Identification of Best Plane-of-Array Irradiance Sensor for PV System Performance Analytics

As the amount of available photovoltaic (PV) system performance data increases, there is an increased need to develop automated tools to handle data preparation tasks historically performed by hand. One such task is determining the best irradiance sensor to use as a reference when multiple sensor data streams (or derived/modeled data streams) are available. We present an algorithm for identifying the best irradiance data stream to use, given a PV system power generation signal. This algorithm is based on the concept of cross-validation from the discipline of statistical learning, and is implemented in the Solar Data Tools Python package.