CN114954963A - Hypersonic aeroengine air inlet channel pneumatic instability early warning method - Google Patents
Hypersonic aeroengine air inlet channel pneumatic instability early warning method Download PDFInfo
- Publication number
- CN114954963A CN114954963A CN202210743131.3A CN202210743131A CN114954963A CN 114954963 A CN114954963 A CN 114954963A CN 202210743131 A CN202210743131 A CN 202210743131A CN 114954963 A CN114954963 A CN 114954963A
- Authority
- CN
- China
- Prior art keywords
- early warning
- starting
- air inlet
- boundary
- inlet channel
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000013528 artificial neural network Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims abstract description 12
- 238000011156 evaluation Methods 0.000 claims abstract description 10
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 6
- 238000003062 neural network model Methods 0.000 claims abstract description 3
- 238000010606 normalization Methods 0.000 claims description 9
- 238000002485 combustion reaction Methods 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 3
- 238000007635 classification algorithm Methods 0.000 claims description 2
- 238000001514 detection method Methods 0.000 claims description 2
- 238000012706 support-vector machine Methods 0.000 description 10
- 238000009826 distribution Methods 0.000 description 8
- 238000004088 simulation Methods 0.000 description 5
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000000546 chi-square test Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000012938 design process Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000035939 shock Effects 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B64—AIRCRAFT; AVIATION; COSMONAUTICS
- B64D—EQUIPMENT FOR FITTING IN OR TO AIRCRAFT; FLIGHT SUITS; PARACHUTES; ARRANGEMENT OR MOUNTING OF POWER PLANTS OR PROPULSION TRANSMISSIONS IN AIRCRAFT
- B64D33/00—Arrangements in aircraft of power plant parts or auxiliaries not otherwise provided for
- B64D33/02—Arrangements in aircraft of power plant parts or auxiliaries not otherwise provided for of combustion air intakes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B64—AIRCRAFT; AVIATION; COSMONAUTICS
- B64D—EQUIPMENT FOR FITTING IN OR TO AIRCRAFT; FLIGHT SUITS; PARACHUTES; ARRANGEMENT OR MOUNTING OF POWER PLANTS OR PROPULSION TRANSMISSIONS IN AIRCRAFT
- B64D33/00—Arrangements in aircraft of power plant parts or auxiliaries not otherwise provided for
- B64D33/02—Arrangements in aircraft of power plant parts or auxiliaries not otherwise provided for of combustion air intakes
- B64D2033/0253—Arrangements in aircraft of power plant parts or auxiliaries not otherwise provided for of combustion air intakes specially adapted for particular type of aircraft
- B64D2033/026—Arrangements in aircraft of power plant parts or auxiliaries not otherwise provided for of combustion air intakes specially adapted for particular type of aircraft for supersonic or hypersonic aircraft
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biophysics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Aviation & Aerospace Engineering (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Evolutionary Biology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Measuring Fluid Pressure (AREA)
Abstract
The invention discloses a hypersonic aeroengine air inlet channel pneumatic instability early warning method, after normalizing the characteristics of preprocessed on-way wall surface pressure data, performing two classifications of starting and non-starting state data based on a linear SVM algorithm, determining an optimal classification surface as a non-starting boundary of the current state of the air inlet passage, defining eta% stability margin to determine the non-starting early warning boundary, using the non-starting boundary to realize the air inlet channel instability judgment, on the basis, the training and the evaluation of the back pressure multiple attribute are carried out on the on-way wall surface pressure data based on the BP neural network, the on-way wall surface pressure at the non-starting boundary and the early warning boundary thereof is used as regression input, forecasting the non-starting state of the air inlet channel and the back pressure multiple boundary value of the early warning state of the air inlet channel through a BP neural network model, the method can be used for realizing pneumatic instability early warning of the hypersonic aeroengine within a specific Mach number and flight altitude range. A new feasible scheme is provided for the pneumatic instability early warning of the air inlet passage of the hypersonic aeroengine.
Description
Technical Field
The invention relates to the field of pneumatic instability of a hypersonic air inlet passage, in particular to a pneumatic instability early warning method of a hypersonic aeroengine air inlet passage.
Background
The air inlet channel is used as an important part of the hypersonic aeroengine and plays a vital role in the aspects of ensuring the working efficiency, normal work, thrust and the like of the engine. The problem of aerodynamic instability of the current hypersonic inlet channel is mainly due to factors of overhigh back pressure of a combustion chamber, overlow incoming flow Mach number and overlarge attack angle, and the factors can cause the non-starting of the inlet channel. The starting failure working state of the hypersonic air inlet cannot be solved in the air inlet design process, and the current measures are taken to avoid the starting failure of the air inlet or restart the air inlet. Therefore, a reasonable early warning mechanism is established on the basis of completing instability judgment aiming at the factors of non-starting, and the method has important research significance and engineering application value for ensuring the stable flight of the hypersonic thrust system.
Aiming at the problem of non-starting of the hypersonic air inlet, the excessively high back pressure of the combustion chamber is used as an important factor for triggering pneumatic instability (non-starting) of the hypersonic air inlet, and the hypersonic air inlet is characterized in that a normal shock wave appears at the downstream of the air inlet and can be pushed out of the air inlet under the action of high back pressure excitation, so that the flow of a channel is reduced, and the non-starting of the air inlet is further triggered. From the pneumatic instability angle, based on a certain hypersonic air inlet channel model, numerical simulation is carried out under different back pressure multiples to obtain on-way wall surface pressure data of an air inlet channel, and the prediction of instability early warning boundaries is completed by utilizing a learning algorithm, so that the instability early warning of the hypersonic air inlet channel has the following characteristics in the aspect of the method: in data processing, the wall pressure data volume of the air inlet channel in the pneumatic instability state is small, so that the stable state data and the instability state data are unbalanced in distribution, and a sample has the characteristics of high dimension and small scale; and secondly, from the angle of the back pressure multiple, the back pressure multiple boundary of the air inlet channel pneumatic instability can be defined between the existing starting state data and the existing non-starting state data through a regression method because the high back pressure multiple can form high back pressure to cause the air inlet channel not to start, a certain stability margin is set, and the back pressure multiple boundary of the air inlet channel pneumatic instability early warning is obtained and serves as the early warning condition of the air inlet channel pneumatic instability.
Disclosure of Invention
The invention aims to provide a hypersonic air inlet channel aerodynamic instability early warning method based on machine learning, aiming at the problem of aerodynamic instability of an air inlet channel, and the hypersonic air inlet channel aerodynamic instability early warning method is a novel aerodynamic instability early warning method at present.
The application provides a hypersonic aeroengine air inlet channel pneumatic instability early warning method which is characterized in that after the characteristics of preprocessed on-way wall surface pressure data are normalized, performing two classifications of starting and non-starting state data based on a linear SVM algorithm, determining an optimal classification surface as a non-starting boundary of the current state of the air inlet passage, defining eta% stability margin to determine the non-starting early warning boundary, using the non-starting boundary to realize the air inlet channel instability judgment, on the basis, the training and the evaluation of the back pressure multiple attribute are carried out on the on-way wall surface pressure data based on the BP neural network, the on-way wall surface pressure at the non-starting boundary and the early warning boundary thereof is used as regression input, forecasting the non-starting state of the air inlet channel and the back pressure multiple boundary value of the early warning state of the air inlet channel through a BP neural network model, the method can be used for realizing pneumatic instability early warning of the hypersonic aeroengine within a specific Mach number and flight altitude range.
In some specific embodiments, the method for early warning of aerodynamic instability of an air inlet channel of an aircraft engine is characterized by specifically comprising the following steps:
step 4, calculating the non-starting boundary B of the two-dimensional feature classification of the wall pressure NS And no start warning boundary B NSW ；
Step 5, carrying out back pressure multiple regression training and index evaluation on the wall pressure;
step 6, estimating back pressure multiples of the non-starting state and the early warning state;
and 7, realizing instability early warning conditions.
In some specific embodiments, the method for early warning of aerodynamic instability of an air inlet passage of an aircraft engine is characterized in that the specific implementation method in step 1 is as follows: a pressure signal sampling point is set for a combined air inlet model of a hypersonic aeroengine model, each modal channel is divided into a plurality of on-way wall surfaces, 100 pressure signal acquisition points are set for each on-way wall surface, and actual counter pressure conditions of a combustion chamber are simulated by setting a group of counter pressure multiples at typical working points of the combined engine, so that pressure sample data of the current air inlet starting and non-starting are obtained.
