Vasudha - solar panels with wind turbines in the distance

Vasudha

AI & IoT for Sustainability

Over the last decade, there has been a significant increase in the development and deployment of photovoltaic solar energy generation across the globe. As of 2020, nearly 105 countries have invested in Solar Energy with a total of 580GW of installed capacity.

Photo Voltaic (PV) modules are the commercially avaialble basic building block in the solar deployment. One PV module typically consists of multiple PV cells which are arranged in matrix form of 12×6 or 12×5 fashion. With the increasing capacity of PV modules, it is important to monitor the quality and performance of these modules during manufacturing time as well as during their operations. Due to various real-world conditions during the PV modules manufacturing process, there is a possibility of certain PV cells inside PV modules getting damaged or causing an abnormal behavior with a fall in performance. If these defects are identified during early stages of manufacturing process, the affected cells can be replaced, thereby effectively saving major losses in plant efficiency and performance as well as monetary losses and warranty issues.

Many of these defects are extremely small in size and are localized  to a small part of a cell. So, such defects are difficult to be detected with bare eyes or through imaging techniques like infrared imaging. Therefore, Electroluminescence (EL) Imaging is more commonly used practice in detecting faults and degradation in cells from an EL image of PV cell.

In this work we design and develop an “AI-assisted system towards quality assurance of PV modules“. The proposed system accurately classifies (accuracy > 96%) the faulty modules form the working ones, as well as correctly identifies the PV cell (inside the PV module) that contains the fault and the type of fault. It also helps in correctly detecting the location of the fault inside the PV cell.  The performance of this system is evaluated on open EL image dataset as well as on real-world manufacturing dataset.

This is a joint collaboration with Azure Global team.

Vision AI algorithm pipeline for detection of faults in a solar panel. The pipeline includes preprocessing of EL image, segmenting the solar panel into individual cell images and then running a classification algorithm to identify whether the cell is working or faulty with a specific fault condition.