Predictive process control for a manufacturing process
Aspects of the disclosed technology encompass the use of a deep-learning controller for monitoring and improving a manufacturing process. In some aspects, a method of the disclosed technology includes steps for: receiving control values associated with a process station in a manufacturing process, predicting an expected value for an article of manufacture output from the process station, and determining if the deep-learning controller can control the manufacturing process based on the expected value. Systems and computer-readable media are also provided.
1. A computer-implemented method for training a deep-learning controller to control a manufacturing process executed in a manufacturing system comprising a plurality of process stations, comprising:
training, by a computing system, a deep learning processor to predict an expected intermediate value or an expected final output for an article of manufacture undergoing the manufacturing process, wherein the training comprises:
learning, by the deep learning processor, to generate the expected intermediate value or the expected final output based on a set of control values output from a process station from the plurality of process stations;
identifying, by the computing system, key influencers in the set of control values that most impact the expected final output;
based on the key influencers, generating, by the computing system, a robust training data set for continual training of the deep learning processor, wherein the robust training data set comprises new control inputs comprising conditions for generating intermediate values that exceed an expected fluctuation but yields final outputs that are in specification; and
re-training, by the computing system, the deep learning processor to predict the expected final output based on the robust training data set.
2. The computer-implemented method of claim 1 , wherein identifying, by the computing system, the key influencers in the set of control values that most impact the expected final output comprises:
ranking aspects of the manufacturing process that impact the expected final output.
3. The computer-implemented method of claim 1 , wherein identifying, by the computing system, the key influencers in the set of control values that most impact the expected final output comprises:
identifying aspects of the manufacturing process that satisfy a threshold percentage of contribution to the expected final output.
4. The computer-implemented method of claim 1 , wherein identifying, by the computing system, the key influencers in the set of control values that most impact the expected final output comprises:
generating a quantitative score for each aspect of the manufacturing process and determining a relative weight of each aspect based on the quantitative score.
5. The computer-implemented method of claim 1 , wherein training, by the computing system, the deep learning processor further comprises:
comparing the expected final output to an actual value;
determining that the expected final output is within a threshold confidence level of the actual value; and
based on the determining, deploying the deep learning processor within the manufacturing process.
6. The computer-implemented method of claim 1 , wherein the robust data set is generated by adjusting one or more station controllers to produce intermediate values that exceed a specified range from a mean of values.
7. The computer-implemented method of claim 1 , further comprising:
after re-training, receiving, by the deep learning processor, a first set of control values from a first process station for a target article of manufacture; and
predicting, by the deep learning processor, a first expected final output for the article of manufacture based on the first set of control values.
8. A system for training a deep-learning controller to control a manufacturing process executed in a manufacturing system comprising a plurality of process stations comprising:
a processor; and
a memory having programming instructions stored thereon, which, when executed by the processor, cause the system to perform operations comprising:
training a deep learning processor to predict an expected intermediate value or an expected final output for an article of manufacture undergoing the manufacturing process, wherein the training comprises:
learning, by the deep learning processor, to generate the expected intermediate value or the expected final output based on a set of control values output from a process station from the plurality of process stations;
identifying key influencers in the set of control values that most impact the expected final output;
based on the key influencers, generating a robust training data set for continual training of the deep learning processor, wherein the robust training data set comprises new control inputs comprising conditions for generating intermediate values that exceed an expected fluctuation but yields final outputs that are in specification; and
re-training the deep learning processor to predict the expected final output based on the robust training data set.
9. The system of claim 8 , wherein identifying the key influencers in the set of control values that most impact the expected final output comprises:
ranking aspects of the manufacturing process that impact the expected final output.
10. The system of claim 8 , wherein identifying the key influencers in the set of control values that most impact the expected final output comprises:
identifying aspects of the manufacturing process that satisfy a threshold percentage of contribution to the expected final output.
11. The system of claim 8 , wherein identifying the key influencers in the set of control values that most impact the expected final output comprises:
generating a quantitative score for each aspect of the manufacturing process and determining a relative weight of each aspect based on the quantitative score.
12. The system of claim 8 , wherein training the deep learning processor further comprises:
comparing the expected final output to an actual value;
determining that the expected final output is within a threshold confidence level of the actual value; and
based on the determining, deploying the deep learning processor within the manufacturing process.
13. The system of claim 8 , wherein the robust data set is generated by adjusting one or more station controllers to produce intermediate values that exceed a specified range from a mean of values.
14. The system of claim 8 , wherein the operations further comprise:
after re-training, receiving, by the deep learning processor, a first set of control values from a first process station for a target article of manufacture; and
predicting, by the deep learning processor, a first expected final output for the article of manufacture based on the first set of control values.
15. A non-transitory computer-readable storage medium for training a deep-learning controller to control a manufacturing process executed in a manufacturing system comprising a plurality of process stations, the non-transitory computer-readable storage medium comprising instructions stored therein, which when executed by one or more processors, cause the one or more processors to perform operations comprising:
training, by a computing system, a deep learning processor to predict an expected intermediate value or an expected final output for an article of manufacture undergoing the manufacturing process, wherein the training comprises:
learning, by the deep learning processor, to generate the expected intermediate value or the expected final output based on a set of control values output from a process station from the plurality of process stations;
identifying, by the computing system, key influencers in the set of control values that most impact the expected final output;
based on the key influencers, generating, by the computing system, a robust training data set for continual training of the deep learning processor, wherein the robust training data set comprises new control inputs comprising conditions for generating intermediate values that exceed an expected fluctuation but yields final outputs that are in specification; and
re-training, by the computing system, the deep learning processor to predict the expected final output based on the robust training data set.
16. The non-transitory computer-readable storage medium of claim 15 , wherein identifying, by the computing system, the key influencers in the set of control values that most impact the expected final output comprises:
ranking aspects of the manufacturing process that impact the expected final output.
17. The non-transitory computer-readable storage medium of claim 15 , wherein identifying, by the computing system, the key influencers in the set of control values that most impact the expected final output comprises:
identifying aspects of the manufacturing process that satisfy a threshold percentage of contribution to the expected final output.
18. The non-transitory computer-readable storage medium of claim 15 , wherein identifying, by the computing system, the key influencers in the set of control values that most impact the expected final output comprises:
generating a quantitative score for each aspect of the manufacturing process and determining a relative weight of each aspect based on the quantitative score.
19. The non-transitory computer-readable storage medium of claim 15 , wherein training, by the computing system, the deep learning processor further comprises:
comparing the expected final output to an actual value;
determining that the expected final output is within a threshold confidence level of the actual value; and
based on the determining, deploying the deep learning processor within the manufacturing process.
20. The non-transitory computer-readable storage medium of claim 15 , wherein the robust data set is generated by adjusting one or more station controllers to produce intermediate values that exceed a specified range from a mean of values.