Krypton Cloud devised a unique approach to apply IoT strategies to close the data divide between modern and legacy application sources. The goal: more timely, accurate insights to improve utilization of capital-intensive industrial assets with timely, accurate information for operational intelligence. Krypton turned to CloudGeometry to create a scalable cloud-native architecture deployed on AWS, so Krypton’s customers could benefit by shifting from reactive to condition-based maintenance. That means better asset availability, avoiding unnecessary maintenance, and continued reduced costs.
Avoiding costly downtime for large-scale equipment infrastructure depends on how well remote operations works with data-intensive alert streams. The complexity of data operations at SunPower’s global Remote Ops starts with processing billions of data points per hour, ingesting streams from equipment types across hundreds of different manufacturers and collected by Krypton Cloud. At the same time, vital historical data needed to be reconciled across incompatible formats from legacy systems, to ensure analytics signals could be modeled with proper context. Using machine learning to drive AI diagnostics could give managers and field technicians alike a clear and actionable picture of the state of each piece of machinery and plant operation health as a whole.
Krypton Cloud took advantage of a range of CloudGeometry services and expertise to build a SaaS platform that could transform data integrated across their customers’ global operations. This model-driven analysis of patterns that maximizes equipment utilization and streamlines physical plant service operations. Krypton Cloud relied on CloudGeometry experts to transform and enrich data across a wider range of sources across geographically distributed availability zones. Signal data was also aggregated every 15 seconds via Spark micro-batch processing. The data flows extended to historic data, cached in a heterogeneous data warehouse infrastructure that runs Amazon Aurora for SQL data and DynamoDB for non-SQL data sets.
For example, Krypton customer Sunpower also harnessed from CloudGeometry dataflows, with data pipelines adapted to profiles of specific equipment types. The ML engine uses this to check device health data, discover anomalies and flag them to operators and technicians. Data Integration services provided by CloudGeometry generated continuously up-to-date datasets and fed a unified analytics dashboard to display health status for power plants across the planet.
With a data-driven real-time analytics pipeline at the heart of its global remote operations, Krypton Cloud successfully delivered machine learning for its customers to unlock the power of IoT for clear ROI. Operations data delivered reliably yields continuously up-to-date insights revealing equipment anomalies sooner and ensuring they are fixed at far less cost.
Continuous data flow management
Kinesis for Real-time data flow integration & management to rapidly recover from unexpected problems in source data
Change data structures and types with functional expressions, within a data flow, using S3 Athena to keep costs low
Operationalize Machine Learning
Managed data integration and ongoing workflow for model building, training, and deployment with AWS Sagemaker
Sunpower selects Krypton asset intelligence platform to manage 700 power plants around the world.Network World