Sensor Identification

Sensor Identification Module

This module contains a class for choosing which irradiance sensor best describes a PV power or current data set. We assume a linear model between irradiance and power/current, and we use k-fold cross validation to assess which irradiance sensor provides the best predictive power.

Generally speaking, we can try to assess a sensor’s distance from an array and its plane-of-array mismatch. Hopefully, there exists a sensor that is both close by the array and well aligned; however, this is not always the case. We use clear sky data to assess POA mismatch and cloudy sky data to assess distance from array. If there is a discrepancy in which sensor is “best” under these two data filtering schemes, the algorithm alerts the user.

class solardatatools.sensor_identification.SensorIdentification(data_handler_obj)

Bases: object

identify(n_splits=20, model='least-squares', epsilon=1.35, max_iter=100, alpha=0.0001)

solardatatools.sensor_identification.rmse(residuals)