Aircraft aerodynamic characteristic intelligent reconstruction method based on multi-source data fusion

Abstract

The invention discloses an aircraft aerodynamic characteristic intelligent reconstruction method based on multi-source data fusion, and relates to the field of intelligent fluid mechanics prediction.The aircraft aerodynamic characteristic intelligent reconstruction method comprises the steps that aircraft multi-source aerodynamic characteristic data are prepared, and an aircraft flow characteristic intelligent dimension reduction model is established; the method comprises the following steps: preparing a data set of an aircraft aerodynamic characteristic intelligent prediction model for multi-source data fusion, establishing the aircraft aerodynamic characteristic intelligent prediction model for multi-source data fusion, carrying out iterative training on the aircraft aerodynamic characteristic intelligent prediction model, and carrying out verification by adopting a corresponding verification data set. And the weight model with the optimal performance is selected to predict the subsequent to-be-tested working condition. According to the aircraft aerodynamic characteristic intelligent reconstruction method based on multi-source data fusion provided by the invention, efficient fusion of multi-source data information can be realized, so that a high-precision aircraft aerodynamic characteristic intelligent reconstruction prediction model is constructed.

Description

Technical Field

[0001] This invention relates to the field of intelligent fluid dynamics prediction. More specifically, this invention relates to a method for intelligent reconstructing the aerodynamic characteristics of aircraft based on multi-source data fusion. Background Technology

[0002] Currently, the prediction of aerodynamic characteristics in the design and development of aerospace vehicles mainly relies on wind tunnel testing and numerical calculation methods, with some simple corrections made based on flight test data. Wind tunnel testing is the most reliable method for obtaining limited sample data under finite operating conditions, but it is characterized by high cost, long cycle time, and small sample size. High-fidelity CFD numerical simulations are usually computationally intensive and time-consuming. The significant improvement in the advanced indicators of aircraft places higher demands on the accuracy of aerodynamic characteristic prediction, and no single method can accurately predict the aerodynamic characteristics of an aircraft.

[0003] Current methods for reconstructing aircraft aerodynamic characteristics based on multi-source data fusion mainly rely on building machine learning models to map input parameters to output aerodynamic characteristics. Most studies only use flow field parameters and aircraft attitude as input parameters, without fully considering the aircraft's flow field structural features (such as flow field structure and aircraft aerodynamic shape). Furthermore, when using numerical computation data fusion, the flow field order reduction and feature selection methods employed to simplify highly complex nonlinear flow field data still have certain limitations.

[0004] In summary, data mining of numerically calculated images of aircraft flow field structures with rich flow characteristics, exploration of dimensionality reduction methods for complex high-dimensional data, and deep fusion of multi-source data are essential requirements for improving the accuracy of predicting the aerodynamic characteristics of new aircraft. Summary of the Invention

[0005] One object of the present invention is to solve at least the above-mentioned problems and / or defects, and to provide at least the advantages described below.

[0006] To achieve these objectives and other advantages of the present invention, a method for intelligent reconstructing the aerodynamic characteristics of an aircraft based on multi-source data fusion is provided, comprising: S1. Based on the aerodynamic characteristics of the aircraft, select the corresponding wind tunnel test data and numerical simulation data; S2. Based on the image classification algorithm as the network architecture for extracting aircraft flow features, and combined with the t-distributed random neighborhood embedding algorithm t-SNE as the nonlinear dimensionality reduction method for aircraft flow features, an intelligent dimensionality reduction model for aircraft flow features is established, and the corresponding dimensionality reduction data is obtained through iterative training. S3. Using a multi-input-multi-output deep artificial neural network as the basic architecture, an intelligent prediction model for the aerodynamic characteristics of aircraft based on multi-source data fusion is established. S4. The numerical simulation data in S1 and the dimensionality reduction data in S2 are used as input parameters for the intelligent prediction model of the aerodynamic characteristics of the aircraft. Iterative training is performed to select the weight model with the best performance for prediction of subsequent test conditions, thus completing the intelligent reconstruction of the aerodynamic characteristics of the aircraft.

