Reliability Evaluation Through Analytics and Data for Emerging Technology in Photovoltaics

ML Tool

Virtual Access

  • Adaptive multiple linear regression with online learning
  • Weather inputs (T, irradiance) → I-V parameters
  • Degradation rate & metastability monitoring
  • Validated on PSK/CIGS tandem

Successful application of this tool has already been demonstrated for perovskite/CIGS tandem technology. Using simulated weather data, degradation evolution may also be projected in advance, with predictions later validated against real measurements as they become available. By tracking actual device performance under continuously shifting environmental conditions, the model supports assessment of meta-stable behaviour and calculation of degradation rates. Training the model requires adequate outdoor data—even a single device can sufficent, provided its dataset includes a stretch without significant degradation trends. Overall, the tool enables ongoing monitoring of device degradation over time.