AI powered Solar Panels Inspection
The Solar industry is booming and emerging as the fastest growing source of renewable energy. Like most other industries, efficiency plays a crucial role for the sustenance of a company in this field. A time taking and labour intensive process which consumes a significant chunk of operating costs are the Solar Plant inspections. Regular inspections ensure that potential defects that could reduce power yield are caught early and repaired.
Traditionally the inspections are carried out manually where electricians check modules for electrical soundness, PV module specialists check each panel individually using thermal cameras and technicians record all faults. This manual process can take 3 people more than a month to cover a Solar plant containing 200,000 PV modules spread over 250 hectares. In addition to the task being tedious and highly susceptible to human error, it often results in interruptions in power generation.
Inspections using Drones
Given the abilities of drones and the drawbacks of manual inspections, drones are a natural fit for inspecting solar farms. The use of drones in the Solar industry has taken off rapidly in the past few years. A drone mounted with a thermal camera can fly over pre-planned routes covering one section of the solar farm at a time and capturing thermal images along with geospatial metadata. The same 250 hectares can be covered by drones well within 2 days.
Inspecting solar farms using drones can be seperated into 3 phases:
I. Fly over the solar farm and collect thermal images along with geospatial metadata
II. Analyse the thermal images and locate faults
III. Generate actionable reports
While the first phase is highly automated using drones and is adopted by a large number of companies in this field, The same cannot be said about the second phase of analysing these thermal images. A large number of companies hire expensive skilled experts and have them manually go through the captured images looking for defects.
Nanonets APIs greatly boost efficiency in the second phase, in terms of both speed and accuracy. Nanonets makes use of state-of-the-art Image Processing/Computer Vision algorithms to build custom models which accurately identify and locate defects in each thermal image. These identified defects can range from disconnected panels, thermal hotspots to diode failures. Each model is fined tuned to custom data (thermal images) to ensure maximum accuracy. The custom models are capable of analysing entire solar farms within a couple of hours which would otherwise take weeks (for domain experts looking at images). This speed can potentially save millions of dollars in module replacement costs by identifying faulty panels faster within the warranty period.
Rate of Inspecting Solar Panel Images
Manually with domain experts
6000 PV modules/man/hour
With NanoNets API
200,000 PV modules/hour
Cost of Analysing Images
Manually with domain experts
After integrating NanoNets API