safety, preservation of material resources and the eco-system. Remote detection by electromagnetic wave in time and space is not sufficiently developed for non-stationary, stochastic and chaotic phenomena and processes. In such cases, due to inability of an unambiguous finding about signal parameters and variables, and functional formalization, detection systems must be distributed in time and space. Modern detection and discrete analytical methods in temporal, spatial and energetic fields enable information extraction and knowledge creation from complex signals. The proposed research anticipates a scientific elaboration of the detection problem and creation of knowledge about the optimum algorithm for data and information extraction and interpretation, by the application of the expanded spectrometric method with multidiscrete detection. That is done by introducing the dynamic analysis in the temporal field (domain), as an additional new possibility of quality determining of dynamic characteristic features of non-stationary optical radiation sources. The purpose is to optimize the whole detection system in case of spatial, spectral and temporal discretizing for achieving the maximum detect ability of the non-stationary optical electromagnetic radiation source. It is expected that the scientific-research applied project will bring new findings in the field of signal detection and processing, and technologies of detection systems. As the significance of the proposed research an increasing interest is noticed and a noticeable improvement of new findings and technologies in detection and algorithmic computer information processing, and a growing need for developing an effective remote fire detection system for protecting the resources (forests) and preserving the environment. The proposed research is important because, beside the scientific-technological use, it has an exceptional role in creating of preconditions for improvement of safety, preserving of material resources and eco-systems (ecology) of the living environment.
Keywords
daljinska detekcija šumskih požara
multidiskretna obrada signala
senzori infra-crvenog spektralnog područja
frekvencijska analiza titranja plamena
multirezolucijska analiza
Wavelet transformacija
brza Fourier transformacija
IC sustavi za daljinsku detekciju
Keywords (english)
Remote forest fire detection
signal processing
infrared sensors
flame flickering frequency analysis
flickering
Wavelet decomposition
fast Fourier transform
Language croatian URN:NBN urn:nbn:hr:168:351727 Study programme Title: Doctoral study programme "Electrical Engineering and Computing" Type of resource Text File origin Born digital Access conditions Closed access Terms of use Created on 2019-04-30 07:38:14