[0007] Preferably, in S1, the wind tunnel test data includes: axial force I, normal force I, lateral force I, roll moment I, yaw moment I, and pitch moment I; The numerical simulation data includes: numerical simulation results of the aerodynamic characteristics of the aircraft and pressure distribution cloud maps of the flow field in the numerical simulation of the aircraft; The numerical simulation results of the aerodynamic characteristics of the aircraft include: axial force II, normal force II, lateral force II, roll moment II, yaw moment II, and pitch moment II. The pressure distribution cloud map of the flow field simulated by the numerical calculation of the aircraft is classified according to Mach number and subjected to data augmentation processing when in use.

[0008] Preferably, in S2, the aircraft flow feature extraction network includes: a feature extraction module and a classification module; The feature extraction module includes a 5-level structure layer. Each level includes two consecutive 3×3 convolutional layers and a ReLU activation layer. Adjacent levels are downsampled through a 2×2 max pooling layer. The classification module consists of three fully connected layers, with the output features of the last fully connected layer serving as the input to the t-SNE algorithm.

[0009] Preferably, in S2, before iteratively training the intelligent dimensionality reduction model of the aircraft flow characteristics, the data consisting of the pressure distribution cloud map of the flow field of the aircraft numerical calculation simulation in S1 is divided into training set, validation set and test set based on the corresponding operating conditions. When iteratively training the intelligent dimensionality reduction model of aircraft flow characteristics, the training set is used as input for iterative training, and the corresponding validation set is used for validation to extract the high-dimensional features corresponding to the last fully connected layer of the optimal weight model. The t-SNE algorithm is used to perform nonlinear dimensionality reduction of the high-dimensional features.

[0010] Preferably, in S3, the intelligent prediction model for the aerodynamic characteristics of the aircraft employs a stochastic search strategy to automatically train the network and hyperparameters, and the hyperparameters of the stochastic search strategy are set as follows: The network has a minimum of 3 layers and a maximum of 10 layers. The minimum number of neurons in a hidden layer is 5, and the maximum number of neurons in a hidden layer is 10. The minimum order of magnitude of the regularization parameter is 2, and the maximum order of magnitude of the regularization parameter is 5. The learning rate is 0.001, and the total number of iterations is 8000.

[0011] This invention offers at least the following advantages: Addressing the limitations of relying on single methods and the low efficiency of multi-source data fusion in predicting aircraft aerodynamic characteristics, this invention first performs data mining on the numerical calculation results of the aircraft flow field structure, specifically extracting features and performing dimensionality reduction analysis on flow field images rich in flow characteristics and aircraft geometric configuration information. Then, based on this, wind tunnel test data is fused to achieve efficient fusion of multi-source data information, thereby constructing a high-precision intelligent reconstruction and prediction model for aircraft aerodynamic characteristics.

[0012] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating the intelligent reconstruction method for aircraft aerodynamic characteristics based on multi-source data fusion in this invention. Figure 2 This is an extended example diagram of the pressure distribution cloud map in the flow field of the numerical simulation of the aircraft of the present invention; Figure 3 This is a schematic diagram illustrating the visualization results of the t-SNE dimensionality reduction algorithm of this invention. Detailed Implementation

[0014] The present invention will now be described in further detail with reference to the accompanying drawings, so that those skilled in the art can implement it based on the description.

[0015] like Figure 1 As shown, a method for intelligent reconstructing the aerodynamic characteristics of an aircraft based on multi-source data fusion includes: Step 1: Prepare multi-source aerodynamic characteristic data for the aircraft, including wind tunnel test data and numerical simulation data for corresponding operating conditions (such as flow field information parameters and aircraft attitude information).

[0016] For example, in one embodiment, for the aerodynamic characteristics study of an aircraft, the wind tunnel test data includes: axial force, normal force, lateral force, roll moment, yaw moment, and pitch moment. The numerical simulation data includes: (1) numerical simulation results of the aircraft's aerodynamic characteristics: axial force, normal force, lateral force, roll moment, yaw moment, and pitch moment; (2) numerical simulation flow field pressure distribution cloud map of the aircraft, which is classified according to Mach number in subsequent specific use. In addition, in order to enhance the robustness of the deep learning model, the data is augmented. For example Figure 2 As shown, under the operating conditions of Mach numbers of 0.4, 0.6, and 0.8 at a certain angle of attack, the flow field pressure distribution cloud map is output at four different angles to complete the data augmentation process.

