CN115270606B - Flow distribution prediction method for integrated stress application rear frame of aero-engine - Google Patents
Flow distribution prediction method for integrated stress application rear frame of aero-engine Download PDFInfo
- Publication number
- CN115270606B CN115270606B CN202210773973.3A CN202210773973A CN115270606B CN 115270606 B CN115270606 B CN 115270606B CN 202210773973 A CN202210773973 A CN 202210773973A CN 115270606 B CN115270606 B CN 115270606B
- Authority
- CN
- China
- Prior art keywords
- stress application
- flow distribution
- prediction model
- model
- prediction
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000012549 training Methods 0.000 claims abstract description 23
- 238000013528 artificial neural network Methods 0.000 claims abstract description 15
- 238000013461 design Methods 0.000 claims abstract description 11
- 238000012360 testing method Methods 0.000 claims abstract description 9
- 230000008859 change Effects 0.000 claims abstract description 8
- 238000012546 transfer Methods 0.000 claims abstract description 7
- 238000012795 verification Methods 0.000 claims abstract description 5
- 238000002474 experimental method Methods 0.000 claims abstract description 4
- 238000005070 sampling Methods 0.000 claims description 7
- 210000002569 neuron Anatomy 0.000 claims description 6
- 238000003062 neural network model Methods 0.000 claims description 4
- 230000003068 static effect Effects 0.000 claims description 4
- 239000000112 cooling gas Substances 0.000 claims description 3
- 230000007547 defect Effects 0.000 abstract description 3
- 239000007789 gas Substances 0.000 description 5
- 238000001816 cooling Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 239000003381 stabilizer Substances 0.000 description 3
- 238000002485 combustion reaction Methods 0.000 description 2
- 238000002790 cross-validation Methods 0.000 description 2
- 239000000446 fuel Substances 0.000 description 2
- 239000002737 fuel gas Substances 0.000 description 2
- 238000002347 injection Methods 0.000 description 2
- 239000007924 injection Substances 0.000 description 2
- 239000000243 solution Substances 0.000 description 2
- 238000010200 validation analysis Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000003416 augmentation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000013401 experimental design Methods 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005312 nonlinear dynamic Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/08—Fluids
-
- 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)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a method for predicting flow distribution of an integrated stress application rear frame of an aero-engine, and relates to the field of air systems of aero-engines. The method overcomes the defects that modeling of the framework structure is complex after integral stress application, and solving is long in time consumption and difficult to predict. The method comprises the following steps: step 1, determining input parameters and change intervals of a frame flow distribution prediction model after integral stress application; step 2, carrying out Latin hypercube numerical experiment design and solving a three-dimensional flow heat transfer physical model, and providing a training sample and a test sample for the prediction model; step 3, establishing a flow distribution prediction model based on a generalized regression neural network, determining experience coefficients of the model based on a cross verification method, and training the established prediction model; and 4, predicting the frame flow distribution after the integrated stress application by using the prediction model trained in the step 3. The prediction cost is low, the prediction precision is high, and the dynamic learning capability is provided.
Description
Technical Field
The invention relates to the field of aero-engine air systems, in particular to a method for predicting flow distribution of an integrated stress application rear frame of an aero-engine.
Background
The afterburner is an important thrust augmentation device of the jet engine, and the thrust-weight ratio of the engine can be greatly improved and the flight envelope can be enlarged by injecting fuel into the gas flow (or the mixed flow of the fuel and the air of the external culvert) after the turbine for combustion; have found widespread use in military turbojet and turbofan engines. In order to realize stable flame combustion, the traditional afterburner is characterized in that a flame stabilizer and an oil injection pipeline are directly arranged in a main gas flow path, so that the main gas flow is inevitably blocked, the thrust-weight ratio of an engine is reduced, and the oil consumption rate is increased. In order to meet the high-performance requirements of the new generation of aero-engines, western developed countries such as Europe and America propose a concept of an integrated stress application rear frame, namely, a turbine rear support plate, an oil injection rod and a flame stabilizer are integrated, flame stabilization is realized by directly utilizing the blunt body configuration of the tail part of the support plate, and the novel aero-engine has the advantages of low flow resistance, high efficiency and low weight, and is considered as the key development direction of the new generation of aero-power technology.
With the continuous improvement of the performance of the engine, the average afterburning temperature exceeds 2100K, and the temperature resistance limit of the metamaterial is far, so that the thermal protection is a key difficulty for restricting the technical development of the afterburner. The development of efficient cooling means in combination with the use of high temperature resistant materials is considered to be critical in breaking through the technological bottlenecks that restrict afterburners. The establishment of the high-precision low-cost cooling air flow distribution model for the integrated stress application rear frame heat flow environment has important theoretical and practical significance for the design of the efficient cooling structure.
