Fault Injection Block about omsimulator HOT 10 CLOSED

Yes that's true, but from a user perspective it may be tedious to insert FIBs as separate FMUs, especially for large composite models. Master functionality for activating FIBs for any combination of signals without modifying the actual composite model structure would be a fully industy relevant feature. As far as I know, among FMI masters this feature would be unique for OMSimulator (but for the moment not super critical).

from omsimulator.

This is how to create an fib:

--oms_faultInjection("PI_Controller.co_sim.I.y", oms_fault_type_bias, 0.2)
oms_faultInjection("PI_Controller.co_sim.I.y", oms_fault_type_gain, 10.0)

from omsimulator.

If I understood it correctly, this functionally could be easily realised using an FMU as fault injection block. It basically would have one input and one output and put in the middle of the connection(s) to investigate. It could support all the different fault types described in the document.

from omsimulator.

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

from omsimulator.

Should we assign a FIB to an output, input, or connection?

Proposal
One option would be to directly assign the fault injection to an output. The original output, e.g. model.root.unit.y, could then be redirected to model.root.unit.y.$original, and model.root.unit.y would be calculated based on its original value and the fault type.

API: oms_faultInjection(<output name>, <fault type>, <fault value>).

from omsimulator.

Yes I think it shoul be on the output, seems to be in line with the reference mentioned above. A typical example is to simulate sensor failure.

from omsimulator.

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

from omsimulator.

I think current proposal above (on output) is good.

from omsimulator.

I will consider the following fault types:

typedef enum {
  oms_fault_type_none,      ///< y := y.$original
  oms_fault_type_bias,      ///< y := y.$original + faultValue
  oms_fault_type_gain,      ///< y := y.$original * faultValue
  oms_fault_type_const      ///< y := faultValue
} oms_fault_type_enu_t;

from omsimulator.

It is now available for testing. Documentation and Python bindings will be added soon.

from omsimulator.