CN108205310A - Gas path failure recognition methods in a kind of aero-engine envelope curve based on ELM filtering algorithms - Google Patents
Gas path failure recognition methods in a kind of aero-engine envelope curve based on ELM filtering algorithms Download PDFInfo
- Publication number
- CN108205310A CN108205310A CN201810013686.6A CN201810013686A CN108205310A CN 108205310 A CN108205310 A CN 108205310A CN 201810013686 A CN201810013686 A CN 201810013686A CN 108205310 A CN108205310 A CN 108205310A
- Authority
- CN
- China
- Prior art keywords
- elm
- engine
- sample
- envelope curve
- gas path
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0243—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/24—Pc safety
- G05B2219/24065—Real time diagnostics
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses gas path failure recognition methods in a kind of aero-engine envelope curve based on ELM filtering algorithms, this method includes:ELM model topology parameters are trained using Kalman filtering algorithm;By the trained ELM models of filtering algorithm for gas path failure identification in engine envelope curve.The present invention is solved in existing envelope curve in Fault Diagnosis of Aircraft Engine Gas Path, the engine diagnosis of conventional data-driven is not strong in different operating point generalization ability, the problem of precision is not high, suitable for the engine failure pattern-recognition of the different operating point in flight envelope, play the role of actively promoting for engine health control, reduction maintenance cost.
Description
Technical field
The invention belongs to Fault Diagnosis of Aircraft Engine Gas Path technical fields more particularly to one kind to be based on ELM filtering algorithms
Aero-engine envelope curve in gas path failure recognition methods.
Background technology
Heart of the aero-engine as aircraft, complicated and working environment are severe.Engine Failure Diagnostic Technology
It is to ensure engine performance and reliability, reduces the important means of working service cost.During aero-engine military service, portion
Part performance can occur slowly to degenerate.Furthermore, it is also possible to generation part health parameters are mutated.Gas path component failure affects aviation hair
The performance and reliability of motivation, it is necessary to be diagnosed to it.Extreme learning machine (Extreme Learning Machine,
ELM) it is a kind of Fast Learning method for growing up on the basis of neural network theory, in necks such as data mining, pattern-recognitions
Domain has been widely used.The essence of ELM is a Single hidden layer feedforward neural networks (Single hidden-layer Feed-
forward Networks,SLFNs).Traditional feedforward neural network is made of input layer, hidden layer and output layer.ELM models
The part different from traditional SLFNs is that the corresponding input weights of ELM and biasing are generated by random assignment, so as to shape
The linear system is solved into a fixed linear system of parameter, then with least square method.Huang et al. by theoretical and
It is a kind of efficient and effective learning algorithm that a large amount of emulation experiment, which demonstrates ELM,.
ELM algorithms there is also some defects, randomly generate input weight and biasing way brought to ELM it is certain
Randomness, different initiation parameters can cause different learning effects.And due to the precision of least-squares algorithm in itself not
Height, so as to affect the stability of ELM models and generalization ability.One of method for making up this defect is to utilize to combine nerve net
Network thought, but this network structure can be network structure redundancy, while increase computation complexity.Utilize genetic algorithm or intersection
It is also to improve a behave of ELM stability, but these algorithms need largely that proof method, which obtains optimal concealed nodes parameter,
Iterative calculation, causes learning efficiency not high.
Kalman filtering algorithm is a kind of recursive form of linear minimum-variance estimation.In random estimation theory, linearly
Minimum variance estimate is the most the superior in all Linear Estimations.The present invention is based on Kalman filtering algorithms, and traditional ELM is solved
The method that algorithm replaces with filtering recurrence estimation, the output that all training samples only need a filtering iteration and can acquire ELM are weighed
Value improves the stability of ELM models.On this basis, the present invention proposes a kind of aviation based on ELM filtering training algorithms
Engine air passage fault recognition method improves the accuracy of identification of engine different operating point failure pattern in flight envelope.