The artificial neural network is a self-adaptive nonlinear dynamic system formed by a large number of processing units through extremely rich and perfect connection, and can autonomously learn mathematical correlation in a data sample to realize the prediction of unknown data. The generalized regression neural network is based on non-parametric kernel regression, uses sample data as a posterior condition, and directly calculates the regression value of the output quantity to the input quantity after obtaining a joint probability density function between the input quantity and the output quantity from an observation sample by executing Parzen non-parametric estimation.
In this way, how to predict the flow distribution of the frame after the integral stress application of the aero-engine by means of the artificial neural network specifically so as to solve the defects that the modeling of the traditional integral stress application frame structure is complex, the solving is long in time consumption and difficult to predict, and the method becomes a technical problem to be solved by the person skilled in the art.
Disclosure of Invention
Aiming at the problems, the invention provides a frame flow distribution prediction method after integrated stress application of an aeroengine, overcomes the defects of complex modeling of a frame structure after integrated stress application, long solution time consumption and difficult prediction, and provides a low-cost and high-precision prediction method.
The technical scheme of the invention is as follows: the method comprises the following steps:
step 1, determining input parameters and change intervals of a frame flow distribution prediction model after integral stress application;
step 2, carrying out Latin hypercube numerical experiment design and solving a three-dimensional flow heat transfer physical model, and providing a training sample and a test sample for the prediction model;
step 3, establishing a flow distribution prediction model based on a generalized regression neural network, determining experience coefficients of the model based on a cross verification method, and training the established prediction model;
and 4, predicting the frame flow distribution after the integrated stress application by using the prediction model trained in the step 3.
Further, the input parameters of the prediction model comprise boundary conditions of the main flow and the outer culvert, and specifically comprise main flow inlet total temperature, main flow inlet total pressure, outer culvert total temperature and main flow outlet static pressure. The output parameters of the prediction model comprise the cooling gas output quantity of each part of the frame after the integral stress application.
Further, training samples in the step 2 are obtained through Latin hypercube sampling design and three-dimensional flow heat transfer physical model solving, the number is 12 XN, N is the number of model input parameters, test samples are generated through a design interval random sampling method, and the number is 1/4 of training samples.
Further, the generalized regression neural network model in the step 3 comprises an input layer, a mode layer, a summation layer and an output layer; the output of neurons in the pattern layer is:
wherein X is the input quantity of the network, X i For the learning sample corresponding to the ith neuron, the output of the summation layer is:
wherein y is i The connection weight of the ith element in the output sample; the output of the output layer is
In step 3, the cross validation method divides the training samples into k classes, each subset data is used as a primary validation set, the rest k-1 sets of subsets are used as training sets, and the average value of the k model prediction precision is used as the overall prediction precision.
The beneficial effects of the invention are as follows:
1) The method has the advantages of low prediction cost: in the prediction process, an integrated stress application back frame structure model and a solution flow heat transfer physical equation are not required to be established, so that the time consumption is short and the cost is low.
2) The method has high prediction precision: the generalized regression neural network has strong learning capability, can deeply mine the inherent numerical correlation between data, and has higher calculation accuracy than the traditional empirical correlation.
3) The method has the dynamic learning capability: training samples can be further enriched along with continuous accumulation of later data, and the prediction precision of the generalized regression neural network is improved, which is not possessed by the traditional physical model solving method.
Drawings
FIG. 1 is a schematic view of the structure of the integrated stress application rear frame;
fig. 2 is a topological structure diagram of the generalized regression neural network in this case.
Detailed Description
In order to clearly illustrate the technical features of the present patent, the following detailed description will make reference to the accompanying drawings.
