Artificial Intelligence in Energy Industry: Image Processing System
Building an AI-powered platform that can detect the quality of the third-party provided electric poles’s images and process them for anomaly detection.
THE CUSTOMER
The customer is the primary electricity supply company for much of Southern California. It provides 14+ million people with electricity across a service territory of 50,000+ square miles.
BUSINESS GOALS
Across the U.S., there are 3,000 incidents of utility poles catching fire each year, many of which are caused by leakages and damages to a pole that went unaddressed for too long. Such damages can be hugely expensive when not dealt with: The average cost of a pole fire on a distribution line is $25,000. But it’s not just money that’s saved when pole fires are avoided — it’s lives and homes, too.
The energy company wanted to leverage AI technologies to enhance their equipment management, customer service and cost management. The idea was, using the AI and Drone technologies to avoid such above mentioned wildfires and the related damage.
The customer asked ThirdEye to build an AI model to build a fully-functional predictive maintenance system that can process thousands of electric pole images per day, identify and report defects, anomalies and potential hazards, then, utility maintenance teams can proactively address problems before they spark and potentially turn into fires.
THE SOLUTION
ThirdEye built an end-to-end platform that picks up third-party provided images of electric poles to detect the quality of the image for further downstream consumptions. Once the quality of the images were determined to be adequate, the platform channeled these images for anomaly detection and other downstream analytics.
ThirdEye developed an Advanced AI Image Processing system with features like collecting data (image format), data classification, object detection, result prediction and notification. The system was powered by AI, backed by machine learning technologies and complex algorithms. Our AI image processing drives the data-driven decision making program to achieve the goals for the customer.
The process steps were as follows:
- Process image format data sent by drone technology.
- Develop various AI models to detect various aspects of the image quality like structure, tags deciphering, occlusion, blurriness, angle etc.
- Once the quality of the images are determined to be adequate, channel these images for anomaly detection and other downstream analytics.
- Added functionality to detect & decipher tags on these electric poles.
- Various AI models were developed to recognize the backgrounds of the tags and then decipher vertically written alphanumeric characters.
- Developed master pipeline that processes all images, branches to appropriate data pipelines based on the types of objects detected in the image invokes appropriate AI models on the image persists all outcomes into appropriate folders for validations by humans.
Technologies Incorporated:
- TensorFlow
- Model Building, deploying and serving
- Google Vertex AI
- Google AI Platform
- Google Kubernetes Services
- Azure Data Factory
- Azure Function Apps
- CosmosDB
VALUE CREATED
The goal was to develop the complex machine learning models and meeting the performance/accuracy metrics. ThirdEye had successfully delivered the machine learning (computer vision) models to automate the image quality assessment for Field Asset Inspection (part of Wildfire Mitigation Program) and met model performance metric targets. The company deployed the models into production and started using the systems seamlessly.
Transforming Enterprises with
Data & AI Services & Solutions.
ThirdEye delivers Data and AI services & solutions for enterprises worldwide by
leveraging state-of-the-art Data & AI technologies.
