Once the inspection program can output a result, check that the receiver reads the right fields and processes the result for the matching part and cycle. Use normal parts and representative fault samples. Record sending, acknowledgment, processing and state updates.

Check duplicates, timeouts and late delivery

In the agreed test environment, reproduce missing acknowledgments, duplicate delivery, late results and mismatched identifiers. Check whether the receiver detects duplicates, the sender retries as agreed and unusable results are recorded and withheld. Keep message records traceable to the same part.

Check restart and site timing

Check restart on both sides, network interruption and recovery. Verify that old and new results remain distinct, and that state and sequence recover as agreed. Measure result arrival, acknowledgment and action time while using the intended concurrent workloads.

Archive the actual test conditions

Keep the interface version, equipment configuration, part identifiers, test steps, expected behavior and logs from both endpoints. Mark untested faults and cycle targets as pending. Record the scope of acceptance and retest affected items after interface or recipe changes.

Related service: Vision algorithms and model engineering.