Can we detect/read barcodes on panels?
We have tested this in the past and concluded todays technology doesn't allow the bar codes to be read. Our current default pixel resolution on of our systems on RGB is 1cm/pixel, to read the barcode you would need 0.1cm/pixel which is possible but it would drastically increase the flight time to a point where it is not commercially viable.
Is Sitemark IEC compliant?
We use IEC standards as a basis to develop our platform. IEC was written for handheld inspections, therefore there are a few elements that are impossible to comply with when using a drone.
A list of the elements we are not compliant with:
Geometric resolution (IEC Page 9): In the standard is written that you need a maximum Geometric resolution of 3cm/pixel. The reason you want a high resolution is that you need to have sufficient measurements/temperature samples of every cell on the solar panel. The average solar cell size is 16x16cm, meaning that the IEC standard wants you to have about 28 measurements per cell to make sure if you have some drifting pixels in your sensor the measurement is not compromised. At Sitemark we fly at 35 meters resulting in a resolution of 4,5cm/pixel or about 12 measurements per cell. However because our system combines thousands of thermal pictures into one orthomosaic it allows us to have about 16 photos of every point on the site. Thus our system has access to 16x12 measurements resulting in an average of nearly 200 temperature measurements per 16x16cm solar cell. Finally we also offer the possibility to fly at 23 meters to comply with the 3cm/pixel resolution. However this would multiply the flight times by 2,5 and therefore more than doubles the cost of acquiring the data. We tend to use this option when our customers have specific rare cases such as difficult warranty claims etc.
Ambient conditions (IEC Page 11): At Sitemark we use satellite based data coming from Solargis for irradiance, temperature & wind speed to balance operational efficiency (cost) and practicality.
Are you able to detect micro-cracks?
Typically they do not show up with IR scans unless the damaged cells are causing a more significant impact on production that would result in the generation of heat.
How do you calculate the mean temperature of the panel?
Every panel in the orthomosaic consist of a set of pixels (defined by the drawn boundaries), each pixel has its own absolute temperature. Each panel has about 300 to 500 pixels in it.
The mean is calculated from this set of pixels, but with the following distinction:
In panels marked as Hot Spots, we exclude 10% of the hottest pixels from the mean generation (to make sure we don’t take the hot spot into account in the mean)
In panels marked as Multi Hot Spots, we exclude 33% of the hottest panels
For the other panels, we leave them mean to be calculated on the full distribution
How accurate are the temperature measurements of the platform?
The temperature measurements on the platform come directly from the camera itself. Absolute temperatures are heavily reliant on changing weather conditions, wind, thermal drift in the sensor, etc. These elements and the accuracy of the temperature measurements are related to the hardware used to acquire the measurements. You will be able to find these specifications under the manufacturers specification document.
We can provide some insights on what we have found and what you can expect from the temperature measurements. In practice, the relative temperature differences are more accurate than the absolute ones. We can count on about 0.5℃ accuracy on the relative ones (difference between two points close to one another), and up to 5℃ in absolute temperatures (the temperature of one specific point). This of course only applies on the solar panels, the absolute temperature measurements on other elements may be very off