Intelligent prediction method for aerodynamic force-heat-structure multi-field coupling of high-speed aircraft

Abstract

The invention relates to an aerodynamic force-thermal-structure multi-field coupling intelligent prediction method for a high-speed aircraft, and belongs to the technical field of high-speed aircraft design. According to the flight path of the aircraft, sample parameters used for calculating aerodynamic force and aerodynamic heat states are extracted; modeling the aircraft, and carrying out analog calculation to obtain an aerodynamic force and aerodynamic heat data set; calculating an aircraft temperature field along the trajectory; training the aerodynamic force data set to obtain an aerodynamic force prediction model; calculating structural stress displacement field data of the aircraft model; extracting concerned aircraft feature point corresponding data from the aerodynamic data set, the aircraft temperature field and the structural stress displacement field data of the aircraft model to obtain a feature point data set; training the feature point data set to obtain an accurate prediction result; according to the method, after artificial intelligence algorithm offline learning modeling is carried out on the aerodynamic force-heat-structure coupling data of analog simulation, the flight path is input, and the data, changing along the flight path, of the feature point data of the output aircraft is rapidly predicted.

Description

Technical Field

[0001] The invention belongs to the technical field of high-speed aircraft design, and relates to an intelligent prediction method for aerodynamic-thermal-structural multi-field coupling of a high-speed aircraft. Background Art

[0002] High-speed aircraft face complex multi-physical field coupling problems, which makes it difficult to predict the structural state online. Each physical field has different characteristics and the analysis methods are different, making it difficult to obtain a more accurate structural response. Therefore, exploring the coupling relationship between different physical fields and accurately predicting the aerodynamic / thermal environment, structural temperature, stress and displacement state of the aircraft are of great significance to the development of high-speed aircraft.

[0003] At present, high-speed aircraft are facing increasingly complex mission requirements, and the flight trajectory is more complex. The high temperature and temperature gradient generated by the aerodynamic heating on the structure will reduce the mechanical properties of the structural materials. The uneven thermal deformation and thermal stress inside the structure will change the stiffness characteristics of the structure, and at the same time change the inherent vibration characteristics of the structure. Under complex trajectories, it is very costly to conduct time-domain dynamic aeroelastic analysis of the entire trajectory. The fluid mechanics problems are characterized by high dimensions, strong nonlinearity, and large amounts of data. Although computational structural mechanics solutions are not as time-consuming as computational fluid dynamics solutions for NS equations from tens of thousands to tens of millions of grids, which usually take hours or even days, the time-consuming solution of differential equations for tens of thousands to hundreds of thousands of grids often reaches minutes or even hours, which cannot be ignored in multi-field coupling calculations. Summary of the invention

[0004] The technical problem solved by the present invention is: to overcome the shortcomings of the prior art, to propose an intelligent prediction method for aerodynamic-thermal-structural multi-field coupling of high-speed aircraft, to implement offline learning modeling of simulated aerodynamic-thermal-structural coupling data using an artificial intelligence algorithm, to input a flight trajectory, and to quickly predict the changes in output aircraft feature point data along the flight trajectory.

[0005] The solution of the present invention is:

[0006] The intelligent prediction method of aerodynamic-thermal-structural multi-field coupling of high-speed aircraft includes:

[0007] Step 1: extract sample parameters for calculating aerodynamic force and aerodynamic thermal state according to the flight trajectory of the aircraft;

[0008] Step 2: Model the aircraft, and simulate and calculate to obtain aerodynamic and aerodynamic thermal data sets according to the sample parameters extracted in step 1;

[0009] Step 3: Calculate the temperature field of the aircraft along the trajectory based on the aerodynamic thermal data set in step 2;

[0010] Step 4: training the aerodynamic data set in step 2 to obtain an aerodynamic prediction model;

[0011] Step 5: Calculate the structural stress displacement field data of the aircraft model according to the aircraft temperature field along the trajectory in step 3 and the aerodynamic force prediction model in step 4;

[0012] Step 6: Obtain feature point data set;

[0013] Step 7: Train the feature point data set to obtain accurate prediction results.

[0014] In the above-mentioned high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method, in the step one, a uniform sampling method is used to extract sample parameters; the sample parameters include the wall temperature, flight altitude, angle of attack and Mach number of the aircraft.

