Automated method for the systematic interpretation of resonance peaks in spectrum data

• US 5,623,579 A
• Filed: 05/17/1995
• Issued: 04/22/1997
• Est. Priority Date: 05/27/1993
• Status: Expired due to Fees

• Current Assignee: Martin Marietta Energy Systems, Inc.
• Original Assignee: Martin Marietta Energy Systems, Inc.
• Inventors: Damiano, Brian, Wood, Richard T.
• Primary Examiner(s): Hafiz, Tariq R.
• Application Number:
  US08/443,292
• Time in Patent Office: 706 Days
• Field of Search: 73/23.34, 382/155-159, 395/20-25, 395/27
• US Class Current: 706/17
• CPC Class Codes: G06F 2218/00 Aspects of pattern recognition specially adapted for signal processing
• CPC Class Codes: G21C 17/00 Monitoring; Testing measuring in general G01; Maintaining
• CPC Class Codes: G21D 3/001 Computer implemented control
• CPC Class Codes: Y02E 30/00 Energy generation of nuclear origin
• CPC Class Codes: Y02E 30/30 Nuclear fission reactors

Abstract

A method for spectral signature interpretation. The method includes the creation of a mathematical model of a system or process. A neural network training set is then developed based upon the mathematical model. The neural network training set is developed by using the mathematical model to generate measurable phenomena of the system or process based upon model input parameter that correspond to the physical condition of the system or process. The neural network training set is then used to adjust internal parameters of a neural network. The physical condition of an actual system or process represented by the mathematical model is then monitored by extracting spectral features from measured spectra of the actual process or system. The spectral features are then input into said neural network to determine the physical condition of the system or process represented by the mathematical. More specifically, the neural network correlates the spectral features (i.e. measurable phenomena) of the actual process or system with the corresponding model input parameters. The model input parameters relate to specific components of the system or process, and, consequently, correspond to the physical condition of the process or system.

Claims

1. A method for spectral signature interpretation comprising the steps of:

measuring and transforming raw sensor signals of a system into frequency spectra;

isolating from said frequency spectra a plurality of resonance frequencies corresponding to operating conditions of components within the system;

developing a neural network from a mathematical model based on modeled resonant frequencies of components of said system, the resonant frequencies being adjusted to characterize the steady state condition of the system;

feeding the plurality of resonance frequencies into the neural network; and

outputting from the neural network a diagnosis of the actual physical conditions of said components of the system.

2. A method according to claim 1, wherein said step of measuring and transforming raw sensor signals comprises measuring and transforming sensor signals derived from neutron noise.

3. A method according to claim 2, wherein said isolating step comprises applying said frequency spectra to a frequency spectrum decomposition algorithm.

4. A method according to claim 3, further comprising the step of adjusting internal parameters of said neural network based upon a training set generated by a mathematical model of neutron noise produced by a nuclear power plant.

5. A method according to claim 4, wherein said step of adjusting internal parameters includes the step of forming said training set comprising table sets of model input parameters defining specific components of said nuclear power plant and measurable phenomena generated by said mathematical model based upon input of said model input parameters.

6. A method according to claim 1, wherein the isolating step comprises applying said frequency spectra to a frequency spectrum decomposition algorithm.

7. A method according to claim 1, further comprising the step of forming a training set for the neural network, said training set being generated by said mathematical model to adjust internal parameters of said neural network.

8. A method according to claim 7, wherein said step of forming said training set comprises generating table sets consisting of model input parameters defining specific components of said system and measurable phenomena generated by said mathematical model based upon input of said model input parameters.

9. A method according to claim 1, further comprising the step of applying said parameters representing actual physical conditions of said system in conjunction with said mathematical model to estimate the actual physical condition of said system.

10. A method according to claim 1, wherein said feeding step comprises the step of performing interpolation.

11. A method diagnosing operations of a system having a plurality of components, the method comprising the steps of:

creating a mathematical model of the system including modeled resonance frequencies of components within the system;

developing a neural network training set from said mathematical model;

adjusting internal parameters of a neural network based upon said neural network training set;

extracting spectral features which include actual resonance frequencies from measured spectra of the system during operation;

inputting said actual spectral features into said neural network to determine actual physical conditions of the system.

12. A method according to claim 11, wherein said step of creating said mathematical model comprises creating a mathematical model of a nuclear power plant, the model including the modeled resonance frequencies.

13. A method according to claim 12, wherein said step of extracting spectral features comprises extracting spectral features of neutron noise produced by said nuclear power plant from measured spectra of said neutron noise.

14. A method according to claim 13, wherein said said step of extracting spectral features further comprises extracting spectral features from measured spectra of said neutron noise based upon a frequency spectrum decomposition algorithm.

15. A method according to claim 14, further comprising the step of adjusting internal parameters of said neural network based upon training sets generated by a mathematical model of said neutron noise produced by said nuclear power plant.

16. A method according to claim 15, wherein said step of developing said neural network training set comprises developing a neural network training set comprising model input parameters defining specific components of said nuclear power plant and measurable phenomena generated by said mathematical model based upon input of said model input parameters.

17. A method according to claim 11, wherein said step of extracting spectral features comprises using a frequency spectrum decomposition algorithm.

18. A method according to claim 11, wherein said step of developing said neural network training set comprises developing a neural network training set of model input parameters defining specific components of said system or process and measurable phenomena generated by said mathematical model based upon input of said model input parameters.

19. A method according to claim 11, further comprising the step of generating numerical values representing actual physical conditions of said system or process through said neural network, said numerical values are used in conjunction with said mathematical model to estimate the actual physical condition of said system or process.

20. A method according to claim 11, said step of inputting said spectral features into said neural network comprises the step of performing interpolation.