In some specific embodiments, the method for early warning of aerodynamic instability of an air inlet passage of an aircraft engine is characterized in that the specific implementation method in step 4 is as follows: and performing linear two-classification on the starting and non-starting states of the wall pressure by using an SVM classification algorithm, determining an optimal classification surface, namely a non-starting boundary, which can be used as a detection basis for the non-starting of the air inlet channel in the current state, defining an eta% stability margin, and determining the non-starting early warning boundary of the air inlet channel.
In some specific embodiments, the method for early warning of aerodynamic instability of an air inlet passage of an aircraft engine is characterized in that the specific implementation method in step 6 is as follows: and (3) performing regression training and evaluation feedback of the inlet channel on the back pressure multiple by using a BP neural network according to the sample set and the back pressure conditions thereof, then, predicting the inlet channel non-starting back pressure multiple boundary value and the eta% early warning condition back pressure multiple boundary value thereof by using a BPNN regression model based on the determined non-starting boundary in the step (3) and the on-way wall pressure at the early warning boundary thereof as regression input.
In some specific embodiments, the method for early warning of aerodynamic instability of an air inlet passage of an aircraft engine is characterized in that the specific implementation method in step 7 is as follows: according to the back pressure multiple boundary estimation values of the typical working point non-starting state and the early warning state, under the condition of not considering modal conversion, fitting out a rough instability boundary and an early warning boundary curve in the flight Mach number and height range to serve as a condition for early warning of aerodynamic instability of an air inlet channel on a flight track.
Compared with the prior art, the invention has the following advantages and beneficial effects:
according to the method, from the angle of air intake duct pneumatic instability (non-starting) caused by overhigh back pressure of a combustion chamber, a back pressure multiple is set for a typical working point on one flight envelope of the combined cycle engine so as to simulate the back pressure condition of the hypersonic air intake duct, and the on-way wall pressure data of the duct is obtained through computer simulation. And performing two classifications of the starting state data and the non-starting state data by using a Support Vector Machine (SVM) algorithm, wherein the classification accuracy reaches 100.00 percent, obtaining the non-starting boundary of each working point, finishing the instability judgment of the non-starting state data, defining eta percent stability margin, and determining the non-starting early warning boundary of each working point. On the basis, the on-way wall pressure is subjected to regression of the back pressure multiple value, and training and evaluation are carried out by using a BP neural network. And predicting the backpressure multiple boundary value of the air inlet channel non-starting state and the early warning state by taking the pressure data of the non-starting boundary and the early warning boundary as input through the regression model with the well adjusted learning rate alpha. And finally, fitting the instability boundary and early warning boundary value predicted by the working point under each mode (turbine, ejector rocket and stamping) without considering mode conversion to obtain the non-starting boundary and the early warning boundary of the air inlet channel aerodynamic instability within a specific Mach number and flight height range, and taking the non-starting boundary and the early warning boundary as the air inlet channel aerodynamic instability early warning condition.
Compared with the research of other traditional air inlet pneumatic instability early warning problems, the method considers the situation that the air inlet is not started due to overhigh back pressure of a combustion chamber of a thrust system, and completes instability boundary judgment and early warning scheme realization on the on-way wall surface pressure data of an air inlet channel and the corresponding starting state and back pressure multiple conditions by using a machine learning method.
The invention provides the distribution of the aerodynamic instability boundary and the early warning boundary of the hypersonic aeroengine in a flight track. Aiming at turbine, ejector rocket and stamping modes, a thrust system aerodynamic instability and early warning boundary curve is fitted in the flight track (without considering mode conversion stage) by predicting back pressure multiple instability boundary and early warning boundary value of typical working point under each mode, and the curve is used for early warning the aerodynamic instability of the air inlet channel on the flight track.
Drawings
FIG. 1 is a flowchart of an early warning method for aerodynamic instability of an air intake duct according to an embodiment of the present invention;
FIG. 2 shows the distribution of wall pressure along the way at different channel positions of a certain operating point in the embodiment of the present invention;
FIG. 3 is a SVM two classification of the inactive and active states at an operating point according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of the BP neural network structure used;
FIG. 5 is a schematic diagram of a boundary of instability of an inlet port and its early warning boundary curve distribution in a flight path.
