Inverter began to fail frequently, and the root cause was unidentified.

The supervised ML models detected early patterns of inverter failure. Overheating issues were prompted as the probable cause. Our predictive intervention avoided one week time which it would have taken to find the root cause and fix the failure.

Some inverters, intermittently, were observed to produce less energy than the others.

Patterns of late inverter wakeup was detected due to RISO fault, due to moisture, which would turn off that string. Our predictive intervention avoided string disconnection and hence string shutdown time of ~3-4 days were saved.

Sandstorm covered the panels, production reduced by 62%. Client scheduled the site for cleaning 1 day after the sandstorm. This would cost EUR 20,000.

Our Dynamic Cleaning Schedule forecasted the rain (150mm) within 2 days would be enough to reach cleanliness effectiveness. The 3rd day from sandstorm, there were heavy rains at the site which was sufficient to get back to the PR level prior to the sandstorm.

Production loss of up to 18% for inverters, increasing over the months.

Our system utilised advanced pattern recognition algorithms to detect evolution of horizons profile. Our supervised ML models then identified pattern of vegetation and optimised trimming interventions based on projected vegetation growth trends.

Heavy penalty levied due to deviation in forecasted production values.

Advanced integrated approach combining models on plant’s health prediction and long-term climate change scenarios deployed to forecast resources and production with 99% accuracy.