Incorrect fault timing with ScenarioLoader -a

[Fredrik Heintz]

There seems to be a problem with computing the fault time when running
in fast mode (ScenarioLoader -a). I use the Exp_1127_002f_pb_ADAPT-
Lite scenario.

When I use ScenarioLoader I get the correct result:
Scenario: ADAPT-Lite - Exp_1127_002f_pb_ADAPT-Lite.scn
Isolation @ time: 79
diagnosis @1281375442442 <DA> detection = true, isolation = true,
candidates = { { E240 = Offset [ FaultTime = 110078.0, Offset =
-1.01637] } [weight = 1.0] }, notes = 'diagnosis-1';
faultInjection @1281375442363 <SDS> isInjection = true, fault = { E240
= Offset [ FaultTime = 110000.0, Offset = ' -1.0']};
Classification Errors: 2.0

When I use ScenarioLoader -a I get incorrect results:
Scenario: ADAPT-Lite - Exp_1127_002f_pb_ADAPT-Lite.scn
Isolation before fault injection.
Isolation before fault injection.
diagnosis @1281375728546 <DA> detection = true, isolation = true,
candidates = { { E240 = Offset [ FaultTime = 110078.0, Offset =
-1.01637] } [weight = 1.0] }, notes = 'diagnosis-1';
faultInjection @1281375728590 <SDS> isInjection = true, fault = { E240
= Offset [ FaultTime = 222.0, Offset = ' -1.0']};
In this run the time stamp of the sensor data that caused the fault is
1281375838446, which is much later than both the injected and reported
time.

The expected fault time also varies between runs:
Scenario: ADAPT-Lite - Exp_1127_002f_pb_ADAPT-Lite.scn
Isolation before fault injection.
Isolation before fault injection.
diagnosis @1281375902756 <DA> detection = true, isolation = true,
candidates = { { E240 = Offset [ FaultTime = 110078.0, Offset =
-1.01637] } [weight = 1.0] }, notes = 'diagnosis-1';
faultInjection @1281375902803 <SDS> isInjection = true, fault = { E240
= Offset [ FaultTime = 392.0, Offset = ' -1.0']};
In this run the time stamp of the sensor data that caused the fault
was 1281376012489, again much later than the expected injected fault
time and the reported isolation time.

I used the latest version of the DXC framework (v2.1).

/Fredrik

[Jeremy Mange]

We have experienced this same phenomenon; I am not sure exactly how
the FaultTime is calculated in abbreviated mode, but we are assuming
the actual scoring will take place with the ScenarioLoader being run
in normal mode.

[Scott Poll]

Yes, I believe the fault time calculation is not applicable in fast
time mode (David G. can correct me if I'm wrong). Scoring will take
place in normal mode.