Synthesizing high-fidelity signals with spikes for prognostic-surveillance applications

1. A method for performing prognostic-surveillance operations on a monitored system, comprising:

receiving original time-series signals from sensors in the monitored system;

detecting and removing spikes from the original time-series signals to produce despiked original time-series signals, wherein detecting the spikes involves,

using the original time-series data to optimize a damping factor, which is applied to a threshold for a spike-detection technique, and

using the spike-detection technique with the optimized damping factor to detect the spikes;

generating despiked synthetic time-series signals, which are statistically indistinguishable from the despiked original time-series signals;

including synthetic spikes, which have the same temporal distribution, amplitude distribution and width distribution as the spikes in the original time-series signals, in the despiked synthetic time-series signals to produce synthetic time-series signals with spikes;

using the synthetic time-series signals with spikes to train an inferential model; and

using the inferential model to perform prognostic-surveillance operations on subsequently-received time-series signals from the monitored system.

2. The method of claim 1 , wherein optimizing the damping factor comprises:

using the spike-detection technique with an initial damping factor to detect and remove the spikes from the original time-series signals to produce despiked original time-series signals;

generating despiked synthetic time-series signals, which are statistically indistinguishable from the despiked original time-series signals;

including synthetic spikes in the despiked synthetic time-series signals to produce ground truth synthetic time-series signals, wherein the synthetic spikes have expanded and contracted temporal distributions, amplitude distributions and width distributions with respect to the spikes in the original time-series signals;

performing tests, which involve varying the damping factor while using the spike-detection technique to detect spikes in the ground truth synthetic time-series signals; and

determining an optimized damping factor based on true and false detections resulting from the tests.

3. The method of claim 1 , wherein the spike-detection technique comprises one of the following:

a phase-space thresholding technique, which uses a phase-space-related threshold;

an acceleration-thresholding technique, which uses an acceleration threshold; and

a wavelet-thresholding technique, which uses a wavelet-space-related threshold.

4. The method of claim 1 , wherein generating the despiked synthetic time-series signals includes:

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

using the deterministic and stochastic components to produce the despiked synthetic time-series signals.

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

6. The method of claim 5 , wherein the stochastic content includes one or more of the following:

a mean;

a variance;

a skewness;

a kurtosis; and

Kolmogorov-Smirnov test statistics.

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

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

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

selecting a portion of the despiked 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 despiked original time-series signals;

selecting the maximum N mode Fourier modes; and

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

9. The method of claim 1 , wherein using the inferential model to perform prognostic-surveillance operations on the subsequently-received time-series signals comprises:

using the prognostic inferential model to generate estimated values for the subsequently-received time-series sensor data;

performing a pairwise differencing operation between actual values and the estimated values for the subsequently-received time-series sensor data to produce residuals; and

performing a sequential probability ratio test (SPRT) on the residuals to detect the incipient anomalies.

10. The method of claim 9 , wherein detecting the incipient anomalies comprises detecting one or more of the following:

an impending failure of the monitored system; and

a malicious-intrusion event in the monitored system.

11. A non-transitory computer-readable storage medium storing instructions that when executed by a computer cause the computer to perform a method for performing prognostic-surveillance operations on a monitored system, the method comprising:

receiving original time-series signals from sensors in the monitored system;

detecting and removing spikes from the original time-series signals to produce despiked original time-series signals, wherein detecting the spikes involves,

using the original time-series data to optimize a damping factor, which is applied to a threshold for a spike-detection technique, and

using the spike-detection technique with the optimized damping factor to detect the spikes;

generating despiked synthetic time-series signals, which are statistically indistinguishable from the despiked original time-series signals;

including synthetic spikes, which have the same temporal distribution, amplitude distribution and width distribution as the spikes in the original time-series signals, in the despiked synthetic time-series signals to produce synthetic time-series signals with spikes;

using the synthetic time-series signals with spikes to train an inferential model; and

using the inferential model to perform prognostic-surveillance operations on subsequently-received time-series signals from the monitored system.

