Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Modeling method and application of aeroengine performance based on fitting sensitivity

An aero-engine and modeling method technology, applied in special data processing applications, geometric CAD, etc., can solve problems such as under-fitting and over-fitting, and achieve the effect of reasonable fitting degree and good generalization ability.

Active Publication Date: 2020-06-09
HARBIN INST OF TECH
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0008] The purpose of the present invention is to solve the problems of overfitting and underfitting in existing models, and propose a generalized approximate modeling method based on the performance of aeroengines with fitting sensitivity

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Modeling method and application of aeroengine performance based on fitting sensitivity
  • Modeling method and application of aeroengine performance based on fitting sensitivity
  • Modeling method and application of aeroengine performance based on fitting sensitivity

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0036] Specific implementation mode one: the specific process of the generalized approximate modeling method based on the aeroengine performance of fitting sensitivity is:

[0037] Fit Sensitivity Analysis

[0038] Denote the training samples as X=[x 1 ,x 2 ,...,x n ], the fitting value is expressed as Y=[y 1 ,y 2 ,...,y n ], then when X and Y have the same initial value (x 1 =y 1 ), the degree of fit can be expressed as the fitted sensitivity model dY / dX.

[0039] (a) When dY / dX→1, y k to x k Overfitting, that is, ΔY≈ΔX. Fitted value y k The change trend of x k consistent, such as figure 1 shown.

[0040] (b) When dY / dX>1, y k to x k Underfitting and ΔY>ΔX, such as figure 2 shown. Fitted value y k Expanded training sample x k The trend of change, at this time y k is unstable and with x k Oscillation due to the change of , thus obtaining unstable prediction results. This state is called "over-underfitting".

[0041] (c) When 0k to x k Underfitting an...

specific Embodiment approach 2

[0068] Specific embodiment two: the difference between this embodiment and specific embodiment one is: the specific process of establishing the generalized approximate model of the aeroengine performance based on the fitting sensitivity in the step one is:

[0069] Set the value of dY / dX in the interval (0, 1), when x 1 =y 1 When building a fitted sensitivity model:

[0070]

[0071] where X=[x 1 ,x 2 ,...,x n ] is the training sample, Y=[y 1 ,y 2 ,...,y n ] is the fitting value;

[0072] (a) when|x k -y k | becomes larger, the fitted value y k Deviate from the training sample x k . because the fitted value y k Should reflect the main trend of the training sample, so the training sample x k Contains strong noise and fluctuations. in order to make y k Get the slower main trend, y k to x k The sensitivity of dy k / dx k should be lowered.

[0073] (b) When|x k -y k | becomes smaller, the fitted value y k Approximate training samples x k . because the ...

specific Embodiment approach 3

[0089] Embodiment 3: This embodiment differs from Embodiment 1 or Embodiment 2 in that the value of p is set in step 21: 1

[0090] When training the prediction model parameters, due to the large number of training samples, it is difficult to ensure that any sample segment will not be over-fitting. In order to improve the prediction accuracy, only the last p sample points of the training samples are constrained to satisfy the suppression of over-fitting. Underfitting constraints. The maximum length of p can be taken to the training sample length n, and the minimum can be taken to be 1, that is, p∈[1,n] and p∈N.

[0091] When p→1, the model only constrains the last few points, so that the fitted value of the training sample does not fall into overfitting and underfitting, but few points do not contain x k trend information, leading to inaccurate forecast results.

[0092] When p→n, the model constrains the entire training sample segment so that the fitted value of the tr...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention provides a generalized approximate modeling method based on fitting sensitivity for aero-engine performance, and model application. The invention relates to a generalized approximate modeling method for aero-engine performance, and model application. The over-fitting and under-fitting problems in an existing model are solved. The generalized approximate modeling method comprises the steps of (1) establishing a generalized approximate model based on fitting sensitivity for aero-engine performance; and (2) solving parameters of the established generalized approximate model based on fitting sensitivity for aero-engine performance. The step (2) comprises the steps of (21) setting a value of p; (22) setting an adjusting coefficient [epsilon] of suppressing excessive under fitting; (23) setting an integral compression coefficient a for xk according to the generalized approximate model based on fitting sensitivity for aero-engine performance, and a compression coefficient b for xk according to |xk-yk|; and setting an integral migration amount c1 according to the step (21) to the step (23). The generalized approximate modeling method is used in the aero-engine operation, maintenance and safety engineering field.

Description

technical field [0001] The invention relates to a generalized approximate modeling method for the performance of an aero-engine and an application of the model. Background technique [0002] Aeroengine performance prediction technology is extremely important in the field of operation and maintenance and safety engineering. In January 2008, the electrical system of a Qantas Boeing 747-400 failed during flight and all four engines failed. In the United States, a study of 7,571 transport aircraft that failed due to mechanical failure from 1980 to 2001 found that landing gear and turbine engines were the most likely to fail. In addition, the maintenance cost of commercial aircraft in the world in 2007 reached tens of billions of dollars, but 31% of the expenditure was used on engine maintenance. Therefore, the performance prediction technology of aero-engine is of great significance to reduce the probability of aircraft crash and reduce the maintenance cost. [0003] Due to t...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06F30/17
CPCG06F30/17
Inventor 林琳王芳钟诗胜
Owner HARBIN INST OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products