Proactive Error Detection System

Operational Logic

After 7 days of machine learning, the system can start forecasting the normal generation curve, and proactively notify the designated maintenance personnel via mobile phone or email at the moment the error is detected and diagnosed.

The Result of Applying the System

By using AI machine learning to estimate the power generation and voltage status of each device in the power plant, the system may diagnose whether the equipment has inverter failure, fuse burnout, heat drop, shading, etc., and send reports to facilitate early repair, adjustment, or other improvement.


This system has been applied in more than 300 solar photovoltaic sites across Taiwan. The performance results from 2018 to 2020 showed that the accuracy rate of the default system is 98.6% and the error type correctness is 94%. The power generation of the sites applying the system has increased by around 4.7%, and reduced the transportation cost of maintenance by 41%.


The operational effectiveness of this intelligent debugging, diagnosis and dispatching system has been proven in practice, that the team has been invited to present at international technology research forums and trade shows in the industry during 2019-2021.


  • The world's largest solar photovoltaic forum "EU Photovoltaic Solar Energy Conference" ranked our research 23rd among the 900+ papers in various fields of solar photovoltaics in the world in 2019. The team was invited to have oral presentation in 2019 & 2020.
  • In 2021, the team was also invited to share research results online at "Intersolar", the world's largest solar energy trade show.

The Research Papers

The Service Process

Search

You can enter the keyword or select one of the common searches for convenience

Common Searches

+