Synthesizing high-fidelity time-series sensor signals to facilitate machine-learning innovations

1. A method for facilitating development of machine-learning techniques to perform prognostic-surveillance operations on time-series data from a monitored system, comprising:

receiving original time-series signals comprising sequences of observations obtained from sensors in the monitored system;

decomposing the original time-series signals into deterministic and stochastic components;

using the deterministic and stochastic components to produce synthetic time-series signals, which are statistically indistinguishable from the original time-series signals; and

enabling a developer to use the synthetic time-series signals to develop machine-learning (ML) techniques to perform prognostic-surveillance operations on subsequently received time-series signals from the monitored system.

2. The method of claim 1 , wherein using the synthetic time-series signals to develop the ML techniques to perform the prognostic-surveillance operations includes determining one or more of the following for the prognostic-surveillance operations:

false-alarm probabilities (FAPs);

missed-alarm probabilities (MAPs);

sensitivity for anomaly discovery;

time-to-detection (TTD) metrics for discovering anomalies; and

overall compute cost for the ML techniques.

3. The method of claim 1 , wherein the synthetic time-series signals have the same serial-correlation structure, cross-correlation structure, and stochastic content as the original time-series signals.

4. The method of claim 1 , wherein the stochastic content includes one or more of the following:

a mean;

a variance;

a skewness;

a kurtosis; and

Kolmogorov-Smirnov test statistics.

5. The method of claim 1 , wherein if the original time-series signals include spikes, producing the synthetic time-series signals involves:

determining a temporal distribution, an amplitude distribution and a width distribution for the spikes in the original time-series signals; and

including synthetic spikes in the synthetic time-series signals, which have the same temporal distribution, amplitude distribution and width distribution as the spikes in the original time-series signals.

6. The method of claim 1 , wherein if there exists a suspect pair of records, which have possibly reversed timestamps, in the original time-series signals, the method further comprises:

training an MSET model on the original time-series signals without the suspect pair of records;

applying the trained MSET model to the original time-series signals to produce residuals r 1 ;

applying the trained MSET model to the original time-series signals with the suspect pair of records reversed to produce residuals r 2 ; and

if r 1 >r 2 , reversing the suspect pair of records in the original time-series signals.

7. The method of claim 1 , wherein producing the synthesized time-series signals involves using a telemetry parameter synthesis (TPSS) technique, wherein the TPSS technique creates a high-fidelity synthesis equation, and then uses the high-fidelity synthesis equation to produce the synthetic time-series signals.

8. The method of claim 7 , wherein using the TPSS technique to produce the synthetic time-series signals includes:

using an autocorrelation technique to determine a longest period for each signal in the original time-series signals;

selecting a portion of the original time-series signals that contains an integer number of periods;

determining a number of Fourier modes N mode , which equals a number of peaks in a spectral-density function for the original time-series signals;

selecting the maximum N mode Fourier modes; and

using the selected N mode Fourier modes to produce the synthetic time-series signals.

9. The method of claim 1 , wherein the monitored system includes one or more of the following:

a power plant;

a power-distribution system;

an enterprise computing system;

a motorized vehicle;

a factory; and

an oil refinery.

10. A non-transitory computer-readable storage medium storing instructions that when executed by a computer cause the computer to perform a method for facilitating development of machine-learning techniques to perform prognostic-surveillance operations on time-series data from a monitored system, the method comprising:

receiving original time-series signals comprising sequences of observations obtained from sensors in the monitored system;

decomposing the original time-series signals into deterministic and stochastic components;

using the deterministic and stochastic components to produce synthetic time-series signals, which are statistically indistinguishable from the original time-series signals; and

enabling a developer to use the synthetic time-series signals to develop machine-learning (ML) techniques to perform prognostic-surveillance operations on subsequently received time-series signals from the monitored system.

11. The non-transitory computer-readable storage medium of claim 10 , wherein using the synthetic time-series signals to develop the ML techniques to perform the prognostic-surveillance operations includes determining one or more of the following for the prognostic-surveillance operations:

false-alarm probabilities (FAPs);

missed-alarm probabilities (MAPs);

sensitivity for anomaly discovery;

time-to-detection (TTD) metrics for discovering anomalies; and

overall compute cost for the ML techniques.

12. The non-transitory computer-readable storage medium of claim 10 , wherein the synthetic time-series signals have the same serial-correlation structure, cross-correlation structure, and stochastic content as the original time-series signals.

13. The non-transitory computer-readable storage medium of claim 10 , wherein the stochastic content includes one or more of the following:

a mean;

a variance;

a skewness;

a kurtosis; and

Kolmogorov-Smirnov test statistics.

14. The non-transitory computer-readable storage medium of claim 10 , wherein if the original time-series signals include spikes, producing the synthetic time-series signals involves:

determining a temporal distribution, an amplitude distribution and a width distribution for the spikes in the original time-series signals; and

including synthetic spikes in the synthetic time-series signals, which have the same temporal distribution, amplitude distribution and width distribution as the spikes in the original time-series signals.

15. The non-transitory computer-readable storage medium of claim 10 , wherein if there exists a suspect pair of records, which have possibly reversed timestamps, in the original time-series signals, the method further comprises:

training an MSET model on the original time-series signals without the suspect pair of records;

applying the trained MSET model to the original time-series signals to produce residuals r 1 ;

applying the trained MSET model to the original time-series signals with the suspect pair of records reversed to produce residuals r 2 ; and

if r 1 >r 2 , reversing the suspect pair of records in the original time-series signals.

16. The non-transitory computer-readable storage medium of claim 10 , wherein producing the synthesized time-series signals involves using a telemetry parameter synthesis (TPSS) technique, wherein the TPSS technique creates a high-fidelity synthesis equation, and then uses the high-fidelity synthesis equation to produce the synthetic time-series signals.

17. The non-transitory computer-readable storage medium of claim 16 , wherein using the TPSS technique to produce the synthetic time-series signals includes:

using an autocorrelation technique to determine a longest period for each signal in the original time-series signals;

selecting a portion of the original time-series signals that contains an integer number of periods;

determining a number of Fourier modes N mode , which equals a number of peaks in a spectral-density function for the original time-series signals;

selecting the maximum N mode Fourier modes; and

using the selected N mode Fourier modes to produce the synthetic time-series signals.

18. The non-transitory computer-readable storage medium of claim 10 , wherein the monitored system includes one or more of the following:

a power plant;

a power-distribution system;

an enterprise computing system;

a motorized vehicle;

a factory; and

an oil refinery.

19. A system that facilitates development of machine-learning techniques to perform prognostic-surveillance operations on time-series data from a monitored system, comprising:

at least one processor and at least one associated memory; and

a development mechanism that executes on the at least one processor, wherein during operation, the development mechanism:

receives original time-series signals comprising sequences of observations obtained from sensors in the monitored system;

decomposes the original time-series signals into deterministic and stochastic components;

uses the deterministic and stochastic components to produce synthetic time-series signals, which are statistically indistinguishable from the original time-series signals; and

enables a developer to use the synthetic time-series signals to develop machine-learning (ML) techniques to perform prognostic-surveillance operations on subsequently received time-series signals from the monitored system.

20. The system of claim 19 , wherein while using the synthetic time-series signals to develop the ML techniques to perform the prognostic-surveillance operations, the system determines one or more of the following for the prognostic-surveillance operations:

false-alarm probabilities (FAPs);

missed-alarm probabilities (MAPs);

sensitivity for anomaly discovery;

time-to-detection (TTD) metrics for discovering anomalies; and

overall compute cost for the ML techniques.