Edge machine learning-based industrial fault detection
Detekce poruch v průmyslu s použitím strojového učení v koncovém zařízení
Authors
Supervisors
Reviewers
Editors
Other contributors
Journal Title
Journal ISSN
Volume Title
Publisher
České vysoké učení technické v Praze
Czech Technical University in Prague
Czech Technical University in Prague
Date of defense
Abstract
Táto práca sa zaoberá realizáciou koncového zariadenia s detekciou porúch založenou na strojovom učení. Hlavnou súčasťou systému je mikrokontrolér STM32F413ZH, ktorý vykonáva zber a spracovanie dát a vyhodnotenie pomocou dopredu naučeného modelu. Pre demonštráciu tohto zariadenia bola 3D tlačou vyrobená prevodovka s vymeniteľnými nepoškodenými a poškodenými ozubenými kolesami. Okrem toho demonštračná jednotka obsahuje elektromotor, mikrokontrolér, ktorý ho ovláda, osobitný mikrokontrolér zabezpečujúci ethernetovú komunikáciu a displej. Navrhnutý systém zbiera dáta z inkrementálneho rotačného enkodéru, predspracováva signál a extrahuje príznaky založené na jeho frekvenčných aj časových charakteristikách. Boli otestované a porovnané rôzne modely natrénované v NanoEdge AI Studio a určený limit, pri ktorom systém ešte dokáže spoľahlivo detekovať chyby. S projektom založenom na detekcií anomálií boli dosiahnuté hodnoty $1$ pre mieru skutočne pozitívnych detekcií a $0.74$ pre mieru skutočne negatívnych detekcií. Model klasifikujúci viacero tried mal bezchybné výsledky, keď bol vstup obmedzený iba na normálny stav a dve veľké poškodenia.
This thesis describes the implementation of machine learning-based fault detection on an edge device. The main part of the system is built on an STM32F413ZH microcontroller that performs data acquisition and processing and inference by a pre-trained machine-learning model. A gearbox was 3D printed for demonstration of this device with interchangeable undamaged and damaged wheels. Apart from this, the designed demonstration unit is composed of an electric motor, a microcontroller to control it, a separate microcontroller to enable Ethernet communication and a display. The proposed system collects data from an incremental rotary encoder, preprocesses the signal, and extracts features based on both its frequency and time domain characteristics. Various models trained in NanoEdge AI Studio were tested and compared, and the limit where the system can still reliably detect faults was determined. With anomaly detection, a true positive rate of $1$ and a true negative rate of $0.74$ were achieved. With multiclass classification, a perfect score was obtained when considering only the healthy state and two large faults.
This thesis describes the implementation of machine learning-based fault detection on an edge device. The main part of the system is built on an STM32F413ZH microcontroller that performs data acquisition and processing and inference by a pre-trained machine-learning model. A gearbox was 3D printed for demonstration of this device with interchangeable undamaged and damaged wheels. Apart from this, the designed demonstration unit is composed of an electric motor, a microcontroller to control it, a separate microcontroller to enable Ethernet communication and a display. The proposed system collects data from an incremental rotary encoder, preprocesses the signal, and extracts features based on both its frequency and time domain characteristics. Various models trained in NanoEdge AI Studio were tested and compared, and the limit where the system can still reliably detect faults was determined. With anomaly detection, a true positive rate of $1$ and a true negative rate of $0.74$ were achieved. With multiclass classification, a perfect score was obtained when considering only the healthy state and two large faults.
Description
Keywords
Citation
Underlying research data set URL
Permanent link
Rights/License
A university thesis is a work protected by the Copyright Act of the Czech Republic. Extracts, copies and transcripts of the thesis are allowed for personal use only and at one`s own expense. The use of thesis should be in compliance with the Copyright Act.
Vysokoškolská závěrečná práce je dílo chráněné autorským zákonem. Je možné pořizovat z něj na své náklady a pro svoji osobní potřebu výpisy, opisy a rozmnoženiny. Jeho využití musí být v souladu s autorským zákonem v platném znění.
Vysokoškolská závěrečná práce je dílo chráněné autorským zákonem. Je možné pořizovat z něj na své náklady a pro svoji osobní potřebu výpisy, opisy a rozmnoženiny. Jeho využití musí být v souladu s autorským zákonem v platném znění.