Machine Learning Applications for Utility-Scale PV

current

The project aims to develop a commercial monitoring system for solar photovoltaic (PV) plants, capable of investigating module and system degradation as well as prediction soiling rates. UNSW will develop machine learning algorithms from historical data provided by its project participants to optimise operations and maintenance (O&M) strategies. This project is funded under the ULCS PV Research and Development Round and will build on ARENA’s previous investment into solar PV that support projects aligning with ARENA’s ‘Solar 30 30 30’ target to improve module efficiency to 30 per cent and reduce total construction costs of utility scale solar farms to 30 cents per watt by 2030. Funding is being made available to focus on commercialisation prospects, which will take place after an initial research and development phase, to assist getting the new technologies into the market.One of the main challenges in determining the degradation rate and extent of PV modules in the field arises from the standard test conditions under which nominal electrical parameters are measured in factories. Other challenges are associated with the use of different weather and irradiance data sources, and methods and metrics when calculating degradation rates of PV modules. Electrical and weather data is regularly collected by PV plant operators for forecasting purposes but has not yet been utilised to improve plant performance. The project will address these challenges by using the large datasets already collected by the project participants to develop improved statistical methods and performance metrics to determine accurate degradation rates. The project will develop an:Automated decision-making platform that can be integrated into monitoring systems.Australian-wide PV performance database containing site information, weather data, and electrical measurements.The project aims to undertake commercialisation activities including the development of a user-friendly product interface that combines the outcomes of the earlier work packages into a commercial product that automates the O&M decision-making process. The project aims to improve the technology readiness and commercial readiness of solar PV technologies, through improved reliability of utility-scale PV power plants. It will reduce the cost of renewable energy through the reduction of O&M costs through automated maintenance technology and intelligent plant monitoring systems, contributing towards the stretch goal of $15/MWh. The project will increase the value delivered by renewable energy by lowering the construction costs of PV plants in Australia by utilising site-specific data to improve the design and performance of utility-scale plants.

High