[0015] In the above-mentioned high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method, in the step 2, the method for simulating and calculating aerodynamic force and aerodynamic heat is:

[0016] The aircraft model is divided into aerodynamic grids; the sample parameters are input as boundary conditions; numerical simulation is performed to obtain aerodynamic and aerodynamic thermal data sets.

[0017] In the above-mentioned high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method, in step 3, the calculation method of the aircraft temperature field along the trajectory is:

[0018] S31, obtained by interpolation calculation of aerodynamic thermal data set tn Nodal aerodynamic heat at the moment;

[0019] S32, Setting t0 The initial structural temperature field of the aircraft at the moment is known, and the radiation boundary conditions are known; according to the flight trajectory of the aircraft, the heat load is known; iteratively obtain tn The initial structural temperature field of the aircraft at time t;

[0020] S33, Apply S31 to the aircraft model tn The node aerodynamic heat and radiation boundary conditions at time tn The initial structural temperature field of the aircraft at time t is obtained by transient structural heat transfer analysis using NASTRAN software. tn+1 Structural temperature field at time;

[0021] S34, from tn+1 The surface temperature of the aircraft in the structural temperature field at time tn+1 The node aerodynamic heat at time tn+1 The structural temperature field at time tn+1The initial structural temperature field of the aircraft at time t; loop steps S31-S33 to obtain the temperature field of the aircraft along the trajectory.

[0022] In the above-mentioned high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method, in S33, iteratively obtains tn The method for calculating the initial structural temperature field of the aircraft at time t is:

[0023] exist t0 Based on the initial structural temperature field of the aircraft at the moment, the thermal load is input as the load variable of the aircraft model, and the iterative method is used to obtain tn The initial structural temperature field of the aircraft at time .

[0024] In the above-mentioned high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method, in step 4, the method for training the aerodynamic data set of step 2 is:

[0025] S41, using POD method to reduce the order of aerodynamic data;

[0026] S42. According to the mapping relationship between the sample parameters and the reduced-order aerodynamic data, BP neural network training is adopted to obtain an aerodynamic prediction model.

[0027] In the above-mentioned high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method, in step 5, the method for calculating the structural stress and displacement of the aircraft model is:

[0028] S51. Calculate the aerodynamic prediction model. tn Aerodynamic force at time node;

[0029] S52, will tn The aerodynamic force at the time node and the aircraft temperature field along the trajectory in step 3 are applied to the aircraft model, and the structural strength analysis is performed using NASTRAN software to obtain the structural stress and displacement field;

[0030] S53, enter tn+1 Cycle S51-S52 constantly to obtain the aircraft structure stress and displacement field data along the trajectory.

[0031] In the above-mentioned high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method, in S51, the following is obtained: tn The method of aerodynamic force at a time node is:

[0032] Will tn The altitude, angle of attack, and Mach number of the aircraft at time t are used as the input of the aerodynamic prediction model, and the output is n Aerodynamic force at the time node.

[0033] In the above-mentioned high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method, in step 6, the method for obtaining the feature point data set is:

[0034] From the aerodynamic data set in step 2, the aircraft temperature field in step 3, and the structural stress displacement field data of the aircraft model in step 5, the corresponding data of the aircraft feature points of interest are extracted to obtain a feature point data set.

[0035] In the above-mentioned high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method, in step seven, the method for training the feature point data set is:

[0036] S71, using the POD method to reduce the order of the feature point data set;

[0037] S72, using a BP neural network to train the mapping relationship between the flight trajectory and the feature point data set after the reduction.

[0038] The beneficial effects of the present invention compared with the prior art are:

[0039] (1) Based on the aerodynamic-thermal-structural multi-field data set obtained by simulation, the present invention adopts the Proper Orthogonal Decomposition (POD) and back propagation neural network (BPNN) method to conduct offline training and learning on the multi-field data set, and establishes a fast and high-precision intelligent prediction method for high-speed aerodynamic-thermal-structural multi-field coupling effects;

[0040] (2) The present invention uses a large number of data sets obtained through simulation, based on the fact that machine learning methods are good at processing and analyzing large-scale data sets, and can discover patterns and associated characteristics that are difficult for humans to perceive, and introduces a BP network to train the data sets;

[0041] (3) The present invention reduces the order of high-dimensional data by introducing the POD method while ensuring the data characteristics, and the BP network trains the reduced-order data, which greatly reduces the time consumption and memory usage of model training;

[0042] (4) The present invention uses the trained intelligent prediction model as input, the flight trajectory of the aircraft, and achieves the predicted output of the aerodynamic force, aerodynamic heat, temperature, stress and displacement of the characteristic points of the aircraft within seconds. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is a multi-field coupling intelligent prediction flow chart of the present invention.

[0044] Figure 2 The aircraft temperature field calculation flow chart of the present invention;

[0045] Figure 3 This is a flow chart of structural stress displacement calculation of the present invention;

[0046] Figure 4 It is a schematic diagram of the feature point prediction data and simulation data of the present invention;

[0047] Figure 5 A schematic diagram of relative errors between feature point prediction data and simulation data of the present invention;

[0048] Figure 6 This is a schematic diagram of delay time prediction according to the present invention. DETAILED DESCRIPTION

[0049] The present invention will be further described below in conjunction with the embodiments.

[0050] The present invention aims at the problem of high-speed aerodynamic-thermal-structural multi-field coupling, and reduces the online prediction time of the aircraft as much as possible while ensuring the calculation accuracy. A high-speed aerodynamic-thermal-structural multi-field coupling effect intelligent prediction method is proposed. After the simulated aerodynamic-thermal-structural coupling data is modeled through offline learning of artificial intelligence algorithms, the method realizes the input flight trajectory and quickly predicts the output data of the changes of the aircraft feature point data along the flight trajectory.

[0051] Intelligent prediction method for aerodynamic-thermal-structural multi-field coupling of high-speed aircraft, such as Figure 1 As shown, the specific steps include:

[0052] Step 1: Extract sample parameters for calculating aerodynamic force and aero-thermal state according to the flight trajectory of the aircraft.

[0053] The sample parameters are extracted by uniform sampling method; the sample parameters include the wall temperature, flight altitude, angle of attack and Mach number of the aircraft.

[0054] Step 2: Model the aircraft and simulate and calculate the aerodynamic and aerodynamic heat data set based on the sample parameters extracted in step 1. The method for simulating and calculating aerodynamic and aerodynamic heat is:

[0055] The aircraft model is divided into aerodynamic grids; the sample parameters are input as boundary conditions; numerical simulation is performed to obtain aerodynamic and aerodynamic thermal data sets.

[0056] Step 3: Calculate the temperature field of the aircraft along the trajectory based on the aerodynamic thermal data set in step 2.

[0057] like Figure 2 The calculation method of the aircraft temperature field along the trajectory is as follows:

[0058] S31, obtained by interpolation calculation of aerodynamic thermal data set tn Aerodynamic heat of the node at the moment.

[0059] S32, Setting t0 The initial structural temperature field of the aircraft at the moment is known, and the radiation boundary conditions are known; according to the flight trajectory of the aircraft, the heat load is known; iteratively obtain tn The initial structural temperature field of the aircraft at time .

[0060] S33, Apply S31 to the aircraft model tn The node aerodynamic heat and radiation boundary conditions at time tn The initial structural temperature field of the aircraft at time t is obtained by transient structural heat transfer analysis using NASTRAN software. tn+1 The structural temperature field at time.

[0061] Iteration to obtain tn The method for calculating the initial structural temperature field of the aircraft at time t is:

[0062] exist t0 Based on the initial structural temperature field of the aircraft at the moment, the thermal load is input as the load variable of the aircraft model, and the iterative method is used to obtain tn The initial structural temperature field of the aircraft at time .

[0063] S34, from tn+1 The surface temperature of the aircraft in the structural temperature field at time tn+1 The node aerodynamic heat at time tn+1 The structural temperature field at time tn+1 The initial structural temperature field of the aircraft at time t; loop steps S31-S33 to obtain the temperature field of the aircraft along the trajectory.

[0064] Step 4: Train the aerodynamic data set in step 2 to obtain an aerodynamic prediction model.

[0065] The method for training the aerodynamic data set in step 2 is:

[0066] S41. The POD method is used to reduce the order of aerodynamic data.

[0067] S42. According to the mapping relationship between the sample parameters and the reduced-order aerodynamic data, BP neural network training is adopted.

[0068] S43. Check whether the trained aerodynamic prediction model meets the requirements through the test set data. If so, output the final result. If not, modify the reduced-order parameters and network parameters, retrain until it meets the design requirements, and save the aerodynamic prediction model that meets the requirements.

[0069] Step 5: Calculate the structural stress displacement field data of the aircraft model based on the aircraft temperature field along the trajectory in step 3 and the aerodynamic prediction model in step 4.

[0070] like Figure 3 As shown in Figure 2, the method for calculating the structural stress and displacement of the aircraft model is:

[0071] S51. Calculate the aerodynamic prediction model. tn In S51, obtain t n The method of aerodynamic force at a time node is:

[0072] Will tn The altitude, angle of attack, and Mach number of the aircraft at time t are used as the input of the aerodynamic prediction model, and the output is n Aerodynamic force at the time node.

[0073] S52, will tn The aerodynamic force at the time node and the aircraft temperature field along the trajectory in step 3 are applied to the aircraft model, and the structural strength analysis is performed using NASTRAN software to obtain the structural stress and displacement field;

[0074] S53, enter tn+1 Cycle S51-S52 constantly to obtain the aircraft structure stress and displacement field data along the trajectory.

[0075] Step 6: Obtain feature point data set.

[0076] The method to obtain the feature point data set is:

[0077] From the aerodynamic data set in step 2, the aircraft temperature field in step 3, and the structural stress displacement field data of the aircraft model in step 5, the corresponding data of the aircraft feature points of interest are extracted to obtain a feature point data set.

[0078] Step 7: Train the feature point data set to obtain accurate prediction results.

[0079] The method for training the feature point data set is:

[0080] S71. Use the POD method to reduce the order of the feature point data set.

[0081] S72, using a BP neural network to train the mapping relationship between the flight trajectory and the feature point data set after the reduction.

[0082] S73. Check whether the trained intelligent prediction model meets the requirements through the test set data. If so, output the final result. If not, modify the reduction parameters and network parameters, retrain until it meets the design requirements, and save the intelligent prediction model that meets the requirements.

[0083] Through the intelligent prediction model, the test trajectory is input, and the aerodynamic heat, aerodynamic force, temperature, displacement and stress data of the characteristic node are obtained through the intelligent prediction program. Figure 4 , Figure 5 As can be seen from the figure, the prediction results of the intelligent prediction model have a small deviation from the simulation data results, and the maximum relative error does not exceed 10%. The prediction time statistics were performed 100 times. Figure 6 As shown in the figure, the longest prediction time is 0.6423s, the shortest time is 0.3376s, and the average prediction time is 0.3975s. The prediction delay time is short.

[0084] Based on the aerodynamic-thermal-structural multi-field data set obtained by simulation, the present invention adopts the Proper Orthogonal Decomposition (POD) and back propagation neural network (BPNN) methods to perform offline training and learning on the multi-field data set, and establishes a fast and high-precision high-speed aerodynamic-thermal-structural multi-field coupling effect intelligent prediction method. Different from the traditional prediction of multi-field coupling effects of high-speed aircraft, the present invention uses a large number of data sets obtained by simulation, based on the machine learning method, which is good at processing and analyzing large-scale data sets, and can find the characteristics of patterns and associations that are difficult for humans to perceive, and introduces the BP network to train the data set. At the same time, considering that the data obtained by the simulation calculation of the aircraft are all high-dimensional data, direct training and learning will occupy a large amount of memory and take a long time. By introducing the POD method, the high-dimensional data is reduced in order under the premise of ensuring the data characteristics, and the BP network is trained on the reduced data, which greatly reduces the time consumption and memory occupation of the model training. According to the trained intelligent prediction model, the present invention inputs the flight trajectory of the aircraft, and realizes the prediction output of the aerodynamic force, aerodynamic heat, temperature, stress and displacement of the characteristic points of the aircraft within seconds.

