Improved Accuracy for Anomaly & Grid Failure Forecasting with Predictive Analytics
About the Client:
Our client is one of the world’s leading embedded technology service providers
Client uses state of the art automation systems to capture images of insulators/transformers installed at sub stations to monitor dust deposition on them. High dust deposition leads to rise in the insulator temperature and grid failure, thus resulting in revenue losses. To address this business challenge, there is a need to:
- Fetch and manage video feed data from 0.3 million towers
- Build an analytics ecosystem to trigger preventive cleaning alerts on time to avoid grid failure
TekLink data scientists identified the challenges with the storage, access and analysis of the vast amounts of data that is created. Keeping this mind and also reckoning the complexity of factors that the grid functioning is dependent on; TekLink team implemented a holistic solution to cater business needs. This included:
- Pre-processing (i.e. image scaling, background noise removal etc.) the real-time data & images as well as offline data (manual data, image feed, video streams) and it’s migration to data warehouse.
- Data training with predictive models.
- Building Deep learning algorithms like convolutional learning(CNN) using TensorFlow to categorize the clean and dirty images of the insulators.
- Implementing a smart predictive system to provide regular updates and critical warnings to service engineers for preventive maintenance.
Outcome and Benefits:
Leveraging the implemented analytical solution, client was able to:
- Efficiently process and categorize the insulator image data feed
- Minimize consequential losses due to power supply downtime
- Improve the preventive maintenance with the cleaning alerts generated from analytical models
- Leverage AI in data processing, thereby reducing error probabilities eliminating human judgement or intervention
Contact Us to learn more about this experience.