Detection of pump cavitation/blockage and seal failure via current signature analysis

• US 20060071666A1
• Filed: 07/27/2005
• Published: 04/06/2006
• Est. Priority Date: 09/28/1999
• Status: Active Grant

• Current Assignee: Rockwell Automation Technologies Incorporated (Allen-Bradley Co., Inc.)
• Original Assignee: Reliance Electric Technologies LLC (ABB Limited)
• Inventors: Discenzo, Frederick M., Unsworth, Peter J., Babu, Vetcha Sarat
• Granted Patent: US 7,389,278 B2
• Time in Patent Office: Days
• US Class Current: 324/522
• CPC Class Codes: F04D 15/0077 Safety measures F04D15/02 takes precedence
• CPC Class Codes: F04D 15/0088 Testing machines
• CPC Class Codes: F04D 29/669 especially adapted for liquid pumps F04D29/18 takes precedence
• CPC Class Codes: G01R 31/2846 using hard- or software simulation or using knowledge-based systems, e.g. expert systems, artificial intelligence or interactive algorithms
• CPC Class Codes: G01R 31/343 in operation
• CPC Class Codes: G06N 3/088 Non-supervised learning, e.g. competitive learning
• CPC Class Codes: G06N 5/048 Fuzzy inferencing
• CPC Class Codes: Y10S 706/904 Manufacturing or machine, e.g. agricultural machinery, machine tool
• CPC Class Codes: Y10S 706/905 Vehicle or aerospace
• CPC Class Codes: Y10S 706/906 Process plant
• CPC Class Codes: Y10S 706/907 Power plant

Abstract

A system and method is provided for monitoring the operating condition of a pump by evaluating fault data encoded in the instantaneous current of the motor driving the pump. The data is converted to a frequency spectrum which is analyzed to create a fault signature having fault attributes relating to various fault conditions associated with the pump. The fault signature is then input to a neural network that operates in conjunction with a preprocessing and post processing module to perform decisions and output those decisions to a user interface. A stand alone module is also provided that includes an adaptive preprocessing module, a one-shot unsupervised neural network and a fuzzy based expert system to provide a decision making module that operates with limited human supervision.

Claims

1. A system for monitoring the condition of a pump driven by a motor, comprising:

a sensor operatively coupled to a power lead of the motor, the sensor adapted to obtain at least one current signal relating to the operation of the pump; and

an artificial neural network operatively coupled to the sensor, the artificial neural network being adapted to detect at least one fault relating to the operation of the pump from the at least one current signal.

2. The system of claim 1, wherein the sensor is a current transformer.

3. The system of claim 1, wherein the artificial neural network is an unsupervised neural network

4. The system of claim 3, wherein the unsupervised neural network is a one-shot unsupervised neural network.

5. The system of claim 1, further including a processor operatively coupled to the sensor, the processor adapted to generate fast fourier transforms of the at least one current signal.

7. The system of claim 1, wherein the processor is adapted to preprocess the fault signature by eliminating outlayers, and performing scaling and bifurcation of the fault signature data set.

8. The system of claim 1, wherein the artificial neural network is also adapted to detect at least one fault relating to the operation of the motor from the at least one current signal.

18. The method of claim 1, wherein the artificial neural network is also adapted to detect at least one fault relating to the operation of the motor from the at least one current signal.

6. The system of claim 6, wherein the processor is adapted to create a fault signature from a current spectrum created from the fast fourier transformation of the at least one current signal wherein the fault signature contains distinguishable attributes relating to faults of the pump.

9. A method for monitoring the condition of a pump driven by a motor, comprising the steps of:

collecting a first sample of current data signal relating to the operation of the pump;

inputting the first sample of current data signal to a neural network,

collecting a second sample of current data signal relating to the operation of the pump; and

inputting the second sample of current data signal to the neural network, wherein any differences between the first signal and the second signal will be generated as a change in condition signal by the neural network, any change of condition signal representing a pump fault condition.

10. The method of claim 9, wherein the step of collecting at least one sample of current data signal relating to the operation of the machine is accomplished by employing a current transformer.

11. The method of claim 9, wherein the artificial neural network is an unsupervised neural network.

12. The method of claim 9, wherein the unsupervised neural network is a one-shot unsupervised neural network.

13. The method of claim 9, wherein the first sample of current data is a training sample for training the neural network.

14. The method of claim 9, wherein the first and the second sample of current data is transformed to a frequency spectrum by fast fourier transformation.

15. The method of claim 14, wherein the frequency spectrum is transformed to a fault signature and the fault signature is preprocessed by eliminating outlayers, and performing scaling and bifurcation of the fault signature before being fed to the neural network.

16. The method of claim 15, wherein output of the neural network is post processed to determine the specific fault condition represented by a change of condition signal.

17. The method of claim 16, wherein the step of post processing includes comparing the specific fault condition to fault attributes found in a fault signature of the pump.

19. A stand alone decision module adapted to receive a current signal from a machine and facilitate diagnosing the state of the machine by determining if the current signal contains fault data relating to the state of the machine, the decision module comprising:

a neural network operatively coupled to a sensor, the neural network adapted to synthesize a change in condition signal from the sampled current data;

a preprocessing portion operatively coupled to the neural network, the preprocessing portion adapted to condition the current signal prior to inputting the current signal into the neural network; and

a post processing portion operatively coupled to the neural network, the post processing portion adapted to determine whether the change in condition signal is due to a fault condition related to the state of the machine.

20. The module of claim 19, wherein the artificial neural network is an unsupervised neural network.