- Collect IoT data from sensor
- Change point detection using IoT sensor data
- Predicting equipment failure using IoT sensor data
- Sending decisions based on analytics insights to the edge for automated action
The data collected from an IoT system can be used for automated action by applying advanced analytics. Some of the advantages of applying advanced analytics on IoT data is that the traditional method of scheduled maintenance can be replaced with predictive maintenance.
The applications of advanced analytics ranges from simple to complex in nature depending on the application area and outcome. A low complexity univariate analysis can be applied on the time series data of a single sensor to determine if a change point has occurred in the sub-system and if the system has departed from its normal operating range (see the “Detect change points in IoT sensor data” code pattern).
This symptom can be followed up by a more complex multivariate analysis that utilizes random variables from multiple sensor readings to predict if a system failure is likely to occur (see the “Predict equipment failure using IoT sensor data” code pattern).
Once a failure is predicted to be more likely, an automatic action can be triggered to take corrective actions. Such an automated action requires the outcome of predictive models be translated into an action step and then this action translated into a command to be sent to an actuator attached to the edge.
The “Take corrective actions based on predictive analytics of IoT sensor data” code pattern focuses on demonstrating the method for implementing the round tripping of data that involves these steps in an IoT advanced analytics system:
- Collect data from the edge layer by using an attached sensor
- Transport this data to the analytics engine in the cloud
- Run predictive analytics models on this data in the cloud to determine if a corrective action is required
- Determine the corrective action
- Translate this corrective action into a command to be sent to a specific actuator attached to the edge
- Send the command back to the edge layer where the actuator is attached
This composite pattern will help Data Scientists, Solution Architects, and System Integrators to implement a template for a full-fledged advanced analytics sub-system in their IoT solution.