Self-service terminal (SST) maintenance and support processing
Data is collected for Self-Service Terminals (SSTs) including tallies, events, and outcomes associated with servicing the SSTs. Statistical correlations are derived from the tallies and events with respect to the outcomes. Subsequent collected data is processed with the statistical correlations and a probability for a failure of a component or a part of the component associated with a particular SST is reported for servicing the component or part before the failure.
1. A method, comprising:
obtaining, by executable instructions that execute on a hardware processor of a server from a non-transitory computer-readable storage medium, tallies, events, and outcomes associated with service tickets and service reports for a plurality of Self-Service Terminals (SSTs), wherein obtaining further includes obtaining signal data from the SSTs and extracting signal features comprising spectral slope and audio spectrum flatness, wherein obtaining further includes obtaining electromechanical vibrational data associated with rotational and mechanical components of the SSTs and extracting vibration features for the electromechanical vibrational data, wherein obtaining further includes identifying from the service tickets and service reports the signal data and electromechanical vibration data from motors, tracks, sensors, pumps, compressors, microphones, valves, cameras, turbines, and HVAC machines of the SSTs;
finding, by executable instructions, statistical correlations between the tallies, events, signal features, and vibration features with respect to the outcomes;
applying, by executable instructions, the statistical correlations to a new set of tallies, a new set of signal data and electromechanical vibration data having new signal features, new vibration features, and a new set of events for the SSTs by processing a hybrid statistical analysis that is combined with processing a machine learning algorithm and obtaining a probable issue with respect to at least a part of a particular SST, wherein applying further includes continuously training the machine learning algorithm using hierarchical classification trees that classifies parts of a given mechanical component with the given mechanical component, classifies groupings of the mechanical components by component functions, and classifies groupings of the mechanical components with related mechanical components;
reporting, by executable instructions, a probable issue; and
processing hybrid statistical analysis and the machine learning algorithm to provide probable failures in the SSTs and predicted time periods that the probable failures are to occur with continuous refinement based on observed actions taken with respect to the SSTs and the observed actions provided back to the hybrid statistical analysis and the machine learning algorithm for the refinement.
2. The method of claim 1 , wherein obtaining further includes gathering the tallies, events, and outcomes from a plurality of data stores.
3. The method of claim 2 , wherein gathering further includes organizing the tallies, events, and outcomes in a normalized data format within a single data store.
4. The method of claim 3 , wherein organizing further includes assigning hierarchical classifications to collections and combinations of the tallies, events, and outcomes using the hierarchical classification trees within the single data store.
5. The method of claim 4 , wherein finding further includes weighting the hierarchical classifications based on the statistical correlations to the outcomes.
6. The method of claim 1 , wherein finding further includes identifying patterns for collections of the tallies and events with respect to the outcomes.
7. The method of claim 6 , wherein identifying further includes deriving the statistical correlations based on frequencies of the patterns for the collections.
8. The method of claim 7 , wherein deriving further includes producing the statistical correlations based on any given combination for the tallies and events and producing the statistical correlations.
9. The method of claim 1 , wherein applying further includes identifying a set of service actions that resolves a known issue for the particular SST based on the new tallies and the new events and reporting the set of service actions to a service engineer.
10. The method of claim 1 , wherein reporting further includes reporting a future period of time from the current date within which the probable issue is expected to occur.
11. The method of claim 1 further comprising, iterating the method and refining the statistical correlations with ongoing-collected tallies, ongoing-collected events, and ongoing-collected outcomes for the SSTs.
