Predictive and prescriptive analytics for systems under variable operations
A communication system and method that provides predictive and prescriptive analytics for a system running at an edge. In one embodiment, the communication system includes an architect subsystem configured to build, test and deploy a model based on sensor characteristics of the system. The sensor characteristics are from at least one of an operator input, a historical input, a specification input, and a subject-matter expert input. The communication system also includes an edge subsystem configured to receive said model and perform predictive and prescriptive analytics on sensor data from said system running on said model deployed at said edge.
1. A distributed analytics computer system to control the operation of a plurality of monitored systems, comprising:
an architect subsystem and an edge subsystem, said edge subsystem comprising a plurality of edge processing devices to be located proximate to said plurality of monitored systems, wherein:
said architect subsystem is operable to deploy analytic models to said edge processing devices based on sensor characteristics of said monitored systems; and,
each of said edge processing devices is operable to:
1) receive at least one of said analytic models and independently perform predictive and prescriptive analytics on sensor data from at least one monitored system, wherein said prescriptive analytics includes a future action on a component of said monitored system associated with said sensor;
2) provide control signals to said at least one monitored system according to said predictive and prescriptive analytics; and,
3) periodically send information to said architect subsystem, including monitored system responses to said control signals, whereby said architect subsystem can modify said analytic models to improve system performance of said monitored systems.
2. The distributed analytics system as recited in claim 1 , wherein said edge subsystem is configured to perform said predictive and prescriptive analytics on at least a component of said at least one monitored system based on said sensor data.
3. The distributed analytics system as recited in claim 1 , wherein said information includes a current status of a component of said at least one monitored system associated with said sensor.
4. The distributed analytics system as recited in claim 1 , wherein said information includes a characteristic of said at least one monitored system based on said sensor data from a plurality of sensors.
5. The distributed analytics system as recited in claim 1 , wherein said predictive analytics includes an estimate of an operation of a component of said at least one monitored system associated with said sensor.
6. The distributed analytics system as recited in claim 5 , wherein said estimate is the remaining operational life of said component of said at least one monitored system.
7. The distributed analytics system as recited in claim 1 , wherein said prescriptive analytics are based on a threshold performance of a component of said at least one monitored system associated with said sensor.
8. The distributed analytics system as recited in claim 1 , wherein said architect subsystem is remote from said edge subsystem.
9. The distributed analytics system recited in claim 1 , wherein said architect subsystem aggregates the information received from a plurality of edge processing devices to modify the analytic models and redeploy to one or more of said edge processing devices.
10. A method in a distributed analytics system to control the operation of a plurality of monitored systems, said distributed analytics system comprising an architect subsystem and an edge subsystem, said edge subsystem comprising a plurality of edge processing devices located proximate to said plurality of monitored systems, said method comprising the steps of:
deploying, from said architect subsystem, analytic models to said edge processing devices based on sensor characteristics of said monitored systems; and,
receiving, at each of said edge processing devices, at least one of said analytic models, each of said edge processing devices operative to:
1) independently perform predictive and prescriptive analytics on sensor data from at least one monitored system, wherein said prescriptive analytics includes a future action on a component of said monitored system associated with said sensor;
2) provide control signals to said at least one monitored system according to said predictive and prescriptive analytics; and,
3) periodically send information to said architect subsystem, including monitored system responses to said control signals, whereby said architect subsystem can modify said analytic models to improve system performance of said monitored systems.
11. The method as recited in claim 10 , wherein performing said predictive and prescriptive analytics is on at least a component of said at least one monitored system based on said sensor data.
12. The method as recited in claim 10 , wherein said information includes a current status of a component of said at least one monitored system associated with said sensor.
13. The method as recited in claim 10 , wherein said information includes a characteristic of said at least one monitored system based on said sensor data from a plurality of sensors.
14. The method as recited in claim 10 , wherein said predictive analytics includes an estimate of an operation of a component of said at least one monitored system associated with said sensor.
15. The method as recited in claim 14 , wherein said estimate is the remaining operational life of said component of said at least one monitored system.
16. The method as recited in claim 10 , wherein said prescriptive analytics are based on a threshold performance of a component of said at least one monitored system associated with said sensor.
17. The method as recited in claim 10 , wherein said architect subsystem is remote from said edge subsystem.
18. The method recited in claim 10 , further comprising the steps of said architect subsystem aggregating the information received from a plurality of edge processing devices to modify the analytic models and redeploy to one or more of said edge processing devices.