Samoorganizirajući distribuirani upravljački sustav u pametnim zgradama

master's thesis

• Title (english): Self-organizing edge-computing control system in smart buildings
• Author: Martin Požeg
• Mentor: Vinko Lešić (mentor)
• Committee member: Vinko Lešić (predsjednik povjerenstva)
• Committee member: Mario Vašak (član povjerenstva)
• Committee member: Anita Banjac (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-07-06, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Klimatizacija pametnih zgrada jedan je od vodećih informacijsko-komunikacijskih smjerova za postizanje značajnih ušteda energije aktivnim pristupima. Preduvjet za aktivne uštede digitalizacija je zona te međusobna koordinacija sa zajedničkim ciljem smanjenja troška. Pri tome se bežična komunikacija ističe kao brz, efikasan i neinvazivan pristup pogodan za veliki broj distribuiranih senzora i upravljačkih mikroračunala tvoreći složeni IoT sustav. Kako bi se ostvarila automatizacija predviđanjem temperature koristi se strojno učenje na računalnom oblaku.

Izrađeni distribuirani sustav u ovom diplomskom radu svakih 30 minuta očitava podatke o temperaturi, vlagi i tlaku te ih šalje na centralnu upravljačku jedinicu. Raspberry Pi ih zatim prosljeđuje na Google Cloud IoT platformu putem MQTT protokola. Sva mjerenja odnose se na mjerenja u prostoriji dok je prozor otvoren. Podatci se spremaju na računalni oblak u bazu podataka BigQuery. Pomoću Vertex AI modula ti podatci koriste se za izradu modela koji predviđa temperaturu uz uvjet otvorenog prozora. Automatizacija prostorije izvodi se kroz Python skriptu koja svakih sat vremena gleda predikcije za 6h unaprijed i aproksimira predikcije temperature za zatvoreni prozor. Oduzimanjem predikcija od idealne temperature koju korisnik postavi te njihovom srednjom vrijednošću algoritam određuje i upravlja otvorenošću i zatvorenošću prozora paljenjem i gašenjem žarulje.

Abstract (english)

Air conditioning of smart building is one of the leading information and communication directions for achieving significant energy savings through their active approaches. A prerequisite for active savings is digitalisation of the zone and mutual coordination with the same common goal of reducing costs. In doing so, wireless communication stands out as a fast, efficient and non-invasive approach suitable for large number of distributed sensors and central control microcomputers forming a complex IoT system. To achieve automation by temperature prediction, machine learning model is used on the cloud.

The developed distributed system in this thesis reads temperature every 30 minutes and sends it to the central control unit Raspberry Pi which forwards it to the Google Cloud IoT platform via MQTT protocol. All measurements refer to measurements in the room while the window is open. This data is then stored in the BigQuery database on the platform itself. Using the Vertex AI module, this dana is used to create a model that predicts the temperature with the open window. Automatization of the room is performed through a Python script that looks at predictions 6 hours in advance doing it every hour and approximates temperature predictions for a closed window. By subtracting the prediction from the ideal temperature set by the user and by their mean value, the algorithm determines and controls the openness and closure of the window by turning the lamp or off.

• Keywords: IoT; Zigbee senzori; Google Cloud IoT; BigQuery; Python; centralna upravljačka jedinica; pametna kuća; strojno učenje; MQTT
• Keywords (english): IoT; Zigbee sensors; Google Cloud IoT; BigQuery; Python; central control unit; smart home; machine learning; MQTT
• Language: croatian
• URN:NBN: urn:nbn:hr:168:825720
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 13:33:00