System and method for diagnosing jet engine conditions

• US 6,574,613 B1
• Filed: 03/02/2000
• Issued: 06/03/2003
• Est. Priority Date: 02/27/1998
• Status: Expired due to Fees

• Current Assignee: MTU Aero Engines GmbH (MTU)
• Original Assignee: MTU Aero Engines GmbH (MTU)
• Inventors: Moreno-Barragan, Jorge
• Primary Examiner(s): Follansbee, John A.
• Assistant Examiner(s): Booker, Kelvin
• Application Number:
  US09/403,878
• Time in Patent Office: 1,188 Days
• Field of Search: 706/16, 706/20, 706/15, 706/905, 706/913, 706/911, 706/2, 731/17, 73/3, 731/16, 700/48, 700/47, 700/49, 700/291, 701/27, 701/29, 701/106, 244/62
• US Class Current: 706/16
• CPC Class Codes: G01N 33/0034 comprising neural networks or related mathematical techniques
• CPC Class Codes: G05B 19/406 characterised by monitoring or safety G05B19/19 takes precedence
• CPC Class Codes: Y10S 706/905 Vehicle or aerospace
• CPC Class Codes: Y10S 706/913 Vehicle or aerospace

Abstract

A system and a method for diagnosis of engine conditions are proposed. In particular, the system and the method are directed to an extraction of features from different information sources and to their processing. These features, together with a series connection of two neural networks, form the crux of the system and method, so that a dependable diagnosis of engine conditions, particularly an error recognition is possible. As a result thereof, maintenance corresponding to the current engine condition is enabled.

Claims

1. A system for diagnosis of jet engine conditions, comprising:

means for supplying statistical information about an error quota of individual jet engine components from a data bank;

a plurality of measurement sensors for acquiring physical information about the jet engine selected from pressures and temperatures in various engine levels and parameters from a particle analysis in used oil and in engine exhaust gases as well as parameters from an analysis of the gas path;

a plurality of measurement sensors for acquiring vibration information in the time domain from the jet engine;

vibration analysis means for generating vibration information in the frequency domain from the vibration information in the time domain;

a module for feature extraction for processing the physical information and/or the statistical information and the vibration information in the time and frequency domain and for the extraction of a number of features that describe the jet engine condition;

a first neural network to which the features are applied for classification of the features, for identification of relationships and dependencies between features and for corresponding implementation of an information compression and for output of parameters, whereby the first neural network comprises an input layer, an intermediate layer and an output layer of neurons, the input layer comprises more neurons than the intermediate layer and the intermediate layer comprises more neurons than the output layer, and the neurons of the input layer are connected to the neurons of the intermediate layer, which are connected to the neurons of the output layer via a plurality of connecting elements having variable weighting coefficients;

first training means for supplying training input signals to the first neural network and for comparison of an output signal being outputted by the first neural network in response thereto to a training input signal and for the modification of variable weighting coefficients of the first neural network;

a second neural network to which the parameters output by the first neural network are applied for classification of the parameters, for recognition of relationships between the parameters and specific error constellations, for corresponding implementation of an information linkage and for output of a diagnosis signal, the second neural network comprises an input layer, an intermediate layer and an output layer of neurons, the input layer and the output layer comprise fewer neurons than the intermediate layer, and the neurons of the input layer are connected to the neurons of the intermediate layer, which are connected to the neurons of the output layer via a plurality of connecting elements having variable weighting coefficients; and

a second training means for supplying training input signals to the second neural network and for comparing the output signal obtained from the second neural network in response thereto to a training input signal and for modifying variable weighting coefficients of the second neural network.

2. A system according to claim 1, wherein the module for feature extraction employs physical parameters selected from a group consisting of oil consumption for specific engine runs, power reference numbers including pressure and temperature in specific engine levels, parameters from a particle analysis in used oil and in engine exhaust gases as well as parameters from an analysis of the gas path.

3. A system according to claim 1, wherein the module for feature extraction extracts features selected from a group consisting of effective values, properties of envelopes, modulations, absolute values, performance analyses, statistical parameters, distribution functions and wavelet analysis of the vibration information in the time domain.

4. A system according to claim 1, wherein the vibration analysis means handles the vibration signals in the time domain and determines corresponding vibration information in the frequency domain therefrom.