Invention content
For above-mentioned technical problem, the present invention provides gas circuit in a kind of aero-engine envelope curve based on ELM filtering algorithms
Fault recognition method for the unstability for the learning effect that the randomness of ELM network topology parameters is brought, is filtered using Kalman
Wave algorithm replaces traditional least square method for solving, is effectively improved the stability of ELM models.Exist for aero-engine
The fault mode of different operating points devises the recognition methods based on ELM filtering training algorithms in envelope curve, effectively increases hair
The accuracy of identification of motivation operating point fault mode different in flight envelope.
Technical solution:To achieve the above object, the technical solution adopted by the present invention is:
Gas path failure recognition methods in a kind of aero-engine envelope curve based on ELM filtering algorithms, includes the following steps:
Step 1) in the fault mode data of flight envelope internal boundary points, is instructed according to engine using Kalman filtering algorithm
Practice ELM models;
Step 2) knows fault mode of the trained ELM models of filtering algorithm for other common operating points in envelope curve
Not.
Further, in the step 1) according to engine flight envelope internal boundary points fault mode data, use
Kalman filtering algorithm training ELM models are as follows:
Fault data of the engine mockup in boundary point is standardized by step 1.1), obtains training sample, institute
Fault mode data are stated to be made of each sensor measures parameters;
ELM mode input layer weights and hidden layer are biased random initializtion by step 1.2), and hidden layer activation primitive is chosen
Linear correction function ReLU, substitutes into step 1.1 gained training sample, calculates hidden layer output vector;
Step 1.3) is using Kalman filtering algorithm iteration update ELM model output weight vectors.
Further, using Kalman filtering algorithm iteration update ELM model output weight vectors in the step 1.3)
It is as follows:
Step 1.3.1), output weight vector β random initializtions are random for uniform or normal distribution the small value of zero-mean
Variable;
Step 1.3.2), output vector varivance matrix initial method is:
C=ξ I
Wherein, I is unit matrix, and C is output weight vector varivance matrix, and ξ is the decimal of an artificial settings,
Ranging from [0.001,0.1];
Step 1.3.3) substitute into training sample { xk,yk, wherein x is input data, and y is desired output.According to ELM models
Discrete state equations, calculate k-th of sample and correspond to the one-step prediction value of weight vectorWith output node layer outputIts
In, ELM model discrete state equations are written as:
Wherein, hkFor hidden layer node output vector, ωTAnd υTIt is the process noise and observation noise vector being artificially introduced,
Both for white noise;
Step 1.3.4) utilize Kalman filtering algorithm principle, iteration update output weight vector.
Further, step 1.3.4) described in using Kalman filtering algorithm principle, iteration update output weight vector
It is as follows:
Step 1.3.4.1) calculate filtering gain matrix Gk, specific formula is:
Gk=Ck-1hk·[R+hk TCk-1hk]-1
Wherein, Ck-1For the corresponding varivance matrix estimated value of -1 sample of kth, R is the symmetrical of corresponding observation noise
Positive Definite Matrix;
Step 1.3.4.2) update the corresponding weight vector estimated value of k-th of sample and varivance matrix estimated value, tool
Body formula is:
Wherein,It is the corresponding weight vector estimated value of -1 sample of kth,It is the corresponding weight vector of k-th of sample
Estimated value;Ck-1It is the corresponding varivance matrix estimated value of -1 sample of kth, CkIt is the corresponding error variance square of k-th of sample
Battle array estimated value.
Step 1.3.4.3) iteration updates until all sample trainings are completed.
Further, the trained ELM models of filtering algorithm are used for other common work in envelope curve described in step 2)
The Fault Pattern Recognition of point is as follows:
Step 2.1), by engine, the fault data of other common operating points is standardized in envelope curve, is obtained
Test sample, the fault mode data are made of each sensor measures parameters;
Step 2.2), the ELM models that the fault data after standardization is substituted into filtering algorithm training completion carry out engine
Fault Pattern Recognition is tested.
