Electrical signature analysis of electrical rotating machines

• US 10,403,116 B2
• Filed: 06/20/2017
• Issued: 09/03/2019
• Est. Priority Date: 06/20/2017
• Status: Active Grant

• Current Assignee: GE Infrastructure Technology, LLC (GE Vernova, Inc.)
• Original Assignee: General Electric Company
• Inventors: Neti, Prabhakar, Pamulaparthy, Balakrishna, Mishra, Sudhanshu, Vinayagam, Balamourougan, Kanabar, Mitalkumar, Muthukrishnan, Vijayasarathi
• Primary Examiner(s): Yang, James J
• Assistant Examiner(s): Lau, Kevin
• Application Number:
  US15/628,004
• Publication Number:
  US 20180365963A1
• Time in Patent Office: 805 Days
• CPC Class Codes: G01R 23/16 Spectrum analysis; Fourier analysis
• CPC Class Codes: G01R 31/343 in operation
• CPC Class Codes: G08B 21/182 Level alarms, e.g. alarms responsive to variables exceeding a threshold
• CPC Class Codes: H02P 29/0241 the fault being an overvoltage

Abstract

This disclosure relates to systems and methods for electrical signature analysis of electrical rotating machines. In one embodiment of the disclosure, a method includes ascertaining initial information associated with an electrical rotating machine, assigning a plurality of operational conditions associated with the machine to a plurality of buckets, and recording the electrical data to obtain a pre-defined number of sets of learning data. The method further includes determining, based at least on the initial information and the learning data, that the machine is in a healthy condition, obtaining, based on the learning data, baseline data associated with at the at least one bucket, and generating, based on the baseline data, a threshold associated with the bucket and at least one fault frequency associated with the machine. The method further includes generating, based on the threshold and the baseline data, alarms concerning a state of the electrical rotating machine.

Claims

1. A system comprising:

an electrical rotating machine;

a data acquisition device configured to detect and provide at least electrical data associated with the electrical rotating machine; and

an equipment controller communicatively coupled to the data acquisition device, the equipment controller being configured to;

in a learning mode;

ascertain initial information associated with the electrical rotating machine;

assign a plurality of operational conditions associated with the electrical rotating machine to a plurality of buckets, wherein at least one bucket of the plurality of buckets is associated with at least one load window selected from a plurality of load windows;

record the electrical data to obtain a pre-defined number of sets of learning data;

determine, based at least on the initial information and the learning data, that the electrical rotating machine is in a healthy condition; and

based on a determination that the electrical rotating machine is in the healthy condition;

obtain, based on the learning data, baseline data associated with the at least one bucket;

generate, based on the baseline data, a threshold associated with the at least one bucket and at least one fault frequency associated with the electrical rotating machine; and

generate, based on the threshold and the baseline data, at least one alarm concerning a state of the electrical rotating machine, the electrical rotating machine running within operational conditions associated with the at least one bucket, wherein the equipment controller configured to generate the threshold is further configured to;

compute variations for the sets of the learning data associated with the at least one bucket;

select sets with variation less than a pre-determined value;

convert the learning data associated with the selected sets from a time domain to a frequency domain to obtain sets of frequency data;

average the sets of frequency data to obtain the baseline data;

compute a standard deviation of the frequency data for at least one fault frequency; and

determine, based on the standard deviation, the threshold associated with the at least one fault frequency and the at least one bucket, wherein the equipment controller configured to generate the threshold is further configured to compute the threshold by multiplying a standard deviation by a predetermined constant value, the predetermined constant value being selected based on an expected false positive error.

2. The system of claim 1, wherein the equipment controller, while in a monitoring mode, is further configured to:

continue recording the electrical data to obtain incoming data;

determine a current bucket associated with the incoming data;

determine that the baseline data and the threshold are present for the current bucket; and

based on a determination that the baseline data and the threshold are present;

determine that incoming data exceed the baseline data by more than the threshold for the at least one fault frequency; and

based on a determination that the incoming data exceed the baseline data by more than the threshold for the at least one fault frequency, issue an alarm concerning at least one event of the electrical rotating machine.

3. The system of claim 1, wherein the equipment controller configured to determine that the electrical rotating machine is in the healthy condition is further configured to:

obtain, based on the learning data, static data including at least vibration data and temperature signal;

determine that the static data are within a pre-defined range;

based on a determination that the static data are within a pre-defined range, conclude that the electrical rotating machine is in a healthy condition; and

based on a determination that the static data are not within a pre-defined range, issue a recommendation regarding an inspection of the electrical rotating machine.

