Laboratory Studies of Particle Cementation and PV Module Soiling

Research output: Contribution to conferencePaper

2 Scopus Citations


Monitoring of photovoltaic (PV) systems can maintain efficient operations. However, extensive monitoring of large quantities of data can be a cumbersome process. The present work introduces a simple, inexpensive, yet effective data monitoring strategy for detecting faults and determining lost revenues automatically. This was achieved through the deployment of Raspberry Pi (RPI) device at a PV system's combiner box. The RPI was programmed to collect PV data through Modbus communications, and store the data locally in a MySQL database. Then, using a Gaussian Process Regression algorithm the RPI device was able to accurately estimate string level current, voltage, and power values. The device could also detect system faults using a Support Vector Novelty Detection algorithm. Finally, the RPI was programmed to output the potential lost revenue caused by the abnormal condition. The system analytics information was then displayed on a user interface. The interface could be accessed by operations personal to direct maintenance activity so that critical issues can be solved quickly.
Original languageAmerican English
Number of pages4
StatePublished - 2018
Event2017 IEEE 44th Photovoltaic Specialist Conference (PVSC) - Washington, D.C.
Duration: 25 Jun 201730 Jun 2017


Conference2017 IEEE 44th Photovoltaic Specialist Conference (PVSC)
CityWashington, D.C.

NREL Publication Number

  • NREL/CP-5K00-67873


  • data collection
  • fault detection
  • Gaussian process
  • modbus
  • photovoltaics
  • Raspberry Pi
  • support vector machine


Dive into the research topics of 'Laboratory Studies of Particle Cementation and PV Module Soiling'. Together they form a unique fingerprint.

Cite this