The invention, as shown in fig. 1-2, proceeds as follows:
step 1, determining input parameters of an integrated stress application rear frame (the integrated stress application rear frame comprises an integrated support plate, a center cone, a wall type stabilizer and a vibration-proof heat shield) flow distribution prediction model and a change interval of the input parameters; the input parameters comprise total pressure and total temperature of a fuel gas inlet, static pressure of a fuel gas outlet, total pressure and total temperature of an external culvert inlet and static pressure of an external culvert outlet; the output parameters of the prediction model comprise cooling gas output quantities of all parts of the integrated stress application rear frame, including a support plate, a center cone and a heat shield. Wherein the gas temperature change interval is +/-5%, the gas pressure change interval is +/-8%, the culvert pressure change interval is +/-6%, and the culvert temperature change interval is +/-8%;
step 2, latin hypercube numerical experiment design and three-dimensional flow heat transfer physical model solving are carried out, and a training sample and a testing sample are provided for the prediction model; i.e. sampling by Latin hypercube method, the experimental design is carried out, and the references are as follows: afzal A, kim K Y, seo J.effects of Latin hypercube sampling on surrogate modeling and optimization [ J ]. International Journal of Fluid Machinery and Systems,2017,10 (3): 240-253;
training sample design is carried out based on a Latin hypercube method, the sample capacity is 12 XN (N is the number of model input parameters), test samples are generated by a design interval random sampling method, and the number of training samples is 1/4, as shown in table 1;
TABLE 1
Step 3, establishing an integrated back frame flow distribution prediction generalized regression neural network model, training a neural network based on a training sample, wherein the specific training is carried out by adopting a Matlab neural network tool box newgrnn command;
the generalized regression neural network model in the step 3 comprises an input layer, a mode layer, a summation layer and an output layer; the output of neurons in the pattern layer is:
wherein X is the input quantity of the network, X i For the learning sample corresponding to the ith neuron, the output of the summation layer is:
wherein y is i The connection weight of the ith element in the output sample; the output of the output layer is:
in step 3, the cross-validation method divides the training samples into k classes, each subset data is used as a primary validation set, the remaining k-1 sets of subsets are used as training sets, and the average value of the k model prediction accuracy is used as the overall prediction accuracy.
And 4, detecting the generalization capability of the generalized regression neural network prediction model based on the test sample, and realizing the generalized regression neural network prediction model through a matlab neural network tool kit sim command.
Through tests, the prediction error of the established generalized regression neural network prediction model for the cold air flow distribution of the support plate is 3.2%, the prediction error for the cold air flow distribution of the central cone is 2.8%, and the prediction error for the cold air flow distribution of the heat shield is 4.5%.
While there have been described what are believed to be the preferred embodiments of the present invention, it will be apparent to those skilled in the art that many more modifications are possible without departing from the principles of the invention.
Claims (3)
1. The method for predicting the flow distribution of the integrated stress application rear frame of the aero-engine is characterized by comprising the following steps of:
step 1, determining input parameters and change intervals of a frame flow distribution prediction model after integral stress application;
step 2, carrying out Latin hypercube numerical experiment design and solving a three-dimensional flow heat transfer physical model, and providing a training sample and a test sample for the prediction model;
step 3, establishing a flow distribution prediction model based on a generalized regression neural network, determining experience coefficients of the model based on a cross verification method, and training the established prediction model;
the generalized regression neural network model in the step 3 comprises an input layer, a mode layer, a summation layer and an output layer; the output of neurons in the pattern layer is:
wherein X is the input quantity of the network, X i For the learning sample corresponding to the ith neuron, the output of the summation layer is:
wherein y is i The connection weight of the ith element in the output sample; the output of the output layer is
In the step 3, the cross verification method divides the training sample into k classes, each subset data is used as a primary verification set, the rest k-1 sets of subsets are used as training sets, and the average value of the k model prediction precision is used as the integral prediction precision;
and 4, predicting the frame flow distribution after the integrated stress application by using the prediction model trained in the step 3.
2. The method for predicting flow distribution of integrated stress application frame of aeroengine according to claim 1, wherein the input parameters of the prediction model comprise boundary conditions of main flow and outer duct, specifically comprise main flow inlet total temperature, main flow inlet total pressure, outer culvert total temperature and main flow outlet static pressure, and the output parameters of the prediction model comprise cooling gas output quantity of all parts of the integrated stress application frame.
