System and method for unsupervised root cause analysis of machine failures
A system and method for unsupervised root cause analysis of machine failures. The method includes analyzing, via at least unsupervised machine learning, a plurality of sensory inputs that are proximate to a machine failure, wherein the output of the unsupervised machine learning includes at least one anomaly; identifying, based on the output at least one anomaly, at least one pattern; generating, based on the at least one pattern and the proximate sensory inputs, an attribution dataset, the attribution dataset including a plurality of the proximate sensory inputs leading to the machine failure; and generating, based on the attribution dataset, at least one analytic, wherein the at least one analytic includes at least one root cause anomaly representing a root cause of the machine failure.
1. A method for unsupervised root cause analysis of machine failures, comprising:
selecting at least one unsupervised machine learning model based on a plurality of sensory inputs leading to a machine failure, wherein each unsupervised machine learning model is selected for a respective parameter represented by at least one of the plurality of sensory inputs;
analyzing, via at least unsupervised machine learning, the plurality of sensory inputs leading to the machine failure when the machine failure is identified, wherein the output of the unsupervised machine learning includes at least one anomaly, wherein analyzing the plurality of sensory inputs further comprises running the selected at least one unsupervised machine learning model;
identifying, based on the output at least one anomaly, at least one pattern;
generating, based on the at least one pattern and the sensory inputs, an attribution dataset, the attribution dataset including a plurality of the sensory inputs leading to the machine failure; and
generating, based on the attribution dataset, at least one analytic, wherein the at least one analytic includes at least one root cause anomaly representing a root cause of the machine failure.
2. The method of claim 1 , wherein the plurality of sensory inputs leading to the machine failure include sensory inputs received during occurrence of at least one anomalous sequence of the output at least one anomaly.
3. The method of claim 1 , wherein the attribution dataset further includes at least one environmental variable related to operation of the machine.
4. The method of claim 3 , further comprising:
generating, based on the at least one analytic and at least one recommendation rule, a recommendation for avoiding future machine failures.
5. The method of claim 1 , further comprising:
correlating, for each type of sensory input of the monitored sensory inputs, at least one of the at least one anomaly, wherein the at least one pattern is identified further based on the correlation.
6. The method of claim 1 , wherein analyzing the monitored sensory inputs further comprises:
preprocessing the plurality of sensory inputs, wherein the preprocessing includes extracting at least one feature from raw sensory data.
7. The method of claim 1 , further comprising:
generating, based on the at least one anomaly, an anomalies map.
8. The method of claim 1 , further comprising:
generating, based on the running of the selected at least one unsupervised machine learning model, a normal behavior pattern, wherein the at least one anomaly deviates from the normal behavior pattern.
9. The method of claim 1 , wherein the plurality of sensory inputs leading to the machine failure are any of: received within a threshold period of time of the machine failure, and received after a change in behavioral patterns of the sensory inputs and before the machine failure.
10. A non-transitory computer readable medium having stored thereon instructions for causing a processing circuitry to perform a process, the process comprising:
selecting at least one unsupervised machine learning model based on a plurality of sensory inputs leading to a machine failure, wherein each unsupervised machine learning model is selected for a respective parameter represented by at least one of the plurality of sensory inputs;
analyzing, via at least unsupervised machine learning, the plurality of sensory inputs leading to the machine failure when the machine failure is identified, wherein the output of the unsupervised machine learning includes at least one anomaly, wherein analyzing the plurality of sensory inputs further comprises running the selected at least one unsupervised machine learning model;
identifying, based on the output at least one anomaly, at least one pattern;
generating, based on the at least one pattern and the sensory inputs, an attribution dataset, the attribution dataset including a plurality of the sensory inputs leading to the machine failure; and
generating, based on the attribution dataset, at least one analytic, wherein the at least one analytic includes at least one root cause anomaly representing a root cause of the machine failure.
11. A system for unsupervised prediction of machine failures, comprising:
a processing circuitry; and
a memory, the memory containing instructions that, when executed by the processing circuitry, configure the system to:
select at least one unsupervised machine learning model based on a plurality of sensory inputs leading to a machine failure, wherein each unsupervised machine learning model is selected for a respective parameter represented by at least one of the plurality of sensory inputs;
analyze, via at least unsupervised machine learning, the plurality of sensory inputs leading to the machine failure when the machine failure is identified, wherein the output of the unsupervised machine learning includes at least one anomaly, wherein analyzing the plurality of sensory inputs further comprises running the selected at least one unsupervised machine learning model:
identify, based on the output at least one anomaly, at least one pattern;
generate, based on the at least one pattern and the sensory inputs, an attribution dataset, the attribution dataset including a plurality of the proximate sensory inputs leading to the machine failure; and
generate, based on the attribution dataset, at least one analytic, wherein the at least one analytic includes at least one root cause anomaly representing a root cause of the machine failure.
12. The system of claim 11 , wherein the plurality of sensory inputs leading to the machine failure include sensory inputs received during occurrence of at least one anomalous sequence of the output at least one anomaly.
13. The system of claim 11 , wherein the attribution dataset further includes at least one environmental variable related to operation of the machine.
14. The system of claim 13 , wherein the system is further configured to:
generate, based on the at least one analytic and at least one recommendation rule, a recommendation for avoiding future machine failures.
15. The system of claim 11 , wherein the system is further configured to:
correlate, for each type of sensory input of the monitored sensory inputs, at least one of the at least one anomaly, wherein the at least one pattern is identified further based on the correlation.
16. The system of claim 11 , wherein the system is further configured to:
preprocess the plurality of sensory inputs, wherein the preprocessing includes extracting at least one feature from raw sensory data.
17. The system of claim 11 , wherein the system is further configured to:
generate, based on the at least one anomaly, an anomalies map.
18. The system of claim 11 , wherein the system is further configured to:
generate, based on the running of the selected at least one unsupervised machine learning model, a normal behavior pattern, wherein the at least one anomaly deviates from the normal behavior pattern.