Blog

ArrayModeler – III (ML based Power Plant Modeling)

ArrayModeler – III (ML based Power Plant Modeling)

  • On December 1, 2020

Apart from the detailed multi-location mathematical theoretical modeling of a PV site, UNITY offers a Machine Learning-based Modeling analytic that approximates Expected Power by using multivariate time-series prediction techniques.

Taking into account the historical behavior of each site along with current weather conditions, UNITY calculates the power the site is expected to have produced at each recording period. Data used in the algorithm goes through a curation process to ensure that the calculated Expected Power is of maximum accuracy and reliability. Comparing actual production with ML-expected helps to identify underperformance points that may otherwise go undetected.

Additionally, Array Modeler III is the basis for accurate energy loss estimation which can then be attributed to several factors from soiling to grid outages. It may be used as well in performance calculations for reporting and to calculate deemed energy for billing and accounting purposes.