Industry: Energy & Utilities
Our client stands as one of India’s most significant integrated power enterprises, with a presence spanning the complete spectrum of power-related activities. This includes conventional and renewable energy, power services, and forward-looking customer solutions such as solar rooftop installations, electric vehicle (EV) charging stations, and home automation. Collectively, including its subsidiaries and joint ventures, the power company possesses a total generation capacity of 12,772 MW, with 30% of this capacity originating from environmentally friendly and sustainable sources.
Business Challenges
The client used manual testing for solar panels during the quality check, requiring QA engineers to examine each panel individually.
They sought a methodology to eliminate manual intervention, optimize time and space management, and automate object detection.
Manual intervention led to human errors and raised cost-effectiveness concerns in business operations.
Solutions
Data was uploaded from the operator to the datastore, flowed through Azure Data Factory to Staging Blob, triggered an Azure function, and reached an AKS-hosted pipeline (cell splitter, inference modules, and custom vision models) for processing.
Custom Cell Split provided cell coordinates, and the inference module processed binary images to generate results and create a final JSON.
The final inference JSON of the panel image was stored in Cosmos DB and utilized by the UI, containing the blob path of the panel image, cell coordinates, and complete inference results.
Technologies Used
Business Impacts
The collaboration between the prominent power company and Celebal Technologies to automate solar panel inference with AI technology has been a resounding success. By implementing cutting-edge technologies, the power company has improved its solar panel infrastructure’s efficiency and sustainability while reducing operational costs.