Advantageous effect:1st, gas path failure identification in the aero-engine envelope curve based on ELM filtering algorithms that the present invention designs
Method solves the problems, such as that tradition is not high in flight envelope internal fault diagnostic accuracy based on the method for diagnosing faults of data.It is carried
The KF-ELM algorithms gone out while precision is improved, ensure that the stability and generalization ability of model compared with common ELM.Herein
On the basis of propose the fault recognition method of aero-engine, can be very well to aerial engine air passage failure mould in flight envelope
Formula is identified.
Description of the drawings
Fig. 1 is that the present invention is based on the aerial engine air passage fault recognition method flow charts of ELM filtering training algorithms.
Fig. 2 fanjet gas circuits working sections mark figure.
Fig. 3 is the Selection Strategy of flight envelope internal reference training points and test check post.
Fig. 4 is filtering training algorithm (KF-ELM) and the tradition of extreme learning machine on UCI criteria classification problem data collection
Nicety of grading (Accuracy) comparison schematic diagram of ELM algorithms.
Fig. 5 be on Page data sets, the niceties of grading of two kinds of algorithms with number of training and the number of hidden nodes variation
Trend comparison.
Fig. 6 be on Vehicle data sets, the niceties of grading of two kinds of algorithms with number of training and the number of hidden nodes change
Change trend comparison.
Specific embodiment
The specific embodiment of the present invention is further described below in conjunction with the accompanying drawings.
Gas path failure recognition methods in a kind of aero-engine envelope curve based on ELM filtering algorithms that the present invention illustrates, tool
Body includes the following steps:
Step 1) in the fault mode data of flight envelope internal boundary points, is instructed according to engine using Kalman filtering algorithm
Practice ELM models;
Fault data of the engine mockup in boundary point is standardized by step 1.1), obtains training sample, institute
Fault mode data are stated to be made of each sensor measures parameters;
ELM mode input layer weights and hidden layer are biased random initializtion by step 1.2), and hidden layer activation primitive is chosen
Linear correction function ReLU, substitutes into step 1.1 gained training sample, calculates hidden layer output vector;
Step 1.3) is using Kalman filtering algorithm iteration update ELM model output weight vectors.
Step 1.3.1), output weight vector β random initializtions are random for uniform or normal distribution the small value of zero-mean
Variable;
Step 1.3.2), output vector varivance matrix initial method is:
C=ξ I
Wherein, I is unit matrix, and C is output weight vector varivance matrix, and ξ is the decimal of an artificial settings,
Ranging from [0.001,0.1];
Step 1.3.3) substitute into training sample { xk,yk, wherein x is input data, and y is desired output.According to ELM models
Discrete state equations, calculate k-th of sample and correspond to the one-step prediction value of weight vectorWith output node layer outputIts
In, ELM model discrete state equations are written as:
Wherein, hkFor hidden layer node output vector, ωTAnd υTIt is the process noise and observation noise vector being artificially introduced,
Both for white noise;
Step 1.3.4) utilize Kalman filtering algorithm principle, iteration update output weight vector.
Step 1.3.4.1) calculate filtering gain matrix Gk, specific formula is:
Gk=Ck-1hk·[R+hk TCk-1hk]-1
Wherein, Ck-1For the corresponding varivance matrix estimated value of -1 sample of kth, R is the symmetrical of corresponding observation noise
Positive Definite Matrix;
Step 1.3.4.2) update the corresponding weight vector estimated value of k-th of sample and varivance matrix estimated value, tool
Body formula is:
Wherein,It is the corresponding weight vector estimated value of -1 sample of kth,The corresponding weights of k-th of sample to
Measure estimated value;Ck-1It is the corresponding varivance matrix estimated value of -1 sample of kth, CkIt is the corresponding error variance of k-th of sample
Matrix Estimation value.
Step 1.3.4.3) iteration updates until all sample trainings are completed.
Step 2) knows fault mode of the trained ELM models of filtering algorithm for other common operating points in envelope curve
Not.
Step 2.1), by engine, the fault data of other common operating points is standardized in envelope curve, is obtained
Test data, the fault mode data are made of each sensor measures parameters;
Step 2.2), the ELM models that the fault data after standardization is substituted into filtering algorithm training completion carry out engine
Fault Pattern Recognition is tested.