Detailed Description
In order to make the technical problems, technical solutions and advantageous effects to be solved by the present invention clearer and clearer, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The embodiment is a pneumatic instability early warning method for an air inlet passage of a hypersonic aeroengine. Pressure data under the condition that the low back pressure multiple is set to the high back pressure multiple along the wall surface of the air inlet channel is collected, and after preprocessing and data normalization, two pressure characteristics are extracted to complete SVM two classification of wall surface pressure starting and non-starting states. Meanwhile, a BP neural network is used for completing back pressure multiple regression of wall pressure, the wall pressure at the SVM non-starting boundary is used as input regression to obtain a non-starting state back pressure multiple boundary value, eta% stability margin is set, and a non-starting state early warning back pressure multiple boundary value is obtained. And finally, in the flight Mach number and height range, synthesizing an air inlet channel aerodynamic instability boundary curve and an early warning curve on the flight track, and performing early warning processing on the air inlet channel aerodynamic instability on the flight track.
The method of carrying out the present invention is further described below with reference to FIG. 1.
1-1, firstly, setting pressure signal sampling points for a combined air inlet model of the hypersonic aeroengine, dividing each modal channel into a plurality of on-way wall surfaces, and setting 100 pressure signal acquisition points on each on-way wall surface. On a typical working point of the combined engine, a group of back pressure multiples are set to simulate the actual back pressure condition of the combustion chamber, and then pressure sample data of the starting and the non-starting of the current air inlet channel are obtained (a plurality of wall pressure samples with 100-dimensional characteristics can be obtained through simulation under a single back pressure multiple condition).
And 1-2, processing abnormal values and missing values of the pressure sample data acquired by numerical simulation, and merging and sorting the pressure sample data. Fig. 2 shows wall pressure distributions along the way at different channel positions (characteristics) of a certain operating point, each pressure distribution including 100 data points, wherein a red dotted line represents wall pressure data in a non-activated state obtained by simulation under the condition of non-activated backpressure multiple, and a blue solid line represents wall pressure data in an activated state obtained by simulation under the condition of activated backpressure multiple.
And 2, normalizing the input samples.
For an input sample T { (x) k ,y k )|x k ∈R n ,y k ∈R n K is 1,2, …, n, and x is written k (i) (i-1, 2, …, m) is the kth sample data corresponding to the ith feature, where y is k A label indicating a kth sample, m indicating the number of features of the input sample (100 dimensions), and n indicating the number of input samples. The following formula is utilized:
input samples are processed (0,1)And (4) feature normalization, so that each feature sample value is in the same order of magnitude. Wherein x k (i) (max) And x k (i) (min) Respectively representing the maximum and minimum values, x, of sample data corresponding to the ith feature k (i) ' denotes sample data after normalization of the ith feature.
And 3, extracting two-dimensional characteristics of the wall pressure.
For normalized input samples T { (x) k ',y k )|x k '∈R n ,y k ∈R n And k is 1,2, …, n, and the two-dimensional feature { (x) of the wall pressure is selected by using a Chi-square test feature selection algorithm (p) ,x (q) )|p,q＝1,2,…,100,p≠q}。
Step 4, calculating the non-starting boundary B of the two-dimensional feature classification of the wall pressure NS And no start warning boundary B NSW 。
4-1, aiming at the extracted two-dimensional characteristics of the wall pressure, randomly dividing the input samples into 80% training sets and 20% testing sets, completing the linear binary classification of the SVM which starts and stops at each working point, and obtaining an optimal classification surface:
x (p) and x (q) Two characteristic variables representing wall pressure, where ω ═ ω 0 ,ω 1 ]And B is a hyperplane parameter that is linearly separable by the SVM, so that boundary B is not activated NS The equation can be expressed as ω 0 x (p) +ω 1 x (q) And + b ═ 0. FIG. 3 shows SVM linear dichotomy (to 100% classification accuracy) for in-flight wall pressure misfire and misfire behavior at a certain operating point.
4-2, based on the step 4-1, defining eta% stability margin, and determining an early warning boundary B with eta% stability margin left on the non-starting boundary NSW The equation can be expressed as ω 0 x (p) +ω 1 x (q) +b w 0, wherein b w ＝(1-η％)·b。
And 5, carrying out back pressure multiple regression training and index evaluation on the wall pressure.
