Konvolucijske nevronske mreže za odkrivanje napak s pomočjo zvoka

magistrsko delo

Gorazd Fažmon (Author), Marjan Golob (Mentor)

Abstract

V magistrskem delu je predstavljen razvoj sistema za zaznavanje napak v industrijskih procesih, ki temelji na osnovi zaznave zvoka. S pomočjo programskega orodja Audacity, so zajeti zvočni signali proizvodnih postopkov. S programskim orodjem Python je izdelan program za pretvorbo zvočnega signala v sliko. Z uporabo Python knjižnice TensorFlow je program naučen, da prepozna napako. Podan je podroben opis pomembnih pojmov, algoritmov, metod in testiranj sistema. Glavni cilj naloge je implementirati zgrajen sistem na dejanskem proizvodnem postopku.

Keywords

konvolucijske nevronske mreže;kakovost zvoka;spektrogram;Mel frekvenčni kepstralni koeficienti;magistrske naloge;

Data

Language:	Slovenian
Year of publishing:	2020
Typology:	2.09 - Master's Thesis
Organization:	UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher:	[G. Fažmon]
UDC:	681.586.4(043.2)
COBISS:	36699139
Views:	411
Downloads:	80
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Convolutional neural networks for sound-based error detection
Secondary abstract:	The master’s thesis presents the development of an industrial process fault detection system based on sound sensing. Using Audacity software, audio signals from production processes are captured. With the use of Python software, a program for converting audio to image is created. Using the Python TensorFlow library, the program is taught to recognize the error. A detailed description of important system concepts, algorithms, methods, and tests is given. The main objective of the task is to implement the built system on the actual production process.
Secondary keywords:	convolutional neural network;sound quality;spectrogram;Mel-frequency cepstral coefficient (MFCC);TensorFlow;
Type (COBISS):	Master's thesis/paper
Thesis comment:	Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Elektrotehnika
Pages:	X, 72 f.
ID:	11877054