The main aim of the proposal is the development of a commercial product that will assist in maintaining an optimal level of operation of PV plants. In particular, the proposed project is concerned with advancing the field of the automatic identification of performance loss, degradation and failures in monitored PV plants and their classification into various fault types and degradation mechanisms, which are manifested in the field.Specifically, innovative methodologies based on machine learning and statistical analysis will be developed for identifying performance losses and failures in PV plants without disrupting their operation. Such methodologies have the potential to contribute greatly to new standards on PV performance, degradation and reliability.

The scientific and technological objectives of the project are to:

  1. Collect high quality meteorological and PV operational data from fielded PV plants.

  2. Develop data quality routines to inspect and/or filter the monitored measurement data.

  3. Develop an optimum methodology for treating missing and erroneous data.

  4. Perform a sensitivity analysis on the impact of the equipment/sensors’ uncertainty on the estimation of performance indicators of PV systems.

  5. Investigate the impact of different climates on PV degradation rate estimation.

  6. Develop a new methodology for the estimation of degradation rate by taking into account the non-linear trends.

  7. Compare the different statistical techniques (Linear Regression, Principal Components Analysis, Classical time series decomposition, Year on Year, Change point analysis and ARIMA) used for performance loss rate calculation, by applying them on grid-connected PV technologies installed in Cyprus and Israel.

  8. Develop innovative methodologies to detect and classify performance loss and failures from the field data in PV systems.

  9. Test and validate the methodology for detecting and classifying losses and failures on PV systems at the UCY and at least a dozen of other PV plants.

  10. Integrate the developed models into a commercial product (Raycatch) for optimal operations and maintenance (O&M) practices.

  11. Validate the tool at both locations (Cyprus and Israel).

The ultimate technological objective of the proposed work is to extend the capabilities of the Raycatch product. Raycatch is an AI diagnostics technology for solar energy, on a mission to revolutionize the PV market by enabling automated management of solar assets.

The extended tool will include data quality routines, an optimized methodology for the estimation of degradation rate and appropriate algorithms for automatic identification and uninterrupted monitoring of PV systems. As such, the early identification and classification of failures and performance loss mechanisms will be achieved. Therefore, corrective actions will be able to be taken by the corresponding owners or operators in order to safeguard the PV performance and minimize the investment risks.