[0017] Step 2: Establish an intelligent dimensionality reduction model for aircraft flow features. This model uses image classification algorithms from the field of computer vision as the network architecture for aircraft flow feature extraction, and employs the t-distributed stochastic neighbor embedding (t-SNE) algorithm as the nonlinear dimensionality reduction method for aircraft flow features. The aircraft flow feature extraction network includes a feature extraction module and a classification module. The backbone feature extraction module consists of 5 layers, each layer including two consecutive 3×3 convolutional layers and a ReLU activation layer. Downsampling between adjacent layers is achieved using a 2×2 max pooling layer. With each layer, the size (width and height) of the feature map in the next layer is reduced to half that of the previous layer, while the number of channels in the feature map is doubled. The classification module consists of three fully connected layers, and finally, image classification is achieved through the Softmax activation function. The output features of the last fully connected layer serve as the input for the subsequent t-SNE algorithm.

[0018] The design details, operating operators, and output data stream format of the aircraft flow feature extraction network are shown in Table 1. In this example, the image input size is 512 × 512, and the output category is 3. The data dimension provided for subsequent t-SNE algorithm processing is 4096.

[0019] Table 1 During model training, pre-trained weights from the VGG13 network, trained on the large ImageNet dataset, are loaded. Since the VGG13 network shares the same backbone feature extraction module structure as the aircraft flow feature extraction network in this invention, loading these pre-trained weights to initialize features effectively improves training efficiency and performance.

[0020] After multiple model training and hyperparameter optimizations, the optimal hyperparameter settings were determined: the initial learning rate was 1e-3, the minimum learning rate was 1e-4, and the learning rate was adjusted using a step-down method; the Adam optimizer was used to train the model, with a momentum parameter of 0.9; the weight decay coefficient was 5e-4; the total number of training epochs was 50; and the settings for loading the pre-trained model were: the number of training epochs frozen was 5, the batch parameter size for frozen training was 32, and the batch parameter size after unfreezing was 16.

[0021] Once the model is built, the data from the numerical simulation flow field pressure distribution cloud map obtained in step one is divided into training, validation, and test sets according to the corresponding operating conditions and a predetermined ratio. The training set is then input into the model for iterative training, and the model is validated using the corresponding validation dataset. The weight model with the best performance is then selected.

[0022] The high-dimensional features of the last fully connected layer of the optimal weight model are extracted, and the t-SNE algorithm is used for nonlinear dimensionality reduction of these features. In this example, three Mach numbers are used as features for classification. In subsequent research on intelligent reconstruction of aircraft aerodynamic characteristics based on multi-source data fusion, nine embedding space dimensions, ranging from 2 to 10, were set for cross-validation. Ultimately, a 2-dimensional embedding dimension was determined to have optimal performance. The visualization results of the t-SNE dimensionality reduction algorithm are shown below. Figure 3 As shown, flow field structure samples with the same Mach number clearly and intuitively cluster together, indicating that the aircraft flow feature extraction network proposed in this invention has excellent aircraft flow feature extraction capabilities.

[0023] Step 3: Prepare the dataset for the intelligent prediction model of aircraft aerodynamic characteristics using multi-source data fusion. Specifically, the numerical simulation data of aircraft aerodynamic characteristics obtained in Step 1 and the intelligent dimensionality reduction results of aircraft flow characteristics obtained in Step 2 are used as the input parameters of the model, and the wind tunnel test data obtained in Step 1 are used as the output parameters of the model. Table 2 shows a data example for the intelligent prediction model of aircraft aerodynamic characteristics using multi-source data fusion. The input parameters are two dimensionality reduction feature values ​​and six numerical simulation values. The output parameters are six wind tunnel test values.

[0024] Table 2 Step 4: Establish an intelligent prediction model for aircraft aerodynamic characteristics based on multi-source data fusion. This model uses a multi-input multi-output deep artificial neural network as its basic architecture (e.g., Figure 1 The example illustrates an intelligent prediction model for aircraft aerodynamic characteristics based on multi-source data fusion, employing a random search strategy to automatically train the network and hyperparameters. In this example, the hyperparameters for the random search strategy are set as follows: minimum number of network layers: 3; maximum number of network layers: 10; minimum number of neurons in hidden layers: 5; maximum number of neurons in hidden layers: 10; minimum order of magnitude of regularization parameter: 2; maximum order of magnitude of regularization parameter: 5; learning rate: 0.001; total number of iterations: 8000. In practice, these parameters can be adjusted based on factors such as dataset size, hardware computing power, and time efficiency.