4. The system of claim 1, wherein the equipment controller configured to determine that the electrical rotating machine is in the healthy condition is further configured to:

compute load conditions for the pre-defined number of sets of learning data;

determine, based on the load conditions, a bucket containing most frequently occurring load conditions;

evaluate, based on the learning data within the bucket, operational data of the electrical rotating machine, the operational data including one of a speed, a torque, or an efficiency;

determine that the operational data are within a pre-defined range; and

based on a determination that the operational data are not within the pre-defined range, issue a recommendation regarding an inspection of the electrical rotating machine.

5. The system of claim 4, wherein the equipment controller configured to determine that the electrical rotating machine is in the healthy condition is further configured to:

compute variations for the sets of the learning data;

select a set with variation less than a pre-determined value;

assign the learning data of the selected set as baseline data;

convert the baseline data and the learning data associated with the bucket from a time domain to a frequency domain;

determine, based on the baseline data, a first frequency signature for the at least one fault frequency;

determine, based on the learning data associated with the bucket, a second frequency signature for the at least one fault frequency; and

determine that the second frequency signature is higher than the first frequency signature by a pre-determined value;

based on a determination that the second frequency signature is lower than the first frequency signature by the pre-determined value, conclude that the electrical rotating machine is in the healthy condition; and

based on a determination that the second frequency signature is higher than the first frequency signature by the pre-determined value, issue a recommendation regarding an inspection of the electrical rotating machine.

6. The system of claim 1, wherein the equipment controller configured to obtain the baseline data is further configured to:

compute variations for the sets of the learning data associated with the at least one bucket;

select a set with the smallest variation; and

assign the selected set as the baseline data.

7. The system of claim 1, wherein after generating the threshold the equipment controller is further configured to:

determine a noise floor based on the baseline data; and

adjust the threshold based on the noise floor.

8. The system of claim 1, wherein the initial information includes at least one of name plate details, maintenance intervals, a maintenance history, a previous fault history, operational environment details, and a description of load conditions.

9. A method for electrical signature analysis, the method comprising:

providing, by a data acquisition device communicatively coupled to an electrical rotating machine, at least electrical data associated with the electrical rotating machine;

ascertaining, by an equipment controller communicatively coupled to the data acquisition device, initial information associated with the electrical rotating machine;

assigning, by an equipment controller, a plurality of operational conditions associated with the electrical rotating machine to a plurality of buckets, wherein at least one bucket of the plurality of buckets is associated with at least one load window selected from a plurality of load windows;

recording, by the equipment controller, the electrical data to obtain a pre-defined number of sets of learning data;

determining, by the equipment controller, based at least on the initial information and the learning data, that the electrical rotating machine is in a healthy condition; and

based on a determination that the electrical rotating machine is in the healthy condition;

obtaining, by the equipment controller and based on the learning data, baseline data associated with at the at least one bucket;

generating, by the equipment controller and based on the baseline data, a threshold associated with the at least one bucket and the at least one fault frequency associated with the electrical rotating machine; and

generating, based on the threshold and the baseline data, an alarm concerning a state of the electrical rotating machine running within operational conditions associated with the at least one bucket, wherein generating the threshold includes;

computing, by the equipment controller, variations for sets of the learning data associated with the at least one bucket;

selecting, by the equipment controller, sets with variation less than a pre-determined value;

converting, by the equipment controller, the learning data associated with the selected sets from a time domain to a frequency domain to obtain sets of frequency data;

averaging, by the equipment controller, the sets of frequency data to obtain the baseline data;

computing, by the equipment controller, a standard deviation of frequency data at the at least one fault frequency; and

determining, by the equipment controller and based on the standard deviation, the threshold associated with the at least one bucket, wherein the threshold is determined by multiplying a standard deviation by a predetermined constant, the predetermined constant being selected based on an expected false positive error.

10. The method of claim 9, further comprising:

recording, by the equipment controller, the electrical data to obtain incoming data;

determining, by the equipment controller, a current bucket associated with the incoming data;

determining, by the equipment controller, that the baseline data and the threshold are present for the current bucket; and

based on a determination that the baseline data and the threshold are present;

determining, by the equipment controller, that incoming data exceed the baseline data by more than the threshold at the at least one fault frequency; and

based on the determination that that the incoming data exceed the baseline data by more than the threshold for at least one fault frequency, issuing, by the equipment controller, an alarm concerning at least one event of the electrical rotating machine.