In order to verify the Fault Diagnosis of Aircraft Engine Gas Path algorithm based on ELM filtering training algorithms designed by the present invention
Validity, the Digital Simulation that gas path failure in envelope curve identifies has been carried out under MATLAB environment.
First, ELM proposed by the present invention filtering training algorithms are verified on UCI standard data sets.To show this
The validity and novelty of algorithm compare itself and common ELM algorithms and BP neural network.Data illustrate and simulation result
As shown in table 1, table 2 and Fig. 4.Fig. 5 and Fig. 6 is on Page and Vehicle data sets, changes number of training and hidden layer section
Points verify the stability of two kinds of algorithms.As can be seen that ELM filtering training algorithms (KF-ELM) are better than commonly in precision
ELM.And compared with BPNN, the KF-ELM training times significantly reduce.This illustrates KF-ELM algorithms proposed by the invention in essence
It is more advantageous than common ELM on degree, and be significantly improved on the training time compared with BPNN.
Table 1UCI criteria classification problem data collection explanations
Table 2ELM and KF-ELM are in classification problem compared with training time and testing time
For the Fault Pattern Recognition problem of aero-engine different operating point in envelope curve, selected in flight envelope first
8 are taken with reference to training points, the Selection Strategy of training points and test point and is illustrated as shown in Fig. 3 and table 4.As can be seen that instruction
Practice point to be all located on the boundary of flight envelope, test point is some of engine in envelope curve typically common operating point.Institute
The operating point of choosing is injected 13 kinds of fault modes (as shown in table 3) of fanjet, is obtained for trained data sample, should
Sample is made of each sensor measures parameters, main pressure including low pressure rotating speed NL, high pressure rotating speed NH and each section and
The measured value (T22, P22, T3, P3, T43, P43, T6, P5) of temperature.The fanjet that example involved in the present invention is applied
The working sections of gas circuit are as shown in Figure 2.After training sample is obtained, ELM models are trained using Kalman filtering algorithm.Model
After training is completed, the fault mode data of selected 8 training points and 4 test points are substituted into trained model respectively
Fault Pattern Recognition is carried out, for recognition result as shown in table 5, table 6, table 7 and table 8, table 9 and table 10 are two kinds of algorithms in each work
The average accuracy of identification comparison of point.
3 fanjet fault mode explanation of table
The explanation of training points and test point in 4 flight envelope of table
Table 5ELM is in the Fault Pattern Recognition result (correct number/total sample number) of training points
Table 6KF-ELM is in the Fault Pattern Recognition result (correct number/total sample number) of training points
Table 7ELM is in the Fault Pattern Recognition result (correct number/total sample number) of test point
Table 8KF-ELM is in the Fault Pattern Recognition result (correct number/total sample number) of test point
Table 9ELM and KF-ELM is compared in the average accuracy of identification of training points
Table 10ELM and KF-ELM is compared in the average accuracy of identification of test point
KF-ELM methods proposed by the invention put down the fault mode of each operating point it can be seen from table 9 and table 10
The more common ELM of equal accuracy of identification, which is compared, to increase.KF-ELM algorithms are 91.12% in the average accuracy of identification of 8 training points,
The average accuracy of identification of common ELM algorithms is 88.25%;In 4 test points, the average accuracy of identification of KF-ELM is 86.82%,
The average accuracy of identification of common ELM is 83.25%.This explanation, the present invention propose that KF-ELM algorithms have compared to common ELM algorithms
Advantage and generalization ability is more preferable can more meet the requirement of Fault Pattern Recognition work of the engine in flight envelope.
Gas path failure recognition methods in the aero-engine envelope curve based on ELM filtering algorithms that the present invention designs, solves
Tradition is based on the method for diagnosing faults of data flight envelope internal fault diagnostic accuracy is not high the problem of.The KF-ELM proposed is calculated
Method while precision is improved, ensure that the stability and generalization ability of model compared with common ELM.It is proposed on this basis
The fault recognition method of aero-engine can very well be identified aerial engine air passage fault mode in flight envelope.
