Encountering an error during Fault Injection

I am trying to inject faults into a model but repeadetly I am encountering an error: Fault Inport block 'gain/in_Outport1_fault/Fault Inport' is set to read/override invalid signal on the Outport number 1 of block 'xxxxxxxxxxxxxxxx/EM_Torque_Sensor/in'.

Caused by:

Target signal is a continuous-time signal. Continuous-time signals support faults only if its destinations are discrete.

[RAGHUVEER RAJESH]

I tried almost everything I know in the SImulink. Nothing helps... this is the prodcut issue I guess.. Please help urgently..

[Raghava S N]

Hi, @RAGHUVEER RAJESH. It would be helpful if you could share snapshots/files related to your model.

[RAGHUVEER RAJESH]

https://in.mathworks.com/matlabcentral/answers/2105601-how-create-a-degradation-feature-profile-from-available-sensor-data

Answer 1

[Pat Canny]

Hi @RAGHUVEER RAJESH,

I am not sure if you are the same user who asked this on Reddit recently (though that is my assumption). Hopefully the work-around we proposed there is acceptable.

Just in case, here is the full recommendation from Mahesh from our Development team:

Let us understand your use-case a bit more. In a typical plant and controller type of model (in your case may be sensor and controller), the plant side has continuous-timed dynamics, where as the controller side has discrete-timed semantics.

You are trying to add a fault on a signal (For ex: output of a Block A) which has continuous-time semantics. The output of the Block A is connected to EM_Torque_Sensor block's input 1.

Case 1: EM_Torque_Sensor input 1 is a continuous-timed signal which will be discretized immediately either by passing it through a Analog-to-Digital or a Zero-order hold block and then further propagated for other components (blocks)

Case 2: EM_Torque_Sensor input 1 is a continuous-timed signal which will be passed on as continuous signal and then sent other components (blocks) where it will be discretized.

In Case 1, you can add a fault at the output of the Inport block inside of EM_Torque_Sensor and it should work.

However, in Case 2, you are trying to add a fault that has to participate in continuous timed semantics. At the present, we do not have a valid use-case to support this. If you think we should support this, please file an enhancement request explaining what you are trying to do and we will be able to assist you better. We do have couple of workarounds now.

1. Workaround 1: Instead of adding fault at the input 1 of EM_Torque_Sensor, you can add it at the stage where this continuous signal gets discretized
2. Add a zero-order block between output of Block A and input of EM_Torque_Sensor block and then add a fault at the output of the zero-order hold block

If this response did not answer your question, please file a Technical Support escalation. We will be able to handle it better.

[RAGHUVEER RAJESH]

Hello Pat,

Yes, you are right. I have adopted the 2nd workaround for the moment. But, its a workaround and I need to give extra documentation for it to my supervisor. I have a question: How do I find out all the blocks that are discretsing the signal. In other words, what is the easiest way to figure out where the signal is gettign discretized in a huge model???

[RAGHUVEER RAJESH]

https://in.mathworks.com/matlabcentral/answers/2105601-how-create-a-degradation-feature-profile-from-available-sensor-data

[Pat Canny]

Hi @RAGHUVEER RAJESH,

I apologize for the delayed response.

Are you familiar with the Timing Legend? https://www.mathworks.com/help/simulink/ug/how-to-view-sample-time-information.html

[RAGHUVEER RAJESH]

Hi Pat. Thanks for sharing. However, I am now facing another difficulty and I have posted the link of the new post above.

here is the new post: https://in.mathworks.com/matlabcentral/answers/2105601-how-to-create-a-degradation-feature-profile-from-available-sensor-data

and this: https://in.mathworks.com/matlabcentral/answers/2105611-how-to-generate-synthetic-run-to-failure-data-in-simulink

I will be grateful if you assist me or refer me to the right person.

Answer 2

[Raghava S N]

Hi @RAGHUVEER RAJESH,

Without your model details, it is hard to pinpoint exactly why the issue exists. However as per my understanding, the error is a sample time mismatch between 2 or more blocks in your Simulink model.

The best practices page for the Simulink Fault Analyser says – “If you add a fault to a continuous signal, the signal must enter a block with a discrete sample time.”

https://www.mathworks.com/help/fault-analyzer/ug/fault-modeling-best-practices.html#:~:text=If%20you%20add%20a%20fault%20to%20a%20continuous%20signal%2C%20the%20signal%20must%20enter%20a%20block%20with%20a%20discrete%20sample%20time.%20For%20more%20information%20on%20sample%20time%2C%20see%20What%20Is%20Sample%20Time%3F

Going by your model's error trace, the block called “gain” in your model is continuous (has sample time 0) and its output (to which the fault is being injected) is also connected to a continuous block. As mentioned in the best practices documentation, this is not supported.

There are 2 solutions to this:

1. Change the sample time of the block whose input is the output of the gain block so that it is discrete.
2. Change the sample time of the gain block so that it is discrete.

Please refer to this documentation link for more information on sample time - https://www.mathworks.com/help/simulink/ug/what-is-sample-time.html

[Yeldo]

@Pat Canny, @Raghava S N Hello, I have another query regarding the inherited value of sample time used in the fault subsystem block. I am trying to inject a stuck at pervious value fault using simulink fault analyzer. I was getting an error due to the difference in the sample time used in my subsystem model. The simullink Fault Analyzer subsystem (fault subsystem) only support an inherited sample times, but when i make it inherited the fault block is not working as expeceted. But when I add it seperatly as a simulink subsystem its working fine.

NB: I am doing a research on fault injection techniques on model based systems.

[Pat Canny]

Hi @Yeldo,

How are you modeling the fault behavior?

Thanks.

[Yeldo]

Hello @Pat Canny, trying to model a stuck at fault, I am attaching an image of the fault subsystem block to model the fault. the sample time of pulse gnerator and the step function is not inherited from the model. when I model this system in simulink fault subsytem, the correct result are not being driven, but when used alone without fault analyzer the results are correct.

[Pat Canny]

Hi @Yeldo,

How are you modeling the fault behavior in the fault subsystem specifically? That screenshot appears to be using standard Simulink switch blocks. I'm just curious what the specific behavior is (I assume it is the latching logic on the right).

Are you getting any errors or warnings, or just seeing unexpected behavior?

Note: this may best be answered via MathWorks Technical Support.