5-1, aiming at the wall surface pressure data, taking the given back pressure multiple as a data label, performing (0,1) characteristic normalization on the input sample in the mode of the step 2, then performing normalization on the data label to obtain a normalized regression input sample T r '＝{(x k ',y (r)k ')|x k '∈R n ,y (r)k ∈R n ,k＝1,2,…,n}。
5-2, extracting two-dimensional characteristics { x ] from the wall pressure characteristics in the step 3 (p) ,x (q) And f, the wall surface pressure sample data corresponding to the two-dimensional characteristic is T r (p,q) '＝{(x k (i) ',y (r)k ')|x k (i) '∈R n×2 ,y (r)k ∈R n I ═ p, q, k ═ 1,2, …, n, which was divided into an 80% regression training set and a 20% regression test set, and trained and evaluated using the BP neural network.
5-3, FIG. 4 shows the mesh structure of the BP neural network, with inputs (x) from a given BP neural network 1 ,x 2 )＝{x k (p) ',x k (q) ' | k ═ 1,2, …, n }, and output y ═ y (r)k ') | k ═ 1,2, …, n }. Setting 2 layers of hidden layer, the weight and bias coefficient of the connection between the input layer and the hidden layer 1 as { (w) i,1 [1] ,w i,2 [1] ,b i [1] ) I { (1, 2, …,10}, and the weight and bias coefficient for the connection between hidden layer 1 and hidden layer 2 are { (w) i,1 [2] ,w i,2 [2] ,…,w i,10 [2] ,b i [2] ) 1,2, …,20, | i ═ using ReLU as an activation function; the weight and bias coefficient of the hidden layer 2 connected with the output layer are { (w) 1,1 [3] ,w 1,2 [3] ,…,w 1,20 [3] ,b 1 [3] ) And (6) adopting a linear function.
5-4, using typical working point in one flight envelope of hypersonic combined cycle engine as object, using parameter initializationThe BPNN trains and predicts the wall pressure sample data, and finally performs inverse normalization output on the prediction result and determines a coefficient R 2 And the root mean square error is used as an evaluation index of regression training, the learning rate alpha is further adjusted according to an index result to be improved, and an alpha parameter value suitable for the network is determined.
And 6, estimating back pressure multiples of the non-starting state and the early warning state.
6-1, based on the step 4 and the step 5, taking the wall pressure data at the non-starting boundary as regression input of the prediction of the non-starting back pressure multiple, specifically, the non-starting boundary equation omega 0 x (p) +ω 1 x (q) + b-0 and the least square equation x of the normalized two-dimensional feature space sample distribution (q) ＝p 0 x (p) +p 1 And (4) determining. Can obtain its input as
Obtaining a non-starting back pressure multiple boundary value M through BPNN prediction BP (ω,b)。
6-2 boundary equation omega for early warning by no starting 0 x (p) +ω 1 x (q) +b w The early warning boundary input can be determined as 0 through the least square equation of the normalized two-dimensional characteristic space sample distribution
Wherein b is w Obtaining a non-starting early warning back pressure multiple boundary value M through BPNN prediction BP (ω,b w )。
And 7, realizing instability early warning conditions.
By estimating the back pressure multiple boundary of the non-starting state and the early warning state of a typical working point, under the condition of not considering modal conversion, aiming at the working stage of a single-modal (turbine, ejector rocket and stamping mode) thrust system, a rough instability boundary and an early warning boundary curve are fitted in a specific Mach number and flight height range to serve as the condition of the pneumatic instability early warning of an air inlet channel in the flight Mach number and height range. FIG. 5 shows a hypersonic air intake in a flying horseMultiple boundary curve S of non-starting back pressure in Hertz number and height range 1 And η% early warning boundary curve S 2 Is roughly distributed.
Recording the counter pressure multiple value of the combustion chamber when the hypersonic air inlet channel works at the current state point (Ma, H) as M BP At this time, curve S 1 And curve S 2 Corresponding values are respectively S 1 * And S 2 * Then, the method for implementing the intake duct instability warning can be represented by the following form:
the above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.