It should be pointed out that the above description is merely a specific embodiment, but protection scope of the present invention is not
Be confined to this, any one skilled in the art in the technical scope disclosed by the present invention, the change that can be readily occurred in
Change and replace, should all cover within the scope of the present invention.Therefore, protection scope of the present invention should be with the claim
Subject to protection domain.
Claims (5)
1. a kind of gas path failure recognition methods in aero-engine envelope curve based on ELM filtering algorithms, it is characterised in that:Including with
Lower step:
Step 1) in the fault mode data of flight envelope internal boundary points, is instructed according to engine mockup using Kalman filtering algorithm
Practice ELM models;
Step 2) is by the trained ELM models of filtering algorithm for the Fault Pattern Recognition of other operating points in flight envelope.
2. gas path failure identification side in a kind of aero-engine envelope curve based on ELM filtering algorithms according to claim 1
Method, it is characterised in that:Using the fault mode data of the boundary point in flight envelope in the step 1), using Kalman filtering
Algorithm training ELM models are as follows:
Fault mode data of the engine mockup in envelope curve internal boundary points are standardized by step 1.1), obtain training sample
This, the fault mode data are made of the measurement parameter of each sensor;
ELM mode input layer weights and hidden layer are biased random initializtion by step 1.2), and hidden layer activation primitive is chosen linear
Correction function ReLU substitutes into step 1.1 gained training sample, calculates hidden layer output vector;
Step 1.3) is using Kalman filtering algorithm iteration update ELM model output weight vectors.
3. gas path failure identification side in a kind of aero-engine envelope curve based on ELM filtering algorithms according to claim 2
Method, it is characterised in that:It is specific using Kalman filtering algorithm iteration update ELM model output weight vectors in the step 1.3)
Step is as follows:
Step 1.3.1), output weight vector β random initializtions are uniform or normal distribution the small value stochastic variable of zero-mean;
Step 1.3.2), output vector varivance matrix initial method is:
C=ξ I
Wherein, I is unit matrix, and C is output weight vector varivance matrix, the decimal that ξ is manually set for one, range
For [0.001,0.1];
Step 1.3.3) substitute into training sample { xk,yk, wherein x is input data, and y is desired output.According to ELM models from
State equation is dissipated, calculates the one-step prediction value that k-th of sample corresponds to weight vectorWith output node layer outputWherein,
ELM model discrete state equations are written as:
Wherein, hkFor hidden layer node output vector, ωTAnd υTIt is the process noise that is artificially introduced and observation noise vector, the two
It is white noise;
Step 1.3.4) utilize Kalman filtering algorithm principle, iteration update output weight vector.
4. gas path failure identification side in a kind of aero-engine envelope curve based on ELM filtering algorithms according to claim 3
Method, it is characterised in that:Step 1.3.4) described in using Kalman filtering algorithm principle, iteration update output weight vector is specific
Step is as follows:
Step 1.3.4.1) calculate filtering gain matrix Gk, specific formula is:
Gk=Ck-1hk·[R+hk TCk-1hk]-1
Wherein, Ck-1For the corresponding varivance matrix estimated value of -1 sample of kth, R is the symmetric positive definite of corresponding observation noise
Variance matrix;
Step 1.3.4.2) the corresponding weight vector estimated value of k-th of sample and varivance matrix estimated value are updated, it is specific public
Formula is:
Wherein,It is the corresponding weight vector estimated value of -1 sample of kth,It is the corresponding weight vector estimation of k-th of sample
Value;Ck-1It is the corresponding varivance matrix estimated value of -1 sample of kth, CkIt is that the corresponding varivance matrix of k-th of sample is estimated
Evaluation;
Step 1.3.4.3) iteration updates until all sample trainings are completed.
