Predictive maintenance Industry 4.0
Customer Challenges:
- A Gas Pump can fail anytime without any warning, causing significant loss of production
- Ability to predict any possibility of failure is critical
Result:
- Real time monitoring and predictive failure analysis is expected to save 1500 hours of downtime resulting in savings of $100K per month.
- Prescriptive recommendation for informed decision
- 10 % reduction in downtime
CoreView Solution:
- Analysis of existing dataset which consisted of 5000 pump’s historical operating data for the past 2 years.
- Building a data model using Python/Spark/Cassandra pipeline
- Advanced mathematical models based on time series forecasting and classifiers
- Visualization of the results using D3 libraries
Other Considerations:
- Silos data and Scale of the data – pumps, logs, environmental, SOPs etc
- Parameters for pump failures are not well defined
