Method and system for intelligent failure diagnosis center for burn-in devices under test
A mechanism is provided for automatically detecting, diagnosing, transporting, and repairing devices having failed during burn-in testing. Embodiments provide a system that monitors devices undergoing burn-in testing and detecting when a device or a component within a device fails the burn-in test. Embodiments can then alert burn-in-rack monitor personnel of the device failure. Embodiments can concurrently determine the nature of the failure applying a machine learning-based prediction model against log files associated with the failed device. The diagnosis along with a recommended repair strategy can be provided to the repair center as an aid in accelerating the repair process. In addition, the diagnosis can be used to order parts for the repair from a parts depot. In this manner, embodiments can reduce the time for detection, diagnosis, and repair of the failed device.
1. A computer-implementable method for remedying burn-in test failure of devices under test, the method comprising:
monitoring status of one or more devices under burn-in testing;
determining that a device under test (DUT) of a first type of the one or more devices under burn-in testing has failed one or more burn-in tests; and
diagnosing one or more causes of the failure of the one or more burn-in tests by the DUT of the first type, wherein said diagnosing is performed using a probabilistic machine learning system trained from a historical set of burn-in test failure data of devices of the first type, the historical set of burn-in test failure data comprising historical failure logs, configuration file records and historical repair data;
transmitting instructions to an automated guided vehicle (AGV) to report to a burn-in rack associated with the failed DUT and to transport the failed DUT to a selected device repair station; and,
transporting the failed DUT to the selected device repair station via the AGV.
2. The method of claim 1 , wherein
the DUT of the first type comprises a first set of components; and
the failure of the one or more burn-in tests by the DUT of the first type is associated with one or more components of the first set of components.
3. The method of claim 1 , wherein the probabilistic machine learning system comprises a Naïve-Bayes classifier.
4. The method of claim 1 further comprising:
instructing the failed DUT to reboot in response to said diagnosing.
5. The method of claim 1 further comprising, in response to said diagnosing, performing one or more of:
alerting a burn-in test center of the failed DUT;
requesting one or more failed component replacements from a material handling system; and
transmitting the diagnosis regarding the failed DUT to a device repair system.
6. The method of claim 5 , wherein said alerting, requesting, and transmitting are performed in parallel.
7. The method of claim 5 , wherein said alerting comprises:
transmitting an identification of the failed DUT to a burn-in test personnel.
8. The method of claim 7 , wherein said alerting further comprises:
transmitting the identification of the failed DUT to a mobile device associated with the burn-in test personnel.
9. The method of claim 5 , wherein said requesting the one or more failed component replacements comprises:
determining recommended repair strategy in response to said diagnosing;
determining recommended replacement components associated with the recommended repair strategy; and
transmitting an identifier of the recommended replacement components to the material handling system.
10. The method of claim 9 further comprising;
transmitting instructions to a second automated guided vehicle (AGV) to transport the replacement components to the selected device repair station; and,
transporting the replacement components to the selected device repair station via the second ACV.
11. The method of claim 5 , wherein said diagnosing comprises:
identifying one or more failed components of the failed DUT using a log file associated with burn-in testing of the failed DUT as an input to the probabilistic machine learning system; and
identifying one or more repair strategies for the failed DUT in response to said identifying the one or more failed components.
12. The method of claim 11 further comprising:
displaying the diagnosis at the selected device repair station, wherein the failed DUT and replacement components are transported to the selected device repair station.
13. The method of claim 1 further comprising:
selecting the device repair station for the failed DUT.
14. A system comprising:
a processor;
a data bus coupled to the processor;
a network interface, coupled to the data bus and a network, and configured to communicate via the network with a burn-in test monitoring system, a material handling system, and a device repair system, wherein
the burn-in test monitoring system is coupled to one or more devices under burn-in testing, and
the devices under burn-in testing each comprise one or more components; and
a non-transitory, computer-readable storage medium embodying computer program code, the non-transitory, computer-readable storage medium being coupled to the data bus, the computer program code interacting with a plurality of computer operations and comprising instructions executable by the processor and configured to:
monitor status of the one or more devices under burn-in testing,
determine that a device under test (DUT) of a first type of the one or more devices under testing has failed one or more burn-in tests,
diagnose one or more causes of the failure of the one or more burn-in tests by the DUT of the first type, wherein said diagnosing is performed using a probabilistic machine learning system trained from a historical set of burn-in test failure data of devices of the first type, the historical set of burn-in test failure data comprising historical failure logs, configuration file records and historical repair data;
transmit instructions to an automated guided vehicle (AGV) to report to a burn-in rack associated with the failed DUT and to transport the failed DUT to a selected device repair station; and,
transport the failed DUT to the selected device repair station via the AGV.
15. The system of claim 14 further comprising:
a machine learning accelerator processor, coupled to the data bus, and configured to execute instructions configured for the probabilistic machine learning system.
16. The system of claim 15 , wherein the probabilistic machine learning system comprises a Naïve-Bayes classifier.
17. The system of claim 14 wherein the computer program code comprises further instructions executable by the processor further configured to:
alert, using the network interface, the burn-in test monitoring system of the failed DUT;
request, using the network interface, one or more failed component replacements from the material handling system; and
transmit, using the network interface, the diagnosis regarding the failed DUT to the device repair system.
18. The system of claim 17 wherein the instructions configured to request one or more failed component replacements comprise further instructions executable by the processor configured to
determine recommended repair strategy in response to said diagnosing;
determine recommended replacement components associated with the recommended repair strategy; and
transmit, using the network interface, an identifier of the recommended replacement components to the material handling system.
19. The system of claim 17 wherein the instructions configured to diagnose the one or more causes of the failure of the one or more burn-in tests by the DUT comprise further instructions executable by the processor configured to
identify one or more failed components of the failed DUT using a log file associated with the burn-in testing of the failed DUT as an input to the probabilistic machine learning system; and
identify one or more repair strategies for the failed DUT in response to said identifying the one or more failed components.