5. gas path failure identification side in a kind of aero-engine envelope curve based on ELM filtering algorithms according to claim 1
Method, it is characterised in that:By the trained ELM models of filtering algorithm for other common operating points in envelope curve described in step 2)
Fault Pattern Recognition is as follows:
Step 2.1), by engine mockup, the fault mode data of other operating points are standardized in envelope curve, are obtained
Test sample, the fault mode data are made of each sensor measures parameters;
Step 2.2), the ELM models that the fault data after standardization is substituted into filtering algorithm training completion carry out engine failure
Pattern-recognition is tested.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810013686.6A CN108205310B (en) | 2018-01-08 | 2018-01-08 | Aeroengine envelope internal gas circuit fault identification method based on ELM filtering algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810013686.6A CN108205310B (en) | 2018-01-08 | 2018-01-08 | Aeroengine envelope internal gas circuit fault identification method based on ELM filtering algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108205310A true CN108205310A (en) | 2018-06-26 |
CN108205310B CN108205310B (en) | 2020-03-17 |
Family
ID=62606108
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810013686.6A Active CN108205310B (en) | 2018-01-08 | 2018-01-08 | Aeroengine envelope internal gas circuit fault identification method based on ELM filtering algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108205310B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109163911A (en) * | 2018-09-21 | 2019-01-08 | 昆明理工大学 | A kind of fault of engine fuel system diagnostic method based on improved bat algorithm optimization ELM |
CN109635318A (en) * | 2018-11-01 | 2019-04-16 | 南京航空航天大学 | A kind of aero-engine sensor intelligent analytic redundancy design method based on KEOS-ELM algorithm |
CN110084324A (en) * | 2019-05-10 | 2019-08-02 | 杭州电子科技大学 | Kalman filtering parameter adaptive update method based on extreme learning machine |
CN110672328A (en) * | 2019-11-05 | 2020-01-10 | 大连理工大学 | Turbofan engine health parameter estimation method based on random configuration network |
CN112364446A (en) * | 2020-09-30 | 2021-02-12 | 南京航空航天大学 | Engine whole performance attenuation prediction method based on EC-RBELM algorithm |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103983453A (en) * | 2014-05-08 | 2014-08-13 | 南京航空航天大学 | Differentiating method of fault diagnosis of executing mechanism of aeroengine and sensor |
CN104008432A (en) * | 2014-06-03 | 2014-08-27 | 华北电力大学 | Micro-grid short-term load forecasting method based on EMD-KELM-EKF |
CN104200062A (en) * | 2014-08-04 | 2014-12-10 | 南京航空航天大学 | Aircraft engine gas path fault fusion diagnosis method |
CN107045575A (en) * | 2017-04-14 | 2017-08-15 | 南京航空航天大学 | Aero-engine performance model modelling approach based on self-adjusting Wiener model |
-
2018
- 2018-01-08 CN CN201810013686.6A patent/CN108205310B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103983453A (en) * | 2014-05-08 | 2014-08-13 | 南京航空航天大学 | Differentiating method of fault diagnosis of executing mechanism of aeroengine and sensor |
CN104008432A (en) * | 2014-06-03 | 2014-08-27 | 华北电力大学 | Micro-grid short-term load forecasting method based on EMD-KELM-EKF |
CN104200062A (en) * | 2014-08-04 | 2014-12-10 | 南京航空航天大学 | Aircraft engine gas path fault fusion diagnosis method |
CN107045575A (en) * | 2017-04-14 | 2017-08-15 | 南京航空航天大学 | Aero-engine performance model modelling approach based on self-adjusting Wiener model |
Non-Patent Citations (4)
Title |
---|
尤成新等: "航空发动机传感器信号重构的K-ELM方法", 《航空动力学报》 * |
许梦阳等: "用于航空发动机动态辨识的MSMEA-ELM算法", 《传感器与微系统》 * |
陶金伟等: "航空发动机传感器故障隔离与气路健康估计方法", 《自动化应用》 * |
顾嘉辉等: "航空发动机健康估计的神经网络修正卡尔曼滤波算法", 《推进技术》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109163911A (en) * | 2018-09-21 | 2019-01-08 | 昆明理工大学 | A kind of fault of engine fuel system diagnostic method based on improved bat algorithm optimization ELM |
CN109635318A (en) * | 2018-11-01 | 2019-04-16 | 南京航空航天大学 | A kind of aero-engine sensor intelligent analytic redundancy design method based on KEOS-ELM algorithm |
