Net load and minimum demand forecasting for networks

Supervisors

Dr Baran Yildiz (UNSW)

The project 

Given the growing aggregate contribution of rooftop solar, there are increasing periods where Distributed Photovoltaic (D-PV) generation is very high relative to underlying demand, resulting in low demand on the network minimum system load (MSL). 

This research project is focused on optimising machine learning methods for net electricity load forecasting in households equipped with D-PV systems and demand response appliances. Net load monitoring is becoming more popular as most households have smart meters. There is lack of forecast models that focus on the net loads (the underlying demand on the grid) and existing models target gross load (network +solar) or solar generation only, which is a significant gap. The aim is to develop an accurate and effective net load forecasting model which can enhance the accuracy and efficiency of energy management systems and network operations. 

With increasing D-PV uptake, MSL has become more critical for distribution network service providers (DNSPs).  

The project will explore varying spatial levels from household to feeder to state level to optimize network operations and address minimum demand challenges, and temporal levels from 1-minute to daily level. Finding an adaptable solution to different scales and resolutions will maximize the potential financial and environmental benefits for both households and electricity network operators.