Method and apparatus for analyzing time series data

• US 8,346,505 B2
• Filed: 07/16/2010
• Issued: 01/01/2013
• Est. Priority Date: 07/16/2009
• Status: Active Grant

• Current Assignee: European Space Agency
• Original Assignee: European Space Agency
• Inventors: Martinez-Heras, José Antonio, Donati, Alessandro, Yeung, Kar Lam
• Primary Examiner(s): Huynh, Phuong
• Application Number:
  US12/837,882
• Publication Number:
  US 20110035188A1
• Time in Patent Office: 900 Days
• Field of Search: 702121-123, 702176-188, 700 28- 30
• US Class Current: 702/176
• CPC Class Codes: G05B 23/021 adopting a different treatment of each operating region or a different mode of the monitored system, e.g. transient modes; different operating configurations of monitored system
• CPC Class Codes: G05B 23/0221 Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods
• CPC Class Codes: G05B 23/0235 based on a comparison with predetermined threshold or range, e.g. "classical methods", carried out during normal operation; threshold adaptation or choice; when or how to compare with the threshold
• CPC Class Codes: G06F 18/211 Selection of the most significant subset of features
• CPC Class Codes: G06F 2218/12 Classification; Matching

Abstract

The present invention relates to a method and an apparatus for determining which one or more time series parameters of a plurality of time series parameters relating to operation of a system are correlated with a first operation state of the system. According to the invention, the method comprises providing time series data including data relating to a time series of each of the plurality of time series parameters; determining at least two first time periods, wherein the system is in the first operation state during the at least two first time periods; determining at least one second time period, wherein the system is in a second operation state during the at least one second time period; determining, for each respective time series parameter of the plurality of time series parameters, a first characteristic parameter relating to a first characteristic of the time series of the respective time series parameter for each of the at least two first time periods and the at least one second time period; and determining which one or more time series parameters of the plurality of time series parameters relating to the operation of the system are correlated with the first operation state of the system by determining, for each respective time series parameter of the plurality of time series parameters, whether or not the respective time series parameter is correlated with the first operation state of the system based on the first characteristic parameters of the respective time series parameter determined for each of the at least two first time periods and the at least one second time period.

Claims

1. Method for determining which one or more time series parameters of a plurality of time series parameters relating to operation and/or behaviour of a system over time are correlated with a first state and/or first condition of the system, comprising:

providing time series data including data relating to a time series of each of the plurality of time series parameters;

determining at least two first time periods, wherein the system is in the first state and/or first condition during the at least two first time periods;

determining at least one second time period, wherein the system is in a second state and/or second condition during the at least one second time period;

determining, for each respective time series parameter of the plurality of time series parameters, a first characteristic parameter relating to a first characteristic of the time series of the respective time series parameter for each of the at least two first time periods and the at least one second time period; and

determining, using a computer processor, which one or more time series parameters of the plurality of time series parameters relating to the operation and/or behaviour of the system over time are correlated with the first state and/or first condition of the system by determining, for each respective time series parameter of the plurality of time series parameters, whether or not the respective time series parameter is correlated with the first state and/or first condition of the system based on the first characteristic parameters of the respective time series parameter determined for each of the at least two first time periods and the at least one second time period.

2. Method according to claim 1, wherein the respective time series parameter is determined to be correlated with the first state and/or first condition of the system, if the first characteristic parameters of the respective time series parameter determined for the at least two first time periods are similar to each other but not similar to the first characteristic parameter of the respective time series parameter determined for the at least one second time period.

3. Method according to claim 1, wherein the method further comprises determining, for each respective time series parameter of the plurality of time series parameters, at least a second characteristic parameter relating to a second characteristic of the time series of the respective time series parameter for each of the at least two first time periods and the at least one second time period, wherein the respective time series parameter is further determined to be correlated with the first state and/or first condition of the system, if the at least second characteristic parameters of the respective time series parameter determined for the at least two first time periods are similar to each other but not similar to the second characteristic parameter of the respective time series parameter determined for the at least one second time period.

4. Method according to claim 1, wherein the similarity of characteristic parameters of the respective time series parameter determined for the at least two first time periods and the at least one second time period is determined based on a variation coefficient, wherein determined characteristic parameters of the respective time series parameter are determined to be similar if the variation coefficient is smaller or equal to a predetermined variation coefficient threshold value.

5. Method according to claim 1, wherein the first characteristic parameter is one of average value of the time series in the respective time period, standard deviation of the time series in the respective time period, variance of the time series in the respective time period, maximum value of the time series in the respective time period, minimum value of the time series in the respective time period, time series parameter range in the respective time period, slope of the time series in the respective time period, maximum minus slope of the time series in the respective time period, minimum minus slope of the time series in the respective time period, maximum plus slope of the time series in the respective time period, minimum plus slope of the time series in the respective time period, skewness of the time series in the respective time period and kurtosis of the time series in the respective time period.

