AWS Industrial Predictive Maintenance Configure AWS IoT Greengrass using

1
Machine Learning Model and Anomaly Detection Reference Architecture Greengrass Connectors to
communicate with factory machines.
Create a Predictive Maintenance (PdM) Machine Learning (ML) model using Amazon SageMaker with AWS IoT Core
and an anomaly detection application using Amazon Kinesis Data Analytics. Configure rules within AWS IoT Core
2 to trigger events based on MQTT
topics for the factory machines.
Factory AWS Cloud 4 Create an Amazon Kinesis Data
ML 3
Inference Firehouse delivery stream to store the
factory machines data in the data lake
on Amazon Simple Storage Service
5 (Amazon S3).
Amazon
SageMaker 4 Build your Predictive Maintenance
(PdM) Machine Learning (ML) model
2 6 7 with Amazon SageMaker.
1 Lambda function
3 Deploy your Machine Learning model
5
onto your AWS IoT Greengrass Edge
Gateway.
Factory Machines AWS IoT Core Amazon Kinesis Amazon S3
AWS IoT Greengrass Amazon Athena Amazon Build your data queries in Amazon
Data Firehose QuickSight 6
Athena against your AWS Glue Data
Catalog of the data lake on Amazon
S3.
Visualize your analysis using Amazon
7
8 9 QuickSight on the Amazon Athena
Amazon Kinesis data source.
Data Firehose
Create an anomaly detection
8 application in Amazon Kinesis Data
Analytics.
Amazon Kinesis Amazon Kinesis AWS Amazon SNS
Data Streams Data Analytics Lambda Configure AWS Lambda as an output
9 of Amazon Kinesis Data Analytics
application to send anomaly
detections notifications to an Amazon
Reviewed for technical accuracy March 24, 2021
Simple Notification Service (Amazon
© 2021, Amazon Web Services, Inc. or its affiliates. All rights reserved. AWS Reference Architecture SNS) topic