While the present invention has been described with reference to the preferred embodiments, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined by the appended claims. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (6)
1. A hypersonic aeroengine air inlet channel pneumatic instability early warning method is characterized in that, after the feature normalization of the preprocessed on-way wall surface pressure data, performing two classifications of starting and non-starting state data based on a linear SVM algorithm, determining an optimal classification surface as a non-starting boundary of the current state of the air inlet channel, defining eta% stability margin to determine the non-starting early warning boundary, using the non-starting boundary to realize the air inlet channel instability judgment, on the basis, the training and the evaluation of the back pressure multiple attribute are carried out on the on-way wall surface pressure data based on the BP neural network, the on-way wall surface pressure at the non-starting boundary and the early warning boundary thereof is used as regression input, forecasting the non-starting state of the air inlet channel and the back pressure multiple boundary value of the early warning state of the air inlet channel through a BP neural network model, the method can be used for realizing pneumatic instability early warning of the hypersonic aeroengine within a specific Mach number and flight altitude range.
2. The aero-engine inlet channel aerodynamic instability early warning method according to claim 1, characterized by comprising the following steps:
step 1, collecting and preprocessing wall surface pressure data;
step 2, input sample normalization processing;
step 3, extracting two-dimensional characteristics of wall surface pressure;
step 4, calculating the non-starting boundary B of the two-dimensional feature classification of the wall pressure NS And no start warning boundary B NSW ；
Step 5, carrying out back pressure multiple regression training and index evaluation on the wall pressure;
step 6, estimating back pressure multiples of the non-starting state and the early warning state;
and 7, realizing instability early warning conditions.
3. The aero-engine inlet channel aerodynamic instability early warning method according to claim 2, characterized in that the specific implementation method of step 1 is as follows: a pressure signal sampling point is set for a combined air inlet model of a hypersonic aeroengine model, each modal channel is divided into a plurality of on-way wall surfaces, 100 pressure signal acquisition points are set for each on-way wall surface, and actual counter pressure conditions of a combustion chamber are simulated by setting a group of counter pressure multiples at typical working points of the combined engine, so that pressure sample data of the current air inlet starting and non-starting are obtained.
4. The aero-engine inlet channel aerodynamic instability early warning method according to claim 2, characterized in that the specific implementation method of step 4 is as follows: and performing linear two-classification on the starting and non-starting states of the wall pressure by using an SVM classification algorithm, determining an optimal classification surface, namely a non-starting boundary, which can be used as a detection basis for the non-starting of the air inlet channel in the current state, defining an eta% stability margin, and determining the non-starting early warning boundary of the air inlet channel.
5. The aero-engine inlet channel aerodynamic instability early warning method according to claim 2, characterized in that the specific implementation method of step 6 is as follows: and (3) performing regression training and evaluation feedback of the on-way wall pressure of the air inlet channel on the backpressure multiple by using a BP (back propagation) neural network according to the sample set and the backpressure condition thereof, taking the on-way wall pressure at the determined non-starting boundary and the early warning boundary thereof as regression input, and predicting the non-starting backpressure multiple boundary value of the air inlet channel and the backpressure multiple boundary value of the eta% early warning condition thereof through a BPNN (band-pass neural network) regression model.
6. The aero-engine inlet channel aerodynamic instability early warning method according to claim 2, characterized in that the specific implementation method of step 7 is as follows: according to the back pressure multiple boundary estimation values of the typical working point non-starting state and the early warning state, under the condition of not considering modal conversion, fitting out a rough instability boundary and an early warning boundary curve in the flight Mach number and height range to serve as a condition for early warning of aerodynamic instability of an air inlet channel on a flight track.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210743131.3A CN114954963A (en) | 2022-06-27 | 2022-06-27 | Hypersonic aeroengine air inlet channel pneumatic instability early warning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210743131.3A CN114954963A (en) | 2022-06-27 | 2022-06-27 | Hypersonic aeroengine air inlet channel pneumatic instability early warning method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114954963A true CN114954963A (en) | 2022-08-30 |