12. A method, comprising:
organizing, by executable instructions that execute on a hardware processor of a server from a non-transitory computer-readable storage medium, tallies, events, signal data, electromechanical vibration data, and outcomes for Self-Service Terminals (SSTs) in a data store, wherein the data store comprising service tickets and service reports for the SSTs, wherein organizing further includes extracting signal features from the signal data a comprising spectral slope and audio spectrum flatness, wherein organizing further includes extracting vibration features from electromechanical rotational data associated with rotational and mechanical components of the SSTs, wherein organizing further includes identifying from the service tickets, the service reports, the signal data, and electromechanical vibration data for motors, tracks, sensors, pumps, compressors, microphones, valves, cameras, turbines, and HVAC machines of the SSTs;
deriving, by executable instructions, a statistical function between the tallies, the signal features, the vibration features, and events with respect to the outcomes by processing a hybrid statistical analysis that is combined with processing a machine learning algorithm and obtaining the statistical function;
processing, by executable instructions, the statistical function against dynamically collected tallies, dynamically collected signal data and dynamically collected electromechanical vibration data with dynamically extracted signal features, and dynamically extracted vibration features, and dynamically collected events for the SSTs for determining a statistically probable unfavorable outcome;
reporting, by executable instructions, the statistically unfavorable probable outcome and a set of preventive actions to take on a specific SST to proactively avoid the statistically probable unfavorable outcome for the specific SST;
continuously training the machine learning algorithm using hierarchical classification trees that classifies parts of a given mechanical component with the given mechanical component, classifies groupings of the mechanical components by component functions, and classifies groupings of the mechanical components with related mechanical components; and
processing hybrid statistical analysis and the machine learning algorithm to provide probable failures in the SSTs and predicted time periods that the probable failures are to occur with continuous refinement based on observed actions taken with respect to the SSTs and the observed actions provided back to the hybrid statistical analysis and the machine learning algorithm for the refinement.
13. The method of claim 12 , wherein organizing further includes assigning the tallies and the events to classifications using the hierarchical classification trees that are associated with the mechanical components and the parts of the mechanical components associated with the SSTs.
14. The method of claim 13 , wherein deriving further includes processing a Support Vector Machine (SVM) with a linear regression algorithm to derive the statistical function for the statistical analysis.
15. The method of claim 12 further comprising, iterating the organizing and deriving for modifying the statistical function based on ongoing-collected tallies, ongoing-collected events, and ongoing-collected outcomes that are dynamically collected for the SSTs.
16. The method of claim 12 further comprising, collecting new tallies and new events associated with a particular SST that has an existing service ticket assigned to that particular SST, applying the statistical function using the new tallies and the new events as input to the statistical function, and reporting a specific set of actions to resolve the service ticket to a service engineer assigned to the service ticket based on applying the statistical function.
17. A system, comprising:
a server comprising a hardware processor and a non-transitory computer-readable storage medium comprising executable instructions representing a trainer and a predictor;
the trainer when executed by the hardware processor causes the hardware processor to perform processing comprising:
identifying from service reports and service tickets signal data and electromechanical vibration data for motors, tracks, sensors, pumps, compressors, microphones, valves, cameras, turbines, and HVAC machines of Self-Service Terminals (SSTs);
statistically correlating tallies, the signal data, the electromechanical vibration data, and events for SSTs to outcomes associated with outcomes for resolved service tickets on the SSTs; and
extracting signal features from the signal data data comprising spectral slope and audio spectrum flatness;
extracting vibration features from electromechanical vibration data associated with rotational and mechanical components of the SST;
deriving the predictor as a function that predicts a future outcome for any given one of the SSTs based on input associated with particular tallies, particular signal data and particular electromechanical vibration data having particular extracted signal features, particular vibration features, and particular events for a particular SST by processing a hybrid statistical analysis that is combined with processing a machine learning algorithm and obtaining the function;
continuously training the machine learning algorithm using hierarchical classification trees that classifies parts of a given mechanical component with the given mechanical component, classifies groupings of the mechanical components by component functions, classifies groupings of the mechanical components with related mechanical components; and
processing hybrid statistical analysis and the machine learning algorithm to provide probable failures in the SSTs and predicted time periods that the probable failures are to occur with continuous refinement based on observed actions taken with respect to the SSTs and the observed actions provided back to the hybrid statistical analysis and the machine learning algorithm for the refinement; and
the predictor when executed by the hardware processor causes the hardware processor to perform processing comprising:
processing the particular tallies, the particular extracted signal features, the particular vibration features, and the particular events and assign a probability to the future outcome;
identifying actions to take on the particular SST to avoid the future outcome; and
reporting an identifier for the particular SST, the probability, the future outcome, and the actions.
18. The system of claim 17 , wherein the predictor when executed by the hardware processor is further configured to cause the hardware processor to perform additional processing comprising:
predicting a time frame from a current date during which the future outcome will occur and report the time frame.