CN109635318B (en) * | 2018-11-01 | 2023-07-25 | 南京航空航天大学 | Intelligent analysis redundancy design method for aero-engine sensor |
CN110084324A (en) * | 2019-05-10 | 2019-08-02 | 杭州电子科技大学 | Kalman filtering parameter adaptive update method based on extreme learning machine |
CN110084324B (en) * | 2019-05-10 | 2021-05-04 | 杭州电子科技大学 | Kalman filtering parameter self-adaptive updating method based on extreme learning machine |
CN110672328A (en) * | 2019-11-05 | 2020-01-10 | 大连理工大学 | Turbofan engine health parameter estimation method based on random configuration network |
CN110672328B (en) * | 2019-11-05 | 2020-08-14 | 大连理工大学 | Turbofan engine health parameter estimation method based on random configuration network |
CN112364446A (en) * | 2020-09-30 | 2021-02-12 | 南京航空航天大学 | Engine whole performance attenuation prediction method based on EC-RBELM algorithm |
CN112364446B (en) * | 2020-09-30 | 2023-05-12 | 南京航空航天大学 | Engine whole-engine performance attenuation prediction method based on EC-RBELM algorithm |
Also Published As
Publication number | Publication date |
---|---|
CN108205310B (en) | 2020-03-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108205310A (en) | Gas path failure recognition methods in a kind of aero-engine envelope curve based on ELM filtering algorithms | |
CN108416086B (en) | Aero-engine model self-adaptive correction method based on deep learning | |
WO2019144337A1 (en) | Deep-learning algorithm-based self-adaptive correction method for full-envelope model of aero-engine | |
CN105911975B (en) | A kind of aerial engine air passage failure fusion diagnosis method | |
CN108829928B (en) | Turboshaft engine adaptive component-level simulation model construction method | |
CN107045575B (en) | Aero-engine performance model modeling method based on self-adjusting wiener model | |
CN110795780B (en) | XGboost algorithm-based cable-stayed bridge finite element correction method | |
CN110348615B (en) | Cable line fault probability prediction method based on ant colony optimization support vector machine | |
CN109162813A (en) | One kind being based on the modified Aeroengine Smart method for controlling number of revolution of iterative learning | |
CN109800449B (en) | Neural network-based aeroengine compression component characteristic correction method | |
CN109635318A (en) | A kind of aero-engine sensor intelligent analytic redundancy design method based on KEOS-ELM algorithm | |
CN106503746A (en) | A kind of Fault Diagnosis of Aeroengines method based on offset of performance amount | |
CN112883497B (en) | Space valve reliability assessment method based on multi-source information fusion | |
CN111257934A (en) | Seismic oscillation peak acceleration prediction method based on second-order neuron deep neural network | |
CN111581763B (en) | Method for evaluating air path fault diagnosis result of aero-engine | |
CN108897309B (en) | Aero-Engine Sensor Failure diagnosis and partition method based on fuzzy membership | |
CN116106021A (en) | Precision improving method for digital twin of aeroengine performance | |
CN110991024A (en) | Method for monitoring sudden change of gas circuit component under concurrent fault of aircraft engine control system | |
CN110705114B (en) | Ventilation fault diagnosis method without training sample | |
CN112364446B (en) | Engine whole-engine performance attenuation prediction method based on EC-RBELM algorithm | |
CN106301610A (en) | The adaptive failure detection of a kind of superhet and diagnostic method and device | |
CN102788955B (en) | Remaining lifetime prediction method of ESN (echo state network) turbine generator classification submodel based on Kalman filtering | |
CN110674893B (en) | Self-adaptive correction method for diagnosis experience in rotary machine fault diagnosis knowledge base | |
DE102017110795A1 (en) | Method for improved calibration of the control of an internal combustion engine | |
CN115526308A (en) | Time sequence prediction method based on width pruning echo state network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |