Solar Inverter Outlier Analysis
Key Technologies: Databricks, Spark, Machine Learning, Python, SQL
Our client needed to validate the accuracy of solar inverter data and identify any important outliers during operations which might degrade performance on the array. Any problems needed to be monitored and reported to key stakeholders.
Adinkra developed several scripts to automatically ingest and detect outliers across hundreds of inverter devices. We also created multiple event-based alerts for monitoring, and compelling graphics to visualize availability and production data.
Our models helped the client create new alerts that could be used to monitor and identify problems quickly across hundreds of assets. Importantly, the client was able to now hold data providers accountable.