The University of Queensland (UQ) has developed a new software named Solarisᴬᴵ that it claims can analyze data and detect faulty and underperforming solar panels with machine learning algorithms, without the need for additional hardware on site. 

“The challenge with large solar farms is detecting any faulty or underperforming solar panels hidden in a sea of millions,” explains the developer of the software, Associate Professor Dr. Rahul Sharma. “It’s impractical to install monitoring hardware on each panel, inspect every panel for damage or clean every panel to remove dirt. We needed to find a way to automate that process.” 

Developed by the UQ’s School of Electrical Engineering and Computer Science, Solarisᴬᴵ works as a software-as-a-service (SaaS) operations and maintenance (O&M) system for solar farms and large-scale PV installations. 

It was founded by UQ’s commercialization company UniQuest, spearheaded by investment from Uniseed as well as the UniQuest Investment Fund. 

Sharma says Solarisᴬᴵ works at the array and string panel level. It sequentially extracts vital information of the panel’s health, along with monitoring it for degradation, soiling, wiring faults and tracker problems. It alerts for underperforming panels. 

According to Sharma, underperformance in Australian solar farms cost the industry around AUD 400 million/year. With Solarisᴬᴵ, he explains, these losses can be reduced to half. It can also deliver up to 8% improvement in revenues. 

The software was deployed to monitor the 3.3 MW Gatton Solar Farm that hosts 72 strings, each with 15 PV panels for a total of 1,080 panels. Solarisᴬᴵ was able to identify underperforming strings and panels within the arrays ‘uncovering 5-10% reduction in power production’ due to soiling. 

Now the university is in talks with Edify Energy’s Hamilton Solar Farms at Collinsville and Genex Power’s Kidston Solar Farm.