12. The non-transitory computer-readable storage medium of claim 11 , wherein optimizing the damping factor comprises:

using the spike-detection technique with an initial damping factor to detect and remove the spikes from the original time-series signals to produce despiked original time-series signals;

generating despiked synthetic time-series signals, which are statistically indistinguishable from the despiked original time-series signals;

including synthetic spikes in the despiked synthetic time-series signals to produce ground truth synthetic time-series signals, wherein the synthetic spikes have expanded and contracted temporal distributions, amplitude distributions and width distributions with respect to the spikes in the original time-series signals;

performing tests, which involve varying the damping factor while using the spike-detection technique to detect spikes in the ground truth synthetic time-series signals; and

determining an optimized damping factor based on true and false detections resulting from the tests.

13. The non-transitory computer-readable storage medium of claim 11 , wherein the spike-detection technique comprises one of the following:

a phase-space thresholding technique, which uses a phase-space-related threshold;

an acceleration-thresholding technique, which uses an acceleration threshold; and

a wavelet-thresholding technique, which uses a wavelet-space-related threshold.

14. The non-transitory computer-readable storage medium of claim 11 , wherein generating the despiked synthetic time-series signals includes:

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

using the deterministic and stochastic components to produce the despiked synthetic time-series signals.

15. The non-transitory computer-readable storage medium of claim 11 , wherein the despiked synthetic time-series signals have the same serial-correlation structure, cross-correlation structure, and stochastic content as the despiked original time-series signals.

16. The non-transitory computer-readable storage medium of claim 11 , wherein generating the despiked synthetic time-series signals involves using a telemetry parameter synthesis (TPSS) technique, which creates a high-fidelity synthesis equation, and then uses the high-fidelity synthesis equation to produce the despiked synthetic time-series signals.

17. The non-transitory computer-readable storage medium of claim 11 , wherein using the inferential model to perform prognostic-surveillance operations on the subsequently-received time-series signals comprises:

using the prognostic inferential model to generate estimated values for the subsequently-received time-series sensor data;

performing a pairwise differencing operation between actual values and the estimated values for the subsequently-received time-series sensor data to produce residuals; and

performing a sequential probability ratio test (SPRT) on the residuals to detect the incipient anomalies.

18. The non-transitory computer-readable storage medium of claim 17 , wherein detecting the incipient anomalies comprises detecting one or more of the following:

an impending failure of the monitored system; and

a malicious-intrusion event in the monitored system.

19. A system that performs prognostic-surveillance operations on a monitored system, comprising:

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

a prognostic-surveillance mechanism that executes on the at least one processor, wherein during operation, the prognostic-surveillance mechanism:

receives original time-series signals from sensors in the monitored system;

detects and removes spikes from the original time-series signals to produce despiked original time-series signals, wherein while detecting the spikes, the prognostic-surveillance mechanism,

uses the original time-series data to optimize a damping factor, which is applied to a threshold for a spike-detection technique, and

uses the spike-detection technique with the optimized damping factor to detect the spikes;

generates despiked synthetic time-series signals, which are statistically indistinguishable from the despiked original time-series signals;

includes synthetic spikes, which have the same temporal distribution, amplitude distribution and width distribution as the spikes in the original time-series signals, in the despiked synthetic time-series signals to produce synthetic time-series signals with spikes;

uses the synthetic time-series signals with spikes to train an inferential model; and

uses the inferential model to perform prognostic-surveillance operations on subsequently-received time-series signals from the monitored system.

20. The system of claim 19 , wherein while optimizing the damping factor, the prognostic-surveillance mechanism:

uses the spike-detection technique with an initial damping factor to detect and remove the spikes from the original time-series signals to produce despiked original time-series signals;

generates despiked synthetic time-series signals, which are statistically indistinguishable from the despiked original time-series signals;

includes synthetic spikes in the despiked synthetic time-series signals to produce ground truth synthetic time-series signals, wherein the synthetic spikes have expanded and contracted temporal distributions, amplitude distributions and width distributions with respect to the spikes in the original time-series signals;

performs tests, which involve varying the damping factor while using the spike-detection technique to detect spikes in the ground truth synthetic time-series signals; and

determines an optimized damping factor based on true and false detections resulting from the tests.