Plausibility Filtering of PV Outdoor Data

High-quality input data is necessary for calculating performance loss and predicting energy output and the lifetime of photovoltaic (PV) modules. Therefore, filtering PV outdoor data for unplausable measurements is crucial for analyzing PV outdoor performance. Photovoltaic outdoor data consists of electrical measurements (e.g. complete current-voltage (IV-) characteristics or single performance parameters), and meteorological data (e.g. irradiation and temperature). Here, the various electrical and meteorological measurement all constitute different dimensions in the measured data. Currently there is no standard for outdoor PV data filtering. However, commonly data filtering is based on simple thresholds in single dimension of the data. Since thresholding usually results in information loss, we propose using a plausibility filter, which simultaneously considers the various dimensions in the data. To this end we use well known correlations between the various dimensions in the data to compute a measure of plausability by means of the Mahalanobis distance. In this work we demonstrate this concept by combining the solar cell IV parameters, module temperature, and various irradiance measurements (plane-of-array, global-horizontal, and diffuse horizontal irradiance).