Switchover for redundant control systems

1. A redundant control system comprising:

a plurality of channels each comprising a processing system configured to execute a control application for a controlled system; and

a switchover artificial intelligence control operable to:

evaluate a state of the channels of the redundant control system;

monitor a plurality of input/output data and communication data of the channels, wherein the input/output data comprises one or more sensor inputs and one or more outputs to a plurality of components of the controlled system;

apply a fault model to determine one or more component faults and system faults of the channels based on the state, the input/output data, and the communication data, wherein the fault model defines a plurality of relationships between the component faults of the components and the system faults, and the relationships are defined based on a plurality of training data parameters uploaded from a remote analysis system; and

command a switchover of a control function from one of the channels having a lower health status to one of the channels having a higher health status based on the component faults and system faults of the channels.

2. The redundant control system of claim 1 , wherein the training data parameters are adaptively adjusted as the switchover artificial intelligence control observes a plurality of patterns in one or more of the input/output data and the communication data.

3. The redundant control system of claim 1 , wherein the switchover artificial intelligence control is configured to apply trending that identifies inconsistent behavior not matching a known fault mode or an expected result from a model of the controlled system.

4. The redundant control system of claim 1 , wherein the processing system comprises a multi-core processor, and the control application executes in a different processing core from the switchover artificial intelligence control.

5. The redundant control system of claim 4 , wherein the processing system comprises a memory system with a non-volatile memory, and the control application and the switchover artificial intelligence control reside in different portions of the non-volatile memory.

6. The redundant control system of claim 1 , wherein the controlled system is a gas turbine engine.

7. A redundant control system, comprising:

a plurality of channels each comprising a processing system configured to execute a control application for a controlled system; and

a switchover artificial intelligence control operable to:

evaluate a state of the channels of the redundant control system;

monitor a plurality of input/output data and communication data of the channels;

apply a fault model to determine one or more component faults and system faults of the channels based on the state, the input/output data, and the communication data, wherein the switchover artificial intelligence control is trained using artificial intelligence to adapt as one or more cyber threats are characterized in one or more of the input/output data and the communication data; and

command a switchover of a control function from one of the channels having a lower health status to one of the channels having a higher health status based on the component faults and system faults of the channels.

8. A method comprising:

evaluating, by a switchover artificial intelligence control, a state of a plurality of channels of a redundant control system, wherein the channels comprise a processing system configured to execute a control application for a controlled system;

monitoring, by the switchover artificial intelligence control, a plurality of input/output data and communication data of the channels, wherein the input/output data comprises one or more sensor inputs and one or more outputs to a plurality of components of the controlled system;

applying, by the switchover artificial intelligence control, a fault model to determine one or more component faults and system faults of the channels based on the state, the input/output data, and the communication data, wherein the fault model defines a plurality of relationships between the component faults of the components and the system faults, and the relationships are defined based on a plurality of training data parameters uploaded from a remote analysis system; and

commanding, by the switchover artificial intelligence control, a switchover of a control function from one of the channels having a lower health status to one of the channels having a higher health status based on the component faults and system faults of the channels.

9. The method of claim 8 , wherein the training data parameters are adaptively adjusted as the switchover artificial intelligence control observes a plurality of patterns in one or more of the input/output data and the communication data.

10. The method of claim 8 , wherein the switchover artificial intelligence control is configured to apply trending that identifies inconsistent behavior not matching a known fault mode or an expected result from a model of the controlled system.

11. The method of claim 8 , wherein the processing system comprises a multi-core processor, and the control application executes in a different processing core from the switchover artificial intelligence control.

12. The method of claim 11 , wherein the processing system comprises a memory system with a non-volatile memory, and the control application and the switchover artificial intelligence control reside in different portions of the non-volatile memory.

13. The method of claim 8 , wherein the controlled system is a gas turbine engine.

14. A method comprising:

evaluating, by a switchover artificial intelligence control, a state of a plurality of channels of a redundant control system, wherein the channels comprise a processing system configured to execute a control application for the a controlled system;

monitoring, by the switchover artificial intelligence control, a plurality of input/output data and communication data of the channels;

applying, by the switchover artificial intelligence control, a fault model to determine one or more component faults and system faults of the channels based on the state, the input/output data, and the communication data, wherein the switchover artificial intelligence control is trained using artificial intelligence to adapt as one or more cyber threats are characterized in one or more of the input/output data and the communication data; and

commanding, by the switchover artificial intelligence control, a switchover of a control function from one of the channels having a lower health status to one of the channels having a higher health status based on the component faults and system faults of the channels.