5. A system according to claim 1, wherein the module for feature extraction employs an information presentation in the form of a waterfall diagram, handles this information presentation with image processing methods and determines corresponding features therefrom from the vibration information in the frequency domain.

6. A system according to claim 1, wherein the module for feature extraction also implements geometrical considerations of overall image or specific image regions.

7. A system according to claim 1, wherein the module for feature extraction also numerically acquires the vibration information of waterfall diagrams;

and utilizes methods selected from methods from matrix and vector calculation and methods for system identification in the frequency domain for acquiring features from the vibration information in the frequency domain.

8. A system according to claim 1, wherein neural networks in combination with fuzzy logic or pure fuzzy logic circuits are provided instead of the first and second neural networks.

9. A method for diagnosis of jet engine conditions, comprising the steps of:

supplying statistical information about an error quota of individual jet engine components resulting from an evaluation of a data band;

acquiring physical information about the jet engine selected from pressures and temperatures in various engine levels with a plurality of measurement sensors, parameters from a particle analysis in used oil and in jet engine exhaust gases as well as parameters from an analysis of the gas path;

acquiring vibration information in the time domain from the jet engine with a plurality of measurement sensors;

generating vibration information in the frequency domain from the vibration information in the time domain with vibration analysis means;

processing the physical information and/or the statistical information and the vibration information in the time and frequency domain and extracting a number of features that describe the jet engine condition with a module for feature extraction;

classifying the features and identifying relationships and dependencies between features and corresponding implementation of an information compression and output of parameters by a first neural network to which the features are applied, wherein the first neural network comprises an input layer, an intermediate layer and an output layer, the input layer comprises more neurons than the intermediate layer and the intermediate layer comprises more neurons than the output layer and the neurons of each layer are connected to the neurons of adjacent layers via a plurality of connecting elements having variable weighting coefficients;

supplying training input signals to the first neural network and comparing an output signal being outputted in response thereto by the first neural network to a training input signal and modifying the variable weighting coefficients of the first neural network;

classifying the parameters, recognition of relationships between the parameters and specific error constellations, corresponding implementation of an information linkage and output of a diagnosis signal by means of a second neural network to which the parameters being outputted by the first neural network are applied, wherein the second neural network comprises an input layer, an intermediate layer and an output layer of neurons, wherein the input layer and the output layer have fewer neurons than the intermediate layer, and the neurons of each layer are connected to the neurons of adjacent layers via a plurality of connecting elements having variable weighting coefficients;

supplying training input signals to the second neural network and comparing an output signal obtained in response thereto from the second neural network to a training input signal, and modifying the variable weighting coefficients of the second neural network.

10. A method according to claim 9, wherein the acquired, physical parameters are selected from all consumption at specific engine runs, power reference numbers including pressure and temperature in specific engine levels, parameters from a particle analysis in used oil and in jet engine exhaust gases as well as parameters from an analysis of the gas path.

11. A method according to claim 9, wherein specific engine components are classified as susceptible in the feature extraction on the basis of the statistical information and are output in the form of features.

12. A method according to claim 9, wherein the extracted features are selected from a group consisting of effective values, properties of envelopes, modulations, absolute values, power analyses, statistical parameters, distribution functions and wavelet analysis of the vibration information in the time domain.

13. A method according to claim 9, wherein the vibration information in the time domain is processed with vibration analysis means and corresponding vibration information in the frequency domain are determined therefrom.

14. A method according to claim 9, wherein an information presentation in the form of a waterfall diagram is employed when processing vibration information in the frequency domain, this information presentation being handled with image processing methods and corresponding features from the vibration information in the frequency domain being determined therefrom.

15. A method according to claim 9, wherein geometrical considerations of overall image or of specific image regions are implemented as well when processing vibration information in the frequency domain.

16. A method according to claim 9, wherein information of waterfall diagrams are also numerically acquired when processing vibration information in the frequency domain, and methods selected from matrix and vector calculation and methods for system identification in the frequency domain are utilized for acquiring vibration information in the frequency domain.

17. A method according to claim 9, wherein the classification, identification, information compression and output of parameters and the classification, recognition of relationships, information linkage and output of a diagnosis signal is implemented by neural networks in combination with fuzzy logic or by pure fuzzy logic circuits instead of by the first and second neural networks.