At A Glance
The client is a solar O&M specialist in the US working with large solar power producers across north america. Project objective included evaluation of 556 rooftop solar sites across 12 states in north, eastern united states and california state. The installations consisted of parking lot canopies, shopping malls, logistics hubs, universities and stadia. The plant power ratings varied between 0.5 to 1 MW.
The challenge was in measuring plant efficiency and faults at a time when physical access and movements were restricted on account of the Covid pandemic. Given the widely distributed nature of the assets, the client wanted a solution that could ensure a quick turnaround time with minimum human intervention.
Sensehawk was engaged for an automated, cost effective way of identifying solar modules that were non-functional. Sensehawk deployed it’s drone analytics solution which involved drone overflights which would cover the 556 sites. The partner drone operators in each state conducted the drone flights and the RGB camera images and IR sensor images were uploaded to the software.
The software had a digital model of the plant replicated from the design data. This was corroborated with the sensor data from the RGB cameras (visual) and a ‘digital twin’ of the design plant was created in the software. Post this, the drone’s IR (infrared) camera data was overlaid to generate a heatmap of the plant marked to the digital twin. Sensehawk’s AI engine auto-classified elements on the plant such as modules, inverters, combiner boxes, civil structures etc and linked it to the design element in the digital twin.
The thermography data gave the clients individual hotspots, temperature differences and the RCA (root cause analysis) derived from Sensehawk’s domain expertise in thermography. The defective module and the likely reason for the fault – down to the specific component (string failure, diode failure, wiring issue, vegetation etc) were available to the client in easy counts. The client was able to pan across the entire plant without having to worry about the tens of thousands of images (auto-stitched by sensehawk software to create the single thermal orthomosaic) and narrow down to the exact location of the faulty item.
|Drone flights||30 days|
|Daily coverage||~25 sites/ day|
|Execution||Sensehawk partners flew drone, imagery data was uploaded into Sensehawk|
|Defects identified||3,941 issues|
Sensehawk offered the client 3 levels of outputs:
The client identified over 3,941 defects which were most concentrated around module failures, bypass diode active and PID. The dashboard helped the client in both identifying which were the highest priority issues to fix first and also in directing repair teams to the specific module in question with exact geo location on site using the mobile app. The client was able to get the analytics without having to risk personnel in the covid period.