Advancing Photovoltaic Power Plants

(PVPPs) operation efficiency

Problem

Loss of PV generation

due to lengthy diagnostics of technical failures

  • Breakdowns
  • Loss of time
  • Loss of generation
  • Loss of revenue
  • PVPP equipment fails from the first years of operation, and more failure cases with each subsequent year of operation. PVPP with a total capacity of 150 GW have been installed in Europe more than 3 years ago!
  • Diagnosing technical failures and problems is not carried out daily and may take from 5 days to several weeks
  • Shortage of qualified experts and high cost of person-hours in the field of PVPP diagnostics
  • Lack of equipment operation prediction to ensure smooth operation of the PVPP
  • Loss of PV generation, non-fulfillment of obligations under electricity supply contracts, fines

Solution

Software based on neural networks predicts the equipment failure probability with high accuracy and prevents possible losses of PV generation

Product

PV.Predictive Maintenance based on AI and ML

Prediction at all levels of PVPP operation:

  • Module of prediction and detection of the solar combiner box (CB) and string failures
    (exist)
  • Module of a neural network for the CB and string failures classification (detection)
    (to be developed)
  • Module of predicting failures of inverters
    (to be developed)
  • Module of predicting failures of power transformers and cable lines
    (to be developed)

How it works

Operation algorithm:

Data collection and storage of equipment operation parameters on reliable Cloud Storage with mirrors

Data processing using mathematical algorithms

Analysis of received data using AI and ML

Notifications via various secure information channels or on specialized web interface

Why Now?

    “Fit for 55” & “RePower EU” goal

  • Plan “REPowerEU” is a response to energy market disruption induced by Russia’s war against Ukraine. The main focus of the plan is based on the development of renewable sources and increasing efficiency;
  • There is 209 GW installed capacity of PVPP in the EU and it will be scaled to 480 GW till 2026*. KNESS estimates PV plants’ generation losses due to lengthy diagnostics of technical failures in the EU at 1 billion euros per year.

    • *according to the EU Market Outlook by

    Solar Power Europe

    Expert system

  • Knowledge and experience of the team of experts in the PVPP O&M sector which has been gained over the last 8 years (O&M for 135 PVPPs with a total capacity of 1,4 GW);
  • 6 terabytes of data on PVPP equipment operation have been collected during the last 3 years of PVPP operation which improves the performance of neural networks.

    The 4th industrial revolution - AI

  • The 4th industrial revolution is on the threshold. Artificial intelligence (AI) is the next major technological revolution following the growth of mobile and cloud platforms;
  • “The AI industry will contribute an additional $15.7 trillion to the global economy in 2030. But to fully reap these benefits, companies must put these tools into practice now” (Forbes).

Advantages

Advantages of PV. Predictive Maintenance

Our competitive Advantage:

Existing PV plants monitoring systems can detect standard failures but their resolving accuracy accounts for only approximately 10%.

We work not only with panels, strings or СBs, but also predict and carry out diagnostics of all equipment at the PVPPs (inverters, transformers, power cable lines)

  • Saving time
  • Ensuring maximum PV generation
  • Getting maximum revenue

Main Benefits of our solution:

  • Intelligent system: an ongoing development process due to constant learning and self-update of the neural network;
  • Expert system: knowledge and experience of the team of experts in the PVPP O&M sector;
  • Working Ahead: predicting equipment failure using AI and ML;
  • Prevention of PVPP generation losses;
  • Improving the quality and efficiency of PVPP maintenance;
  • Simple integration with various software complexes owing to the АРІ protocol;
  • Optimising the safety management of equipment and systems during their entire lifetime;
  • Reducing time for repairs and optimising maintenance and Spare Parts Management costs.
START 2021

Current Status

Project Timeline

ІI quarter of 2021

PV.Predictive

Maintenance v1.0

MVP

The developed system for predicting and detecting technical defects of CBs and strings:

  • The software module with the specified features is implemented into the PV software SCADA.
  • Dispatchers can review neural network forecasts regarding the technical condition of the equipment on a daily basis.
  • The system automatically notifies employees responsible for the PVPP efficiency about problems via the secure messenger.

ІII quarter of 2021

Software’s pilot use

by a service company

The developed system has been piloted for more than a year in servicing of more than a hundred PVPPs with a total capacity of 1.4 GW by the largest private service company in Ukraine - KNESS Service.

The software module for predicting and detecting technical faults of CBs and strings significantly facilitates the operation of solar power plant equipment, optimizes the use of human resources to detect unwanted deviations from the most efficient operation of the equipment, and minimizes the average response time to technological faults.

- Viktor Terletskyi, Technical Director of KNESS Service

PVPP “Pohrebysche” 9,9 MW

Module of prediction and detection of the solar combiner box (CB) and string failures in use

Number of strings

1951 pcs

Annual output

13 609 MWh

(1200 Full load hours)

Annual revenue

€ 2,1 million

(0,16 € kWh)

Commissioned in

2019

15 strings that did not operate properly for various reasons (blown fuse, damaged cable, etc.) have been detected

15 strings is approximately 105 MWh of generation losses annually

105 MWh in Ukraine is equivalent to € 16,800 of losses per year

Adding to the previous figure, the cost of thermal imaging inspection in Ukraine for this facility is 300 Euro/MW, so we get €16,800 of costs preserved and € 3,000 of saved costs.

The economic effect for the power plant is €19,800 (1% of revenue) annually

Next steps

Project Timeline

ІV quarter of 2023

PV.Predictive

Maintenance v1.1

To be developed: a software module for the CB and strings failures classification

ІI quarter of 2024

PV.Predictive

Maintenance v1.2

To be developed: a software module for predicting inverter failures

ІII quarter of 2024

PV.Predictive

Maintenance Pilot

The pilot launch of PV.Predictive Maintenance (test usage)

ІV quarter of 2024

VPV.Predictive

Maintenance v2.0

Software located in a “cloud” platform with WEB interface for demonstrating the conclusions of the expert system

Each software module will be tested

in the real business conditions!

The Team

Our team has more than 10 years of experience in software and hardware development of diverse complexity and has been actively making use of AI and ML technologies over the last 2 years.

Bohdan Kozachuk

Product Owner

Vadym Dzhyzhula

Software developer

Yuriy Pugachov

Software developer

Maksym Kyrbaba

Hardware developer

Pilot

Pilot PV.Predictive Maintenance v 1.0 byAdvance your PV plants operation efficiency!

in the PVPPs operation

PV.Predictive Maintenance v 1.0

  • The software module with the developed system for predicting and detecting technical defects of CBs and strings in accuracy - 85%
  • Neural network forecasts regarding the technical condition of the equipment on a daily basis
  • The automatic notifications about technical defects of CBs and strings

Since 2009

1,2 GWp installed SPPs

1,4 GWp SPPs under O&M

54,6 MW own SPPs

www.kness.energy