Digital, Data Driven Operation & Maintenance of PV Power Plants

Intelligent PV Monitoring

Performance Monitoring and Machine Learning Algorithms provide a robust basis
© Photo: Fraunhofer ISE, Overlay: metamorworks/Shutterstock
Performance Monitoring and Machine Learning Algorithms provide a robust basis for the early detection of malfunctions of systems.

Efficient and reliable detection of malfunctions and energy losses is essential for reducing costs and guaranteeing successful operation. By promptly identifying issues, timely corrective actions can be taken to minimize downtime and maintain optimal performance. This proactive approach helps to extend the lifespan of the PV system, maximize energy production, and ultimately increase return on investment for solar energy projects.


PV Systems Performance Monitoring

Fraunhofer ISE monitoring system is particularly suited for highly innovative PV projects and technologies: Floating PV, Agri-PV, infrastructure-PV, novel PV technologies (i.e. new PV module technologies, innovative PV systems design, etc.)

Our services comprise:

  • support in sensor selection
  • KPI definition
  • installation & operation
  • data evaluation
  • web-platform for data visualization (widget based, highly customizable)
  • secure data transport and storage on Fraunhofer ISE servers

Enhanced Intelligent Monitoring

Automated fault detection offers a dependable solution for identifying failures and energy losses in PV systems. By continuously monitoring system performance and utilizing advanced algorithms, it can quickly detect anomalies and pinpoint the source of issues. This proactive approach allows for swift corrective action, reducing downtime and ensuring optimal system operation.

For this purpose, multiple methods have been created to:

  • identify shading events using artificial intelligence
  • determine of operating states (i.e. open circuit, inverter outage)
  • compare current and historical Performance Ratio (PR)
  • assess modeled vs. measured energy
  • compare inverters performance (DC current, PR)
  • determine energy losses due to soiling
  • calculate Performance Loss Rate over long-term operation

These techniques have been thoroughly validated using a portfolio of more than a hundred commercial PV systems.

Development of artificial intelligence and deep-learning based algorithms

Within the PV O&M context, the development of AI and deep-learning based algorithms is a groundbreaking approach to enhance performance and efficiency. Fraunhofer ISE offers customized algorithm development as part of research projects tailored to clients' needs. These advanced algorithms analyze vast amounts of PV system data to identify patterns and trends, optimizing system performance and automating decision-making processes. By continually adapting and evolving, AI and deep-learning algorithms play a vital role in advancing solar energy operations and management.

Benchmarking Fault Detection Algorithms

Fraunhofer ISE can assess the client’s own algorithms by contrasting their results with those of other algorithms tackling the same problem

Data-driven Soiling analysis

Standard compliant on-site monitoring of soiling losses
© Fraunhofer ISE
Standard compliant on-site monitoring of soiling losses and time series analyses help identify monetary losses and the best time to clean.

Soiling detection in PV systems is crucial to optimize energy output, prevent losses, and extend lifespan.

Data-driven soiling determination

Using a variety of algorithms (either state-of-the-art from research or Fraunhofer ISE custom developments), experts examine the results to guarantee accuracy and trustworthiness. If necessary, the process includes obtaining and analyzing meteorological data or satellite images.

Sensor based soiling measurements

Our soiling measurement service includes designing a tailored measurement concept, selecting the appropriate sensors, and conducting thorough results analysis. This comprehensive approach ensures the most effective solution for project specific needs.

IEC 61724-1:2021compliant

On site soiling determination through PV array IV curve measurements

Fraunhofer ISE edited an IEA PVPS report on the topic: Soiling Losses – Impact on the Performance of Photovoltaic Power Plants - IEA-PVPS

Determination of best time to clean based on soiling results and economic consideration

Digital Twinning

Digital twinning
© metamorworks/Shutterstock
Digital twinning – the combination of a virtual representation of a PV system and it’s physical counterpart, who are exchanging data in near-realtime – offer the possibility of a multicriteria system behaviour analysis, thus supporting risk assessment and risk mitigation strategies.

The digital twin is based on Fraunhofer’s ZenitTM platform. This Python-based electrical PV simulation tool has been used on a day-to-day basis to provide scientific quality bankability analyses for project developers, investors, EPCs, and many different industrial partners along the solar power value chain. Furthermore, it is regularly developed and validated in research projects, showing a high degree of flexibility in terms of the types of solar power plants capable of being accurately simulated.

The aim of the digital twinning methodology is to provide a baseline performance calculation to help identify deviations from expected behaviour. Utilizing this information, unusual reductions in output power that might be linked to unexpected component faults or degradation can inform maintenance measures such as cleaning or repair. Furthermore, the twin’s forecasting module provides predicted yield for a coming 3-day period based on locally available weather forecasts.

PV Power Forecasting

PV Power Forecasting
© Satellite image: Meteosat Second Generation (EUMETSAT)
We suitably select and combine global weather model data, satellite images and measurements in the best possible way in order to forecast the solar irradiation for individual PV sites or entire regions.

At Fraunhofer ISE, we are working on the development of forecasting models for the reliable prediction of the solar power for time scales from a few minutes up to several days and areas from single locations to regions of many kilometers. For this, our team of physicists, meteorologists and IT specialists combines input data from various sources (cloud camera images, satellite-based data, numerical forecasts) into a variety of different forecast models.

Our services comprise: 


Christian Schill

Contact Press / Media

Christian Schill

Soiling Analysis, Digital Twinning

Fraunhofer ISE
Heidenhofstr. 2
79110 Freiburg

Phone +49 761 4588-5378

David Melgar

Contact Press / Media

David Melgar

Intelligent PV Monitoring

Fraunhofer SE
Heidenhofstr. 2
79110 Freiburg

Phone +49 761 4588-2203

Anna Heimsath

Contact Press / Media

Anna Heimsath

Photovoltaic Power Plants

Fraunhofer ISE
Heidenhofstr. 2
79110 Freiburg

Phone +49 761 4588-5944

Further Information on this Research Topic:

Quick Analysis for PV Power Plants

Troubleshooting in PV Power Plants

Yield losses?
Inverter failures?
System damage? 

Research Project


Cost-optimised operational management of PV systems over their economic lifetime

Research Task

IEA PVPS Task 13

Performance, Operation and Reliability of Photovoltaic Systems