Methods and systems for predicting erroneous behavior of an energy asset using fourier based clustering technique

• US 10,163,062 B2
• Filed: 06/22/2015
• Issued: 12/25/2018
• Est. Priority Date: 03/13/2015
• Status: Active Grant

• Current Assignee: Carrier Corporation (Carrier Global Corporation)
• Original Assignee: Carrier Corporation (Carrier Global Corporation)
• Inventors: Prabha, Baburaj Kaimalilputhenpura, Banerjee, Joy
• Primary Examiner(s): Holmes, Michael B
• Application Number:
  US14/746,535
• Publication Number:
  US 20160267387A1
• Time in Patent Office: 1,282 Days
• 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: G06F 2218/08 Feature extraction
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: H02J 2203/20 Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
• CPC Class Codes: H02J 3/00 Circuit arrangements for ac mains or ac distribution networks

Abstract

This disclosure relates generally to predicting health of an energy asset, and more particularly to methods and systems for predicting erroneous behavior of an energy asset using fourier based clustering technique. In one embodiment, a method for determining predicting erroneous behavior of an energy asset is disclosed. The method includes creating one or more energy signatures by performing frequency domain analysis on historical energy data and subsequent clustering of the energy signatures. Further, live energy data is filtered to generate filtered outputs wherein each of the filtered outputs is mapped to a respective cluster. The outlier cluster is identified to predict the erroneous behavior of the energy asset.

Claims

1. A method for predicting anomaly associated with at least one energy asset via a system that includes a processor comprising a data receiver, a depacketizer and a band pass filter, the method comprising:

receiving, by the data receiver of the processor, time stamped historical energy data associated with the at least one energy asset;

creating, by the processor, one or more frequency components by performing frequency domain analysis on the time stamped historical energy data, each of the one or more frequency components indicative of time stamped energy values associated with the at least one energy asset;

clustering, by the processor, the one or more frequency components to generate one or more clusters based on similarity of time stamped energy values, each of the one or more clusters associated with at least one energy signature, the at least one energy signature being average of time stamped energy values for a cluster;

receiving, by the data receiver of the processor, time stamped energy data in real time from the at least one energy asset over a network, wherein the time stamped energy data is received in a data packet format comprising encrypted data;

translating, by the depacketizer of the processor, the encrypted data from the packet format to a user readable format to generate depacketized data comprising the time stamped energy in the user reasonable format;

filtering, by the band filter of the processor, the depacketized data to filter the time stamped energy data within a predetermined range, wherein band pass filter is configured to reject the time stamped energy data that is outside the predetermined range;

comparing, by the processor, between the filtered time stamped energy data and the at least one energy signature associated with a cluster; and

identifying, by the processor, the cluster comprising outlier data based on the comparison to predict anomaly associated with the at least one energy asset.

2. The method of claim 1, wherein each of the one or more clusters is associated with at least one operating state of the at least one energy asset.

3. The method of claim 2, wherein the at least one operating state comprise at least one of a:

normal working of the at least one energy asset, the energy asset on verge of malfunctioning, and actual malfunctioning of the at least one energy asset.

4. The method of claim 1, wherein the at least one energy asset comprise one or more power meters, one or more drives, one or more motors, one or more capacitor banks, one or more air compressors, one or more refrigerator units, one or more turbines, one or more generators, one or more energy storage devices, one or more photovoltaic cells, one or more robots, one or more reactors, or any combinations thereof.

5. The method of claim 1, further comprising notifying a user about the predicted anomaly associated with the at least one energy asset.

6. The method of claim 1, wherein comparing, by the processor, between the filtered time stamped energy data and the at least one energy signature comprises:

computing Euclidean distance between the filtered time stamped energy data and the at least one energy signature.

7. The method of claim 6, wherein the time stamped energy data having least Euclidean distance between the filtered time stamped energy data and the at least one energy signature is mapped to the cluster associated with the at least one energy signature.

8. A system for predicting anomaly associated with at least one energy asset, the system comprising:

a processor comprising a data receiver, a depacketizer, and a band pass filter; and

a memory storing instructions that when executed by the processor, causes the processor to perform operations comprising;

receiving, by the data receiver of the processor, time stamped historical energy data associated with the at least one asset;

creating, by the processor, one or more frequency components by performing frequency domain analysis on the time stamped historical energy data, each of the one or more frequency components indicative of time stamped energy values associated with the at least one energy asset;

clustering, by the processor, the one or more frequency components to generate one or more clusters based on similarity of time stamped energy values, each of the one or more clusters associated with at least one energy signature, the at least one energy signature being average of time stamped energy values for a cluster;

receiving, by the data receiver of the processor, time stamped energy data in real time from the at least one energy asset over a network, wherein the time stamped energy data is received in a data packet format comprising encrypted data;

translating, by the depacketizer of the processor, the encrypted data from the packet format to a user readable format to generate depacketized data comprising the time stamped energy in the user reasonable format;

filtering, by the band filter of the processor, the depacketized data to filter the time stamped energy data within a predetermined range, wherein band pass filter is configured to reject the time stamped energy data that is outside the predetermined range;

comparing, by the processor, between the filtered time stamped energy data and the at least one energy signature associated with a cluster; and

identifying, by the processor, the cluster comprising outlier data based on the comparison to predict anomaly associated with the at least one energy asset.

9. The system of claim 8, wherein each of the one or more clusters is associated with at least one operating state of the at least one energy asset.

10. The system of claim 9, wherein the at least one operating state comprise at least one of a:

normal working of the at least one energy asset, the energy asset on verge of malfunctioning, and actual malfunctioning of the at least one energy asset.

11. The system of claim 8, wherein the at least one energy asset comprise one or more power meters, one or more drives, one or more motors, one or more capacitor banks, one or more air compressors, one or more refrigerator units, one or more turbines, one or more generators, one or more energy storage devices, one or more photovoltaic cells, one or more robots, one or more reactors, or any combinations thereof.

12. The system of claim 8, further comprising notifying a user about the predicted anomaly associated with the at least one energy asset.

13. The system of claim 8, wherein comparing, by the processor, between the filtered time stamped energy data and the at least one energy signature comprises:

computing Euclidean distance between the filtered time stamped energy data and the at least one energy signature.

14. The system of claim 13, wherein the filtered time stamped energy data having least Euclidean distance between the time stamped energy data and the at least one energy signature is mapped to the cluster associated with the at least one energy signature.