Method and system for modeling the performance of a gas turbine engine

• US 8,306,791 B2
• Filed: 12/21/2009
• Issued: 11/06/2012
• Est. Priority Date: 12/21/2009
• Status: Expired due to Fees

• Current Assignee: United Technologies Corporation (Raytheon Technologies Corporation d/b/a RTX)
• Original Assignee: United Technologies Corporation (Raytheon Technologies Corporation d/b/a RTX)
• Inventors: Yerramalla, Sampath K., Butler, Steven Wayne, Donat, William
• Primary Examiner(s): Jones, Hugh
• Application Number:
  US12/643,579
• Publication Number:
  US 20110153295A1
• Time in Patent Office: 1,051 Days
• Field of Search: 703/2, 703/8
• US Class Current: 703/2
• CPC Class Codes: F02C 9/00 Controlling gas-turbine plants; Controlling fuel supply in air- breathing jet-propulsion plants controlling air intakes F02C7/057; controlling turbines F01D; controlling compressors F04D27/00; controlling in general G05
• CPC Class Codes: F05D 2270/709 with neural networks
• CPC Class Codes: G05B 17/02 electric
• CPC Class Codes: G05B 23/024 Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
• CPC Class Codes: G05B 23/0254 based on a quantitative model, e.g. mathematical relationships between inputs and outputs; functions: observer, Kalman filter, residual calculation, Neural Networks
• CPC Class Codes: G05B 23/0283 Predictive maintenance, e.g. involving the monitoring of a system and, based on the monitoring results, taking decisions on the maintenance schedule of the monitored system; Estimating remaining useful life [RUL] preventive maintenance, i.e. planning maintenance according to the available resources without monitoring the system G06Q10/06
• CPC Class Codes: G06F 2111/10 Numerical modelling
• CPC Class Codes: G06F 30/20 Design optimisation, verification or simulation optimisation, verification or simulation of circuit designs G06F30/30

Abstract

A method for modeling the performance of a gas turbine engine is provided. The method includes the steps of: 1) providing a processor; 2) inputting flight condition parameter data and engine output parameter data into a gas turbine engine model operating on the processor, which model includes a physics-based engine model that uses the flight condition parameter data to produce estimated engine output parameter data, and determines residuals from the engine output parameter data and the estimated engine output parameter data; 3) partitioning the flight condition parameter data and residuals into training data and testing data; 4) performing a correlation reduction on the training data, which analysis produces correlation adjusted training data; 5) performing an orientation reduction on the correlation adjusted training data, which reduction produces orientation adjusted training data; 6) reviewing the orientation adjusted training data relative to at least one predetermined criteria, and iteratively repeating the steps of performing a correlation reduction and an orientation reduction using the orientation adjusted training data if the criteria is not satisfied, and if the criteria is satisfied outputting the orientation adjusted training data; 7) producing estimated corrections to the orientation adjusted training data using one or more neural networks; 8) evaluating the neural adjusted data using the partitioned testing data; and 9) modeling the performance of the gas turbine using the estimated corrections to the orientation adjusted training data.

Claims

1. A method for modeling the performance of a gas turbine engine, comprising the steps of:

providing a processor;

inputting flight condition parameter data and engine output parameter data into a gas turbine engine model operating on the processor, which model includes a physics-based engine model that uses the flight condition parameter data to produce estimated engine output parameter data, and determines residual data from the engine output parameter data and the estimated engine output parameter data;

partitioning the flight condition parameter data and residual data into training data and testing data;

performing a correlation reduction on the training data, which analysis produces correlation adjusted training data;

performing an orientation reduction on the correlation adjusted training data, which reduction produces orientation adjusted training data;

reviewing the orientation adjusted training data relative to at least one predetermined criteria, and iteratively repeating the steps of performing a correlation reduction and an orientation reduction using the orientation adjusted training data, and reviewing if the criteria is not satisfied, and if the criteria is satisfied outputting the orientation adjusted training data;

producing estimated corrections to the orientation adjusted training data using one or more neural networks;

evaluating the neural adjusted data using the partitioned testing data; and

modeling the performance of the gas turbine using the estimated corrections to the orientation adjusted training data.

2. The method of claim 1, wherein the step of performing a correlation reduction includes performing a principal component analysis (PCA) on the training data to reduce correlations within the training data, which analysis produces the correlation adjusted training data, and the method further includes the step of performing a reverse PCA on the correlation adjusted training data.

3. The method of claim 2, wherein the PCA is performed on all of the training data.

4. The method of claim 1, further comprising the step of filtering the flight condition parameter data to remove outliers.

5. The method of claim 4, wherein the filtering step includes performing a PCA on the flight condition parameter data to remove the outliers.

6. The method of claim 5, wherein the flight condition parameter data are organized in matrix of engine event by flight condition parameter, and the filtering step includes removing data from any engine events having outliers.

7. The method of claim 6, wherein flight condition parameter data in less than all of the flight condition parameters is considered in the filtering step.

8. A system for modeling the performance of a gas turbine engine, comprising:

a processor adapted to receive flight condition parameter data and engine output parameter data, and adapted to have a physics based engine model that uses the flight condition parameter data to produce estimated engine output parameter data, and determines residual data from the engine output parameter data and the estimated engine output parameter data, wherein the processor includes;

a partitioning module for partitioning the flight condition parameter data and residual data into training data and testing data;

a training module having a correlation reduction module for reducing one or more correlations in the training data, which module produces correlation adjusted training data, an orientation reduction module for reducing one or more orientations in the correlation adjusted training data, which module produces orientation adjusted training data, and a review module for reviewing the orientation adjusted training data relative to at least one predetermined criteria, and directing the orientation adjusted training data iteratively within the training module if the criteria is not satisfied, and if the criteria is satisfied outputting the orientation adjusted training data;

a neural network module for producing estimated corrections to the orientation adjusted training data using one or more neural networks; and

an evaluation model for evaluating the neural adjusted data using the partitioned testing data; and

wherein the processor is further adapted to model the performance of the gas turbine using the estimated corrections to the orientation adjusted training data.

9. The system of claim 8, wherein the correlation reduction module is adapted to perform a principal component analysis (PCA) on the training data to reduce correlations within the training data, which analysis produces the correlation adjusted training data, and to perform a reverse PCA on the correlation adjusted training data.

10. The system of claim 9, wherein the correlation reduction module is adapted to perform a PCA on all of the training data.

11. The system of claim 8, wherein the processor is further includes a filtering module for filtering the flight condition parameter data to remove outliers.

12. The system of claim 11, wherein the filtering module is adapted to perform a PCA on the flight condition parameter data to remove the outliers.

13. The system of claim 12, wherein filtering module is adapted to consider flight condition parameter data associated with less than all of the flight condition parameters.