Although there have been numerous efforts in this direction, there is still no common approach that allows a systematic classification of faults independently of the monitored system. In this context, the presented method shows an alternative solution for fault detection based on a dimensionless approach which allows simpler anomaly detection for different types of systems. Plant management and maintenance companies, Asset Managers and entities involved in research on photovoltaic systems are to be considered the main recipients of the proposed approach. However, to be economically competitive, a monitoring and fault finding activity should be conducted with the smallest number of measuring instruments possible and should not require complex customizations that depend on the characteristics of the photovoltaic system. This work reports on fault detection conducted on a dimensionless dataset obtained starting from data imported directly from the SCADA system. The proposed approach allows the working points to be classified into clusters, each representing a different operating or fault condition. Based on appropriately chosen thresholds, the method allows obtaining a rapid classification of the working points. The density of points grouped within each cluster represents the weight to be used to determine the state of the system. Preliminary tests conducted on various photovoltaic strings gave positive results, highlighting the characteristics of the method: delocalized, scalable and affordable.