The amount of residential solar photovoltaic (PV) installations in Australia and around the world has significantly increased in the last few years. The resulting reverse power flows are creating significant challenges for distribution companies to manage voltages as customers are now experiencing voltage rise issues. Therefore, distribution companies are recognizing the need for adequate three-phase low voltage (LV) feeder models so detailed studies that identify potential impacts and solutions can be carried out. The growing adoption of smart meters (or similar) across Australia and around the world gives an opportunity to exploit the corresponding data using advanced analytics to identify the phase connectivity. Using only voltage time-series data extracted from smart meters, -this work firstly proposes an accurate algorithm based on clustering techniques that can determine the phase group to which each customer is connected to in a given LV feeder. The large deployment of the smart meters in residential LV distribution networks, provides a useful amount of data that could be beneficial to distribution companies for solving the voltage regulation problem by calculating the voltages of the customers. By only using the active and reactive power extracted from the smart meters of each customer, -this work then proposes an accurate model based on deep learning method that can determine the voltage profile of each customer in a given LV feeder. The work presented in this thesis was done due to a research project focused on artificial intelligence application in residential LV distribution networks, taking as example a LV distribution network in Australia at the University of Melbourne, Australia with constant collaboration and consultation with my Co-Supervisor prof. Luis F. Ochoa.
La quantità di impianti fotovoltaici residenziali in Australia e nel mondo è aumentata in modo significativo negli ultimi anni. I flussi di energia inversa che ne derivano stanno creando sfide significative per le società di distribuzione di energia elettrica per gestire il voltaggio, in quanto i clienti riscontrano problemi di aumento della tensione. Pertanto, le società di distribuzione riconoscono la necessità di adeguati modelli di alimentatori trifase a bassa tensione (BT) per poter realizzare studi dettagliati che identifichino potenziali impatti e soluzioni. La crescente adozione di contatori intelligenti (o simili) in Australia e in tutto il mondo offre l'opportunità di sfruttare i dati utilizzando analisi avanzate per identificare la connettività di fase. Utilizzando unicamente i dati di serie temporali di tensione estratti da contatori intelligenti, questo lavoro, in primo luogo, propone un algoritmo accurato basato su tecniche di clustering in grado di determinare il gruppo di fase al quale ciascun cliente è collegato in un determinato alimentatore BT. Il vasto impiego di contatori intelligenti nelle reti di distribuzione BT residenziali, fornisce una quantità di dati che potrebbero essere utili alle società di distribuzione per risolvere il problema della regolazione della tensione calcolando le tensioni dei clienti. Questo lavoro propone un modello accurato basato su metodi di “deep learning” il quale può determinare il profilo di tensione di ciascun cliente in un determinato alimentatore BT utilizzando solo la potenza attiva e reattiva estratta dai contatori intelligenti di ciascun cliente.
Artificial intelligence applications for residential LV distribution networks
SIMONOVSKA, ANGELA
2018/2019
Abstract
The amount of residential solar photovoltaic (PV) installations in Australia and around the world has significantly increased in the last few years. The resulting reverse power flows are creating significant challenges for distribution companies to manage voltages as customers are now experiencing voltage rise issues. Therefore, distribution companies are recognizing the need for adequate three-phase low voltage (LV) feeder models so detailed studies that identify potential impacts and solutions can be carried out. The growing adoption of smart meters (or similar) across Australia and around the world gives an opportunity to exploit the corresponding data using advanced analytics to identify the phase connectivity. Using only voltage time-series data extracted from smart meters, -this work firstly proposes an accurate algorithm based on clustering techniques that can determine the phase group to which each customer is connected to in a given LV feeder. The large deployment of the smart meters in residential LV distribution networks, provides a useful amount of data that could be beneficial to distribution companies for solving the voltage regulation problem by calculating the voltages of the customers. By only using the active and reactive power extracted from the smart meters of each customer, -this work then proposes an accurate model based on deep learning method that can determine the voltage profile of each customer in a given LV feeder. The work presented in this thesis was done due to a research project focused on artificial intelligence application in residential LV distribution networks, taking as example a LV distribution network in Australia at the University of Melbourne, Australia with constant collaboration and consultation with my Co-Supervisor prof. Luis F. Ochoa.| File | Dimensione | Formato | |
|---|---|---|---|
|
Thesis Angela Simonovska FINAL.pdf
accessibile in internet per tutti
Descrizione: Thesis text
Dimensione
4.71 MB
Formato
Adobe PDF
|
4.71 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/152897