3. The method for predicting flow distribution of integrated stress application framework of aeroengine as claimed in claim 1, wherein the training samples in the step 2 are obtained through Latin hypercube sampling design and three-dimensional flow heat transfer physical model solution, the number is 12×N, N is the number of model input parameters, the test samples are generated through a design interval random sampling method, and the number is 1/4 of the training samples.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210773973.3A CN115270606B (en) | 2022-07-01 | 2022-07-01 | Flow distribution prediction method for integrated stress application rear frame of aero-engine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210773973.3A CN115270606B (en) | 2022-07-01 | 2022-07-01 | Flow distribution prediction method for integrated stress application rear frame of aero-engine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115270606A CN115270606A (en) | 2022-11-01 |
| CN115270606B true CN115270606B (en) | 2024-02-13 |
Family
ID=83763062
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210773973.3A Active CN115270606B (en) | 2022-07-01 | 2022-07-01 | Flow distribution prediction method for integrated stress application rear frame of aero-engine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115270606B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111079920A (en) * | 2019-11-29 | 2020-04-28 | 南京航空航天大学 | Method for predicting uneven flow coefficient of outlet of turbine gas collecting cavity |
| CN111175054A (en) * | 2020-01-08 | 2020-05-19 | 沈阳航空航天大学 | Aeroengine fault diagnosis method based on data driving |
| CN114154234A (en) * | 2021-11-04 | 2022-03-08 | 中国人民解放军海军航空大学青岛校区 | Modeling method, system and storage medium for aircraft engine |
| WO2022068587A1 (en) * | 2020-09-30 | 2022-04-07 | 西南石油大学 | Fused neural network model-based prediction system and method for engine surge fault |
| CN114519299A (en) * | 2022-01-14 | 2022-05-20 | 南京航空航天大学 | Multi-objective optimization method for double-wall structure of integrated support plate flame stabilizer |
-
2022
- 2022-07-01 CN CN202210773973.3A patent/CN115270606B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111079920A (en) * | 2019-11-29 | 2020-04-28 | 南京航空航天大学 | Method for predicting uneven flow coefficient of outlet of turbine gas collecting cavity |
| CN111175054A (en) * | 2020-01-08 | 2020-05-19 | 沈阳航空航天大学 | Aeroengine fault diagnosis method based on data driving |
| WO2022068587A1 (en) * | 2020-09-30 | 2022-04-07 | 西南石油大学 | Fused neural network model-based prediction system and method for engine surge fault |
| CN114154234A (en) * | 2021-11-04 | 2022-03-08 | 中国人民解放军海军航空大学青岛校区 | Modeling method, system and storage medium for aircraft engine |
| CN114519299A (en) * | 2022-01-14 | 2022-05-20 | 南京航空航天大学 | Multi-objective optimization method for double-wall structure of integrated support plate flame stabilizer |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115270606A (en) | 2022-11-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107045575A (en) | Aero-engine performance model modelling approach based on self-adjusting Wiener model | |
| Zheng et al. | Aero-engine on-board dynamic adaptive MGD neural network model within a large flight envelope | |
| Ma et al. | Mathematical modeling and characteristic analysis for over-under turbine based combined cycle engine | |
| Koleini et al. | EGT prediction of a micro gas turbine using statistical and artificial intelligence approach | |
| CN111967202B (en) | Artificial intelligence-based aircraft engine extreme speed performance digital twinning method | |
| Hendricks et al. | Simultaneous propulsion system and trajectory optimization | |
| CN115099165B (en) | Aeroengine modeling method considering performance degradation | |
| Zhewen et al. | A multi-fidelity simulation method research on front variable area bypass injector of an adaptive cycle engine | |
| CN115270606B (en) | Flow distribution prediction method for integrated stress application rear frame of aero-engine | |
| Cai et al. | A new method to improve the real-time performance of aero-engine component level model | |
| Medic et al. | Integrated RANS/LES computations of turbulent flow through a turbofan jet engine | |
| Jia et al. | A novel performance analysis framework for adaptive cycle engine variable geometry components based on topological sorting with rules | |
| Fuksman et al. | Real-time execution of a high fidelity aero-thermodynamic turbofan engine simulation | |
| Dursun et al. | Modeling of performance and thermodynamic metrics of a conceptual turboprop engine by comparing different machine learning approaches | |
| Mohammed et al. | Prediction of turbojet performance by using artificial neural network | |
| CN114961985A (en) | Intelligent prediction method and system for performance of hydrogen fuel aviation rotor engine | |
| CN114519299A (en) | Multi-objective optimization method for double-wall structure of integrated support plate flame stabilizer | |
| Ma et al. | Matching performance prediction between core driven fan stage and high pressure compressor | |
| Zheng et al. | A Research on Aero-engine Control Based on Deep Q Learning | |
| Zhang et al. | Optimization of cycle parameters of variable cycle engine based on response surface model | |
| Zare et al. | Novel Closed-Form Equation for Critical Pressure and Optimum Pressure Ratio for Turbojet Engines | |
| Ighodaro et al. | Off-design modelling of a turbo jet engine with operative afterburner | |
| Ungewitter et al. | CFD capabilities for hypersonic scramjet propulsive flowpath design | |
| Narayanan et al. | Simulation-based engine control for an ignition-assisted diesel engine with varying cetane number fuels | |
| Deng et al. | Direct Multi-Fidelity Integration of 3D CFD Models in a Gas Turbine with Fully Coupled Zooming Method |
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 |