Family
ID=82965096
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210743131.3A Pending CN114954963A (en) | 2022-06-27 | 2022-06-27 | Hypersonic aeroengine air inlet channel pneumatic instability early warning method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114954963A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106446492A (en) * | 2016-05-04 | 2017-02-22 | 北京航空航天大学 | Early warning method for pneumatic stability loss of turbines |
| CN106640722A (en) * | 2017-01-24 | 2017-05-10 | 中国科学院工程热物理研究所 | Gas compressor aerodynamic stability diagnosis and control device and method |
| CN107165850A (en) * | 2017-06-27 | 2017-09-15 | 西北工业大学 | A kind of rotating stall of axial flow compressor method for early warning recognized based on frequency domain hump |
| US20180035606A1 (en) * | 2016-08-05 | 2018-02-08 | Romello Burdoucci | Smart Interactive and Autonomous Robotic Property Maintenance Apparatus, System, and Method |
| CN112131673A (en) * | 2020-09-30 | 2020-12-25 | 西南石油大学 | Engine surge fault prediction system and method based on fusion neural network model |
| CN112651076A (en) * | 2020-11-20 | 2021-04-13 | 南京航空航天大学 | Binary supersonic speed adjustable air inlet duct non-starting boundary prediction method |
-
2022
- 2022-06-27 CN CN202210743131.3A patent/CN114954963A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106446492A (en) * | 2016-05-04 | 2017-02-22 | 北京航空航天大学 | Early warning method for pneumatic stability loss of turbines |
| US20180035606A1 (en) * | 2016-08-05 | 2018-02-08 | Romello Burdoucci | Smart Interactive and Autonomous Robotic Property Maintenance Apparatus, System, and Method |
| CN106640722A (en) * | 2017-01-24 | 2017-05-10 | 中国科学院工程热物理研究所 | Gas compressor aerodynamic stability diagnosis and control device and method |
| CN107165850A (en) * | 2017-06-27 | 2017-09-15 | 西北工业大学 | A kind of rotating stall of axial flow compressor method for early warning recognized based on frequency domain hump |
| CN112131673A (en) * | 2020-09-30 | 2020-12-25 | 西南石油大学 | Engine surge fault prediction system and method based on fusion neural network model |
| CN112651076A (en) * | 2020-11-20 | 2021-04-13 | 南京航空航天大学 | Binary supersonic speed adjustable air inlet duct non-starting boundary prediction method |
Non-Patent Citations (1)
| Title |
|---|
| 王眷卫;屈卫东;: "航空涡轮发动机失稳预警信号提取方法的研究", 微型电脑应用, no. 03, 20 March 2010 (2010-03-20) * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112149316B (en) | Aero-engine residual life prediction method based on improved CNN model | |
| CN108256173B (en) | Gas circuit fault diagnosis method and system for dynamic process of aircraft engine | |
| CN112001128B (en) | Axial flow compressor stall surge prediction method based on deep learning | |
| CN113569338B (en) | Gas compressor rotating stall early warning method based on time expansion convolutional network | |
| WO2024045246A1 (en) | Spike echo state network model for aero engine fault prediction | |
| CN111160457B (en) | Scroll engine fault detection method based on soft-class extreme learning machine | |
| CN110987439B (en) | Aeroengine fault prediction method based on Logitics regression and Xgboost model | |
| CN112446091A (en) | Artificial neural network-based pulsating pressure prediction method | |
| CN110096801A (en) | A kind of aero-engine high-pressure turbine disk dynamic reliability prediction technique | |
| CN113283004A (en) | Aero-engine degradation state fault diagnosis method based on transfer learning | |
| CN115859791A (en) | Method and device for estimating residual life of aero-engine | |
| CN110826587A (en) | Improved weighted support vector machine-based turboshaft engine fault detection method | |
| CN112116101B (en) | Aeroengine fault diagnosis method based on group reduction kernel extreme learning machine | |
| CN114954963A (en) | Hypersonic aeroengine air inlet channel pneumatic instability early warning method | |
| Calvert et al. | Design, test and analysis of a high-pressure-ratio transonic fan | |
| Zhang et al. | Flutter prediction in turbomachinery with energy method | |
| CN114136622B (en) | DBN-DTL-based aeroengine gas circuit fault diagnosis method | |
| CN114857062B (en) | Aircraft engine unstability state detecting system based on TCN | |
| CN112733872A (en) | Aeroengine fault detection method based on dynamic radius support vector data description | |
| CN114861313B (en) | Design method of stall prediction model of axial flow compressor | |
| Yu et al. | EGT baseline model of aeroengine based on kernel principal component analysis and deep belief network | |
| CN112149233B (en) | Echo state network-based aeroengine dynamic thrust estimation method | |
| Zhang et al. | Leveraging Ground Test Data for Aero-Engine Thrust Estimation | |
| De Giorgi et al. | Development of a combined Artificial Neural Network and Principal Component Analysis technique for Engine Health Monitoring | |
| Fenglei et al. | Prediction of engine total pressure distortion in improved cascaded forward 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 |