[0085] Although the present invention has been disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art may make possible changes and modifications to the technical solution of the present invention by using the methods and technical contents disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent prediction method for aerodynamic-thermal-structural multi-field coupling of high-speed aircraft, characterized by: include: Step 1: extract sample parameters for calculating aerodynamic force and aerodynamic thermal state according to the flight trajectory of the aircraft; Step 2: Model the aircraft, and simulate and calculate to obtain aerodynamic and aerodynamic thermal data sets according to the sample parameters extracted in step 1; Step 3: Calculate the temperature field of the aircraft along the trajectory based on the aerodynamic thermal data set in step 2; Step 4: training the aerodynamic data set in step 2 to obtain an aerodynamic prediction model; Step 5: Calculate the structural stress displacement field data of the aircraft model according to the aircraft temperature field along the trajectory in step 3 and the aerodynamic force prediction model in step 4; Step 6: Obtain feature point data set; Step 7: Train the feature point data set to obtain accurate prediction results.

2. The high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method according to claim 1 is characterized by: In the step 1, a uniform sampling method is used to extract sample parameters; the sample parameters include the wall temperature, flight altitude, angle of attack and Mach number of the aircraft.

3. The high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method according to claim 1 is characterized by: In the step 2, the method for simulating and calculating aerodynamic force and aerodynamic heat is: The aircraft model is divided into aerodynamic grids; the sample parameters are input as boundary conditions; numerical simulation is performed to obtain aerodynamic and aerodynamic thermal data sets.

4. The high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method according to claim 1 is characterized by: In step 3, the calculation method of the aircraft temperature field along the trajectory is: S31, obtained by interpolation calculation of aerodynamic thermal data set tn Nodal aerodynamic heat at the moment; S32, Setting t0 The initial structural temperature field of the aircraft at the moment is known, and the radiation boundary conditions are known; according to the flight trajectory of the aircraft, the heat load is known; iteratively obtain tn The initial structural temperature field of the aircraft at time t; S33, Apply S31 to the aircraft model tn The node aerodynamic heat and radiation boundary conditions at time tn The initial structural temperature field of the aircraft at time t is obtained by transient structural heat transfer analysis using NASTRAN software. tn+1 Structural temperature field at time; S34, from tn+1 The surface temperature of the aircraft in the structural temperature field at time tn+1 The node aerodynamic heat at time tn+1 The structural temperature field at time tn+1 The initial structural temperature field of the aircraft at time t; loop steps S31-S33 to obtain the temperature field of the aircraft along the trajectory.

5. The high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method according to claim 4 is characterized by: In S33, iteratively obtain tn The method for calculating the initial structural temperature field of the aircraft at time t is: exist t0 Based on the initial structural temperature field of the aircraft at the moment, the thermal load is input as the load variable of the aircraft model, and the iterative method is used to obtain tn The initial structural temperature field of the aircraft at time .

6. The high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method according to claim 1 is characterized by: In step 4, the method for training the aerodynamic data set of step 2 is: S41, using POD method to reduce the order of aerodynamic data; S42. According to the mapping relationship between the sample parameters and the reduced-order aerodynamic data, BP neural network training is adopted to obtain an aerodynamic prediction model.

7. The high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method according to claim 1 is characterized by: In step 5, the method for calculating the structural stress displacement of the aircraft model is: S51. Calculate the aerodynamic prediction model. tn Aerodynamic force at time node; S52, will tn The aerodynamic force at the time node and the aircraft temperature field along the trajectory in step 3 are applied to the aircraft model, and the structural strength analysis is performed using NASTRAN software to obtain the structural stress and displacement field; S53, enter tn+1 Cycle S51-S52 constantly to obtain the aircraft structure stress and displacement field data along the trajectory.

8. The high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method according to claim 7 is characterized by: In S51, obtain tn The method of aerodynamic force at a time node is: Will tn The altitude, angle of attack, and Mach number of the aircraft at time t are used as the input of the aerodynamic prediction model, and the output is n Aerodynamic force at the time node.

9. The high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method according to claim 1 is characterized by: In step 6, the method for obtaining the feature point data set is: From the aerodynamic data set in step 2, the aircraft temperature field in step 3, and the structural stress displacement field data of the aircraft model in step 5, the corresponding data of the aircraft feature points of interest are extracted to obtain a feature point data set.

10. The high-speed aircraft aerodynamic-thermal-structural multi-field coupled intelligent prediction method according to claim 1 is characterized by: In step 7, the method for training the feature point data set is: S71, using the POD method to reduce the order of the feature point data set; S72, using a BP neural network to train the mapping relationship between the flight trajectory and the feature point data set after the reduction.

Images