Automating Wind Farm Maintenance Using Drones and AWS AI at the Edge
Wind farm owners face high-risk and expensive turbine maintenance operations, estimated to cost over $40 billion across a decade. Traditional rope-based inspections pose safety risks and inefficiencies. AWS customers in wind energy and power utilities implemented an AI-powered drone inspection solution to improve safety, reduce costs, and increase inspection quality. The solution uses Amazon SageMaker to train ML models in the cloud for detecting hazards like corrosion, wear, and icing on turbine blades. The models are optimized with Amazon SageMaker Neo and deployed at the edge using Amazon SageMaker Edge Manager and AWS IoT Greengrass on NVIDIA Jetson devices. Amazon Augmented AI enables a human review loop for continuous model improvement, with integration across AWS IoT Core, AWS IoT SiteWise, AWS IoT Analytics, AWS IoT Events, AWS Lambda, Amazon SNS, Amazon S3, and Amazon QuickSight to manage telemetry data, events, notifications, storage, and reporting. The automated inspection reduces operational costs by up to 70%, downtime losses by up to 90%, and keeps safety high by enabling offline drone-based inspections with expert validation.
- Organization
- AWS customers in wind energy and power utilities
- Industry
- Energy & Utilities
- Location
- United States
- Published
- July 2021
Reported outcomes
−90%
timeTime & speed
Strategic outcomes
Catalog median for time & speed deployments: −60% across 727 reported metrics. Compare benchmarks →
Primary read
Use case focus
Showing 3 of 3
- 1Predictive Maintenance
- 2Edge AI
- 3Human-in-the-loop AI
- Developed an AI/ML drone inspection solution utilizing multiple AWS services to train and deploy models for edge inference.
- Used Amazon SageMaker suite for labeling, training, optimizing, and managing ML models.
- Deployed ML models on edge devices powered by AWS IoT Greengrass to enable offline and continuous operation in remote wind farm scenarios.
- Integrated a continuous learning loop with Amazon Augmented AI for expert human review and model retraining.
- Connected telemetry and event data streams through AWS IoT Core, SiteWise, Analytics, and Events for comprehensive turbine monitoring.
- Implemented serverless event-driven architecture with AWS Lambda, Amazon SNS, and Amazon S3 for data processing and notifications.
- Provided business intelligence and visualization through Amazon QuickSight dashboards.
- Achieved up to 70% reduction in operational maintenance costs and 90% fewer downtime losses compared to rope inspections.
- Enhanced safety through drone-based and offline inspections on rugged terrain.
- Improved inspection quality with human-in-the-loop AI continuous learning.
- Enabled scalable, automated, and cost-efficient turbine maintenance workflows.
Architecture
The architecture includes ML model training and labeling via Amazon SageMaker, model optimization with SageMaker Neo, and deployment on NVIDIA Jetson edge devices running AWS IoT Greengrass. The edge devices run local Lambda functions for inference, triggering human review workflows with Amazon Augmented AI when needed. Telemetry data from turbines is ingested through AWS IoT Core into SiteWise and Analytics for monitoring and alerting via IoT Events and notifications through Amazon SNS. Business intelligence dashboards use Amazon QuickSight.
Sources & evidence1
AI-generated summary. Verify important details with the linked sources before relying on this case.
Explore related AI use cases
Was this useful?