6. Method according to claim 1, wherein the method for determining which one or more time series parameters of a plurality of time series parameters relating to operation and/or behaviour of the system are correlated with the first state and/or first condition of the system is used for the purpose of anomaly investigations, wherein the first state and/or first condition of the system relates to an anomalous operation mode and an anomaly occurs during each of the at least two first time periods, and the second state and/or second condition of the system relates to a normal operation mode and the anomaly does not occur during the at least one second time period.

7. Method according to claim 1, wherein the method for determining which one or more time series parameters of a plurality of time series parameters relating to operation and/or behaviour of the system are correlated with the first state and/or first condition of the system is used for the purpose of period characterizations, wherein the first state and/or first condition of the system during the at least two first time periods relates to a first normal operation mode and the second operation state of the system during the at least one second time period relates to a second normal operation mode different from the first normal operation mode.

8. Method according to claim 1, wherein determining at least two first time periods comprises setting a first of the at least two first target time periods and determining at least a second of the at least two first target time periods based on a shape of the time series for at least one of the plurality of time series parameters in the first of the at least two first time periods, wherein the second of the at least two first time periods is determined such that a shape of the time series for the at least one of the plurality of time series parameters in the second of the at least two first time periods is similar to the shape of the time series for the at least one of the plurality of time series parameters in the first of the at least two first target time periods.

9. Method according to claim 8, wherein the similarity of the shapes of the time series for the at least one of the plurality of time series parameters in the first and second of the at least two first time periods is determined based on one or more dynamic time warping distances.

10. Method according to claim 9, wherein the similarity of shapes of the time series for the at least one of the plurality of time series parameters in the first and second of the at least two first time periods is determined based on a ranking of the lowest dynamic time warping distances of the shapes of the time series for the at least one of the plurality of time series parameters in the first and second of the at least two first time periods.

11. Method according to claim 1, wherein the method for determining which one or more time series parameters of a plurality of time series parameters relating to operation and/or behaviour of the system are correlated with the first state and/or first condition of the system is used for the purpose of anomaly investigations, wherein the first state and/or first condition of the system relates to an anomalous operation mode and an anomaly occurs during one of the two first time periods and the second state and/or second condition of the system relates to a normal operation mode and the anomaly does not occur during the at least one second time period, and wherein the method further comprises detecting the occurrence of an anomaly in real-time time series data relating to a current operation of the system, wherein the occurrence of the anomaly is detected in the real-time time series data, if the shape of the time series for the at least one of the plurality of time series parameters included in the real-time series data is determined to be similar to the shape of the time series for the at least one of the plurality of time series parameters in the one of the at least two first time periods.

12. Method according to claim 1, wherein the system is one or a group of:

a plant, a power plant, a vehicle, a space craft, a space shuttle, a space station, a satellite, a system of a plurality of satellites and/or base stations, a space probe, a system of a plurality of space probes, a plane, an aircraft, or a device relating to robotics, or at least part thereof.

13. Method according to claim 1, wherein the system is one of:

a communication network, a computer network, a traffic network or a power supply network.

14. Method according to claim 1, wherein the system is a financial system such as the stock market.

15. Method according to claim 1, wherein the system is a system of one or more medical devices and the time series data relates to a medical condition of a patient.

16. Method according to claim 1, wherein the time series data is comprised in telemetry data relating to the operation of a technical system over time.

17. Apparatus for determining which one or more time series parameters of a plurality of time series parameters relating to operation and/or behaviour of a system over time are correlated with a first state and/or first condition of the system according to a method according to claim 1, wherein the apparatus comprises:

a time period determining means comprising;

a first time period determining means for determining at least two first time periods, wherein the system is in the first state and/or first condition during the at least two first time periods; and

a second time period determining means for determining at least one second time period, wherein the system is in a second state and/or second condition during the at least one second time period;

a characteristic parameter determining means for determining, for each respective time series parameter of the plurality of time series parameters, a first characteristic parameter relating to a first characteristic of the time series of the respective time series parameter for each of the at least two first time periods and the at least one second time period; and

a correlation determining means for determining which one or more time series parameters of the plurality of time series parameters relating to the operation and/or behaviour of the system are correlated with the first state and/or first condition of the system by determining, for each respective time series parameter of the plurality of time series parameters, whether or not the respective time series parameter is correlated with the first state and/or first condition of the system based on the first characteristic parameters of the respective time series parameter determined for each of the at least two first time periods and the at least one second time period.

18. Apparatus according to claim 17, wherein the time period determining means further comprises a time period setting means for setting at least one of the two first time periods and the at least one second time period.

19. Computer program product comprising a non-transitory computer-readable medium and a computer software program stored therein, wherein the computer software program is stored in a sequence of states corresponding to a sequence of commands, which are configured to be executed by a data processing means of a data processing apparatus such that an apparatus according to claim 17 is provided.