Diagnostic system responsive to learned audio signatures

• US 5,521,840 A
• Filed: 04/07/1994
• Issued: 05/28/1996
• Est. Priority Date: 04/07/1994
• Status: Expired due to Fees

• Current Assignee: Westinghouse Electric Company LLC (Brookfield Asset Management Incorporated)
• Original Assignee: Westinghouse Electric Company LLC (Brookfield Asset Management Incorporated)
• Inventors: Bednar, Fred H.
• Primary Examiner(s): Cosimano, Edward R.
• Assistant Examiner(s): PIPALA, EDWARD J
• Application Number:
  US08/224,467
• Time in Patent Office: 782 Days
• Field of Search: 364/506, 364/507, 364/508, 364/510, 364/550, 364/575, 364/581, 395/914, 395/915, 395/2.41, 395/21, 395/23, 395/24, 137/551, 137/554
• US Class Current: 702/183
• CPC Class Codes: G21C 17/00 Monitoring; Testing measuring in general G01; Maintaining
• CPC Class Codes: Y02E 30/30 Nuclear fission reactors
• CPC Class Codes: Y10S 706/914 Process plant
• CPC Class Codes: Y10T 137/8158 With indicator, register, recorder, alarm or inspection means
• CPC Class Codes: Y10T 137/8242 Electrical

Abstract

Loose parts are sensed moving in a conduit carrying a flowing material, such as the cooling circuit of a pressurized water nuclear reactor. An acoustic pickup produces an electrical signal with vibration of the conduit due to impact of the loose part, and background noise. A signal processor encodes the values of distinct parameters of the electrical signal such as amplitude, amplitude at particular frequencies, etc., in an ongoing manner, producing discrete output values. These outputs are coupled as inputs to a neural network with physical or logical neuron cells loaded with weighting factors affecting the strength and polarity of neural interconnections. The factors represent the acoustic signature of the loose part. Products of the input values and the weighting factors are summed to produce one or more neural network outputs, compared to a threshold. The sum normally varies randomly, but has a strong swing when the pattern is encountered, due to the factors emphasizing the pattern over background noise. The threshold comparison operates a display or alarm. The weighting factors are learned by repeating empirical tests and correlating the factors to the signal to minimize error.

Claims

1. A diagnostic apparatus for detecting a loose part moving in a conduit carrying a flowing material, comprising:

an acoustic pickup operable to produce an electrical signal in response to vibration of the conduit, including vibration caused by the loose part and background vibration;

a signal processing means coupled to the electrical signal of the acoustic pickup, the signal processing means being operable to encode a plurality of values representing distinct parameters of the electrical signal, in an ongoing manner, the signal processing means providing a plurality of discrete outputs representing values of the distinct parameters;

a neural network having a plurality of neuron cells, each defining at least one weighting factor, the neuron cells having inputs coupled to the values of the distinct parameters and being operable to weight the values of the distinct parameters to produce a neuron cell output; and

,

at least one summing cell producing a sum of a product of the weighting factors and the values from each of the neuron cells, and wherein the weighting factors complement at least one combination of the distinct parameters defining a signature of the loose part, whereby upon passage of the loose part the summing cell produces an output that is substantially displaced from a level resulting from the background noise.

2. The diagnostic apparatus according to claim 1, further comprising means for loading the weighting factors by operating the acoustic pickup and signal processing means during known passages of the loose part, and accumulating the weighting factors from observed values of the distinct parameters.

3. The diagnostic apparatus according to claim 2, further comprising a threshold detection means coupled to an output of the summing cell, the threshold detection means producing an output for indicating passage of the loose part when the output of the summing cell exceeds a predetermined value.

4. The diagnostic apparatus according to claim 1, wherein at least certain of the neuron cells are coupled to at least one other of the neuron cells, and wherein an output of at least one of said certain of the neuron cells modifies an output of said at least one other of the neuron cells.

5. The diagnostic apparatus according to claim 4, wherein the neuron cells are arranged in levels proceeding from inputs coupled to the values of the distinct parameters, to the summing cell, and wherein outputs of at least some of the neurons on a higher level are coupled as inputs to neurons on a lower level.

6. The diagnostic apparatus according to claim 5, further comprising means for loading the weighting factors by operating the acoustic pickup and signal processing means during known passages of the loose part, and accumulating the weighting factors from observed values of the distinct parameters.

7. The diagnostic apparatus according to claim 6, wherein the distinct parameters of the electrical signal include at least a subset of variables chosen from the group consisting of signal amplitude, signal amplitude over at least one discrete frequency range, duration of signal amplitude over time, concentration of energy in at least one discrete frequency range, integration of signal amplitude, differentiation of signal amplitude, sum and/or difference functions, logical functions, and simultaneous occurrence of predetermined conditions in two or more of said variables.

8. The diagnostic apparatus according to claim 7, further comprising a threshold detection means coupled to an output of the summing cell, the threshold detection means producing an output for indicating passage of the loose part when the output of the summing cell exceeds a predetermined value.

9. The diagnostic apparatus according to claim 8, comprising a plurality of networks of said neurons, coupled to the values of the distinct parameters, each of the plurality of networks defining weighting factors representing at least one of distinct aspects of vibration of the loose part, and distinct aspects of vibration of loose parts of different types.

10. The diagnostic apparatus according to claim 9, further comprising a summing cell for each of the networks, and means for discriminating for passage of loose parts from outputs of said summing cells.

11. A method for detecting passage of a loose part moving in a conduit carrying a flowing material, comprising:

collecting an acoustic signal from vibration of the conduit caused by the loose part and by background vibration;

processing the acoustic signal to encode a plurality of values representing distinct parameters of the acoustic signal, said processing including determining levels of the distinct parameters for at least a subset of variables chosen from the group consisting of signal amplitude, signal amplitude over at least one discrete frequency range, duration of signal amplitude over time, concentration of energy in at least one discrete frequency range, integration of signal amplitude, differentiation of signal amplitude, sum and difference functions, logical functions, and simultaneous occurrence of predetermined conditions in at least two of said variables;

weighting the values of the distinct parameters by weighting factors representing characteristic aspects of vibration of the loose part;

weighting relationships between the values of the distinct parameters as a function of relatively higher and relatively lower levels of at least two of the distinct parameters defining a pattern characteristic of vibration of the loose part;

summing the weighted levels and the weighted relationships, and comparing a sum thereof to a threshold to discriminate for passage of the loose part, factors for weighting the values and the relationships being chosen to complement an acoustic signal pattern characteristic of the loose part.

12. The method according to claim 11, further comprising accumulating the factors for weighting the values and the relationships by observing known passages of the loose part.