Maintenance is what keep the industries moving. Our goal is to make maintenance as easy as possible. That is why we envision a new approach to condition monitoring and maintenance with the use of Machine Learning.
Our setup uses an array of multiple sensors that are connected to a central unit, which is connected wirelessly to a cloud network. The maintenance supervisor receive notifications thought any wifi enabled device that uses Android, IOS or Windows. The cloud network tracks the behavior of the machine and recognizes what is normal behavior from defective, from analyzing and comparing the inputs from the sensors.
When a customer purchases the setup, he gets:
-Set of sensors (vibration, temperature, voltage, gas and liquids)
-Cloud network user login
How it predicts?
Machine "A" has certain behavior during normal operation. When something is wrong, the behavior has voltage peaks, increased heat generation, irregular vibration and depending on the machine, leaks and gas generation.
Our algorithm analyzes those changes and creates a flag, delivering a notification to the maintenance department wireless devices, with an action menu, where the user can shut down the machine, flag the event as normal or notify that he will check.
Once the event is cleared, the algorithm records the data generated after the action is taken. If the user reports that the machine "A" had a failure of the component "X", then the algorithm now is ready to make a more exact prediction.
If the same behavior is repeated, the notification delivered to the technician suggest that based on previous events component "X" is failing with certain percentage of reliability in the report and creates a shortcut to generate an automated purchase order of component "X"
In the cloud data, the algorithm catalogs the data depending on the machine manufacturer, enabling us to provide this data to the manufacturing so they can develop better services and future updates of those machines.