Systems, and methods for diagnosing an additive manufacturing device using a physics assisted machine learning model
A system for diagnosing an additive manufacturing device is provided. The system includes a first module configured to: obtain one or more parameters for a digital twin of a component of the additive manufacturing device based on raw data from the component of the additive manufacturing device; and generate physics features for the digital twin of the component of the additive manufacturing device based on the one or more parameters and one or more transfer functions, a second module configured to obtain one or more classifiers for classifying the component as a first condition or a second condition based on physics features; and a third module configured to: determine a health of the component based on the generated physics features of the first model and the one or more classifiers.
1 . An apparatus to diagnose an additive manufacturing device, the apparatus comprising:
memory; and
a processor, the processor to execute instructions stored in the memory to at least:
select at least one classifier based on physics features to be monitored for the additive manufacturing device, the at least one classifier formed from a model combined with at least one physics feature extracted from a first output of the additive manufacturing device;
determine a respective threshold for each physics feature, the threshold associated with the respective at least one classifier;
evaluate, using the at least one classifier and the respective threshold for each physics feature, data from operation of the additive manufacturing device to determine whether any respective threshold is exceeded;
classify operation of the additive manufacturing device based on the determination of whether any respective threshold is exceeded;
generate a second output based on the classification; and
adjust the additive manufacturing device based on the second output.
2 . The apparatus of claim 1 , wherein the classification includes: a) normal operation of the additive manufacturing device, and b) at least one of an error or a failure of the additive manufacturing device.
3 . The apparatus of claim 2 , wherein the second output includes a root cause of the at least one of the error or the failure.
4 . The apparatus of claim 3 , wherein the root cause is determined by applying at least one of a trained machine learning model or a statistical model.
5 . The apparatus of claim 3 , wherein the second output includes a correction for the root cause.
6 . The apparatus of claim 5 , wherein the correction includes an adjustment of the additive manufacturing device.
7 . The apparatus of claim 1 , wherein the physics features are extracted from at least one physics-based model of the additive manufacturing device, and wherein the at least one classifier is included in a data science model.
8 . The apparatus of claim 1 , wherein the at least one classifier is updated based on an efficiency drop associated with wear and tear of the additive manufacturing device.
9 . The apparatus of claim 1 , wherein the classification is further based on a weighted average and a transfer function.
10 . At least one non-transitory machine readable medium comprising instructions to cause at least one processor to at least:
select at least one classifier based on physics features to be monitored for an additive manufacturing device, the at least one classifier formed from a model combined with at least one physics feature extracted from a first output of the additive manufacturing device;
determine a respective threshold for each physics feature, the threshold associated with the respective at least one classifier;
evaluate, using the at least one classifier and the respective threshold for each physics feature, data from operation of the additive manufacturing device to determine whether any respective threshold is exceeded;
classify operation of the additive manufacturing device based on the determination of whether any respective threshold is exceeded;
generate a second output based on the classification, the second output to adjust at least one of the additive manufacturing device or the at least one classifier; and
adjust the additive manufacturing device based on the second output.
11 . The at least one non-transitory machine readable medium of claim 10 , wherein the classification includes: a) normal operation of the additive manufacturing device, and b) at least one of an error or a failure of the additive manufacturing device.
12 . The at least one non-transitory machine readable medium of claim 11 , wherein the second output includes a root cause of the at least one of the error or the failure.
13 . The at least one non-transitory machine readable medium of claim 12 , wherein the root cause is determined by applying at least one of a trained machine learning model or a statistical model.
14 . The at least one non-transitory machine readable medium of claim 12 , wherein the second output includes a correction for the root cause.
15 . The at least one non-transitory machine readable medium of claim 14 , wherein the correction includes an adjustment of the additive manufacturing device.
16 . The at least one non-transitory machine readable medium of claim 10 , wherein the physics features are extracted from at least one physics-based model of the additive manufacturing device, and wherein the at least one classifier is included in a data science model.
17 . The at least one non-transitory machine readable medium of claim 10 , wherein the at least one classifier is updated based on an efficiency drop associated with wear and tear of the additive manufacturing device.
18 . The at least one non-transitory machine readable medium of claim 10 , wherein the classification is further based on a weighted average and a transfer function.
19 . A method comprising:
selecting, by executing an instruction using processor circuitry, at least one classifier based on physics features to be monitored for an additive manufacturing device, the at least one classifier formed from a model combined with at least one physics feature extracted from a first output of the additive manufacturing device;
determining, by executing an instruction using processor circuitry, a respective threshold for each physics feature, the threshold associated with the respective at least one classifier;
evaluating, using the at least one classifier and the respective threshold for each physics feature, data from operation of the additive manufacturing device to determine whether any respective threshold is exceeded;
classifying, by executing an instruction using processor circuitry, operation of the additive manufacturing device based on the determination of whether any respective threshold is exceeded;
generating, by executing an instruction using processor circuitry, a second output based on the classification, the second output to adjust at least one of the additive manufacturing device or the at least one classifier; and
adjusting the additive manufacturing device based on the second output.
20 . The method of claim 19 , further including correcting a root cause of an error using the second output.