About the Customer
The client is a leading global manufacturer of industrial UV curing systems. These systems are vital to the production of thousands of items people use every day ranging from furniture and sporting goods, to semiconductors and medical devices. Rather than heat, this advanced technology utilizes microwave energy to generate intense UV light from gas filled electrodeless bulbs. The benefits include faster cure times and reduced costs while promoting environmentally friendly manufacturing practices.
Any failure to the UV light equipment during the manufacturing process can result in costly downtime, diminished product quality, or scrap. Equipment failures and anomalies that might affect product quality need to be predicted in near real-time, before they occur. To achieve this goal, the UV light equipment has been instrumented with an array of over 100 sensors. Each machine is connected to an IoT (Internet of Things) system that transmits the data to the client cloud. The system collects and sends over 200 metrics per machine several times per second. The challenges include:
- Ingesting, cleaning, and processing millions of time series data records
- Fusing instrumentation data with contextual data such as external temperature and factory conditions
- Storing health events in a database for profiling, viewing and developing models
- Developing baseline metrics
- Developing business rules for identifying anomalies in near real-time
- Developing machine learning models for predicting failures well before they occur
- Visualizing data for identifying anomalies and identifying correlations between multiple features
- Designing dashboards for service engineers and operators
Capitis implemented a data analytics platform using compute, storage, database, analytic, and data visualization technologies. The solution included the following:
- Amazon S3 object store as a data lake for storing ingested and processed data
- AWS Lambda (Serverless architecture) for data processing
- Amazon RDS for querying timeseries data
- Amazon EC2 for R Studio
- AWS ML platform for machine learning models
- Amazon QuickSight for online data visualization
- R and python libraries for advanced data analysis and visualization to establish baseline metrics
Capitis was able to establish baseline metrics, develop business rules for anomaly detection and create machine learning models for predicting failures well in advance of their occurrence.
By leveraging AWS infrastructure and platform services Capitis data scientists and software developers assembled a system to ingest data, process the data through machine learning models and visualize the results quickly and cost effectively
Capitis Solutions and AWS
Capitis team is building and managing a highly scalable and reliable IOT analytics platform on AWS for an UV light equipment manufacturing company. The platform is expected to be a game changer for the UV light curing industry by providing advanced preventive and predictive maintenance advisories.