11. The method of claim 9, wherein determining that the electrical rotating machine is in the healthy condition includes:

obtaining, by the equipment controller and based on the learning data, static data including at least vibration data and temperature signal;

determining, by the equipment controller, that the static data are within a pre-defined range;

based on a determination that the static data are within a pre-defined range, establishing that the electrical rotating machine is in the healthy condition; and

based on the determination that the static data are not within the pre-defined range, issuing, by the equipment controller, a recommendation regarding an inspection of the electrical rotating machine.

12. The method of claim 9, wherein the determining that the electrical rotating machine is in the healthy condition includes:

computing, by the equipment controller, load conditions for the pre-defined number of sets of learning data;

determining, by the equipment controller and based on the load conditions, a bucket containing most frequently occurring load conditions;

evaluating, by the equipment controller and based on the learning data within the bucket, operational data of the electrical rotating machine, the operational data including one of a speed, a torque, or an efficiency;

determining, by the equipment controller, that the operational data are within a pre-defined range; and

based on a determination that the operational data are not within a pre-defined range, issuing, by the equipment controller, a recommendation regarding an inspection of the electrical rotating machine.

13. The method of claim 12, wherein the determining that the electrical rotating machine is in the healthy condition, further comprises:

computing, by the equipment controller, variations for the sets of the learning data;

selecting, by the equipment controller, a set with the variation less than a pre-determined value;

assigning, by the equipment controller, the learning data of the selected set as the baseline data;

converting, by the equipment controller, the baseline data and the learning data associated with the at least one bucket from a time domain to a frequency domain;

determining, by the equipment controller and based on the baseline data, a first frequency signature for the at least one fault frequency;

determining, by the equipment controller and based on the learning data associated with the load window, a second frequency signature for the at least one fault frequency; and

determining, by the equipment controller, that the second frequency signature is higher than the first frequency signature by a pre-determined value;

based on the determination that the second frequency signature is not higher than the first frequency signature by the pre-determined value, establishing that the electrical rotating machine is in the healthy condition; and

based on the determination that the second frequency signature is higher than the first frequency signature by the pre-determined value, issuing, by the equipment controller, a recommendation regarding an inspection of the electrical rotating machine.

14. The method of claim 9, wherein obtaining the baseline data includes:

computing, by the equipment controller, variations for sets of the learning data associated with the at least one bucket;

selecting, by the equipment controller, a set with the smallest variation; and

assigning, by the equipment controller, the selected set as the baseline data.

15. The method of claim 9, further comprising, after generating the threshold:

determining, by the equipment controller, a noise floor based on the baseline data; and

adjusting, by the equipment controller, the threshold based on the noise floor.

16. A system for electrical signature analysis, the system comprising:

an electrical rotating machine;

a data acquisition device configured to detect and provide at least electrical data associated with the electrical rotating machine, wherein the electrical data includes a current data and a voltage data; and

an equipment controller communicatively coupled to the data acquisition device, the equipment controller configured to;

in a learning mode;

ascertain initial information associated with the electrical rotating machine;

assign a plurality of operational conditions associated with the electrical rotating machine to buckets, wherein at least one bucket of the plurality of buckets is associated with at least one load window selected from a plurality of load windows;

record the electrical data to obtain a pre-defined number of sets of learning data;

determine, based at least on the initial information and the learning data, that the electrical rotating machine is in healthy condition; and

based on the determination that the electrical rotating machine is in the healthy condition;

obtain, based on the learning data, baseline data associated with at least one bucket; and

generate, based on the baseline data and a predetermined constant, a threshold associated with the at least one bucket and with the at least one fault frequency associated with the electrical rotating machine, the predetermined constant being selected based on an expected false positive error;

determine a noise floor based on the baseline data; and

adjust the threshold based on the noise floor;

in a monitoring mode;

record the electrical data to obtain incoming data;

determine a current bucket associated with the incoming data;

determine that the baseline data and the threshold are present for the current bucket; and

based on the determination that the baseline data and the threshold are present;

determine that incoming data exceed the baseline data by more than the threshold at the at least one fault frequency; and

based on the determination that the incoming data exceed the baseline data by more than the threshold for the at least one fault frequency, issue an alarm concerning at least one event associated with the electrical rotating machine; and

if the baseline data and the threshold are not present;

record the incoming data to the learning data associated with the current bucket; and

switch to the learning mode.