[0025] Step 5: Next, input the training dataset formed in Step 3 into the intelligent prediction model of aircraft aerodynamic characteristics fused from multiple sources for iterative training. Then, use the corresponding validation dataset for validation. Select the weight model with the best performance to predict the subsequent test conditions.

[0026] The prediction results for some operating conditions using the selected optimal multi-source data fusion intelligent prediction model for aircraft aerodynamic characteristics are shown in Table 3. For both extrapolation and interpolation of operating conditions, the error is within 3%, which indicates that the intelligent reconstruction method for aircraft aerodynamic characteristics based on multi-source data fusion established in this invention has good prediction accuracy.

[0027] Table 3 The above solution is merely an illustration of a preferred example and is not limited thereto. When implementing this invention, appropriate substitutions and / or modifications can be made according to the user's needs.

[0028] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. It can be applied to various fields suitable for the present invention. Other modifications can be readily made by those skilled in the art. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and examples shown and described herein.

Claims

1. A method for intelligent reconstruction of aircraft aerodynamic characteristics based on multi-source data fusion, characterized in that: include: S1. Based on the study of the aerodynamic characteristics of the aircraft, select the corresponding wind tunnel test data and numerical calculation simulation data; S2. Based on the image classification algorithm as the network architecture for extracting aircraft flow features, the t-distributed random neighbor embedding algorithm t-SNE is combined as a nonlinear dimensionality reduction method for aircraft flow features to establish an intelligent dimensionality reduction model for aircraft flow features, and obtain the corresponding dimensionality reduction data through iterative training; S3. Using a multi-input and multi-output deep artificial neural network as the basic framework, an intelligent prediction model for aircraft aerodynamic characteristics based on multi-source data fusion is established; S4. The numerical calculation simulation data in S1 and the dimension reduction data in S2 are used as input parameters of the intelligent prediction model of the aerodynamic characteristics of the aircraft, and iterative training is performed to select the weighted model with the best performance to predict the subsequent working conditions to be tested, thereby completing the intelligent reconstruction of the aerodynamic characteristics of the aircraft.

2. The method for intelligent reconstruction of aircraft aerodynamic characteristics based on multi-source data fusion according to claim 1, characterized in that: In S1, the wind tunnel test data include: axial force I, normal force I, lateral force I, rolling moment I, yaw moment I, pitching moment I; The numerical calculation simulation data include: numerical calculation simulation results of aircraft aerodynamic characteristics and pressure distribution cloud diagram of numerical calculation simulation flow field of aircraft; The numerical calculation simulation results of the aerodynamic characteristics of the aircraft include: axial force II, normal force II, lateral force II, rolling moment II, yaw moment II, and pitch moment II; The numerical calculation simulation flow field pressure distribution cloud map of the aircraft is classified according to the Mach number when used, and data expansion processing is performed.

3. The method for intelligent reconstruction of aircraft aerodynamic characteristics based on multi-source data fusion according to claim 1, characterized in that: In S2, the aircraft flow feature extraction network includes: a feature extraction module and a classification module; The feature extraction module includes 5 levels of structural layers, each of which includes two consecutive 3×3 convolutional layers and a ReLU activation layer, and the structural layers of adjacent levels are downsampled through a 2×2 maximum pooling layer; The classification module includes three fully connected layers, and the output features of the last fully connected layer are used as the input of the t-SNE algorithm.

4. The method for intelligent reconstruction of aircraft aerodynamic characteristics based on multi-source data fusion according to claim 2, characterized in that: In S2, before the aircraft flow characteristic intelligent dimension reduction model is iteratively trained, the data consisting of the pressure distribution cloud map of the aircraft numerical simulation flow field in S1 is divided into a training set, a validation set and a test set based on the corresponding working conditions; When the intelligent dimensionality reduction model of aircraft flow characteristics is iteratively trained, the training set is used as input for iterative training, and the corresponding validation set is used for validation to extract the high-dimensional features corresponding to the last fully connected layer of the optimal weight model, and the t-SNE algorithm is used to perform nonlinear dimensionality reduction of high-dimensional features.

5. The method for intelligent reconstruction of aircraft aerodynamic characteristics based on multi-source data fusion according to claim 1, characterized in that: In S3, the aircraft aerodynamic characteristics intelligent prediction model adopts a random search strategy to realize automatic training of the network and hyperparameters, and the hyperparameters of the random search strategy are set as follows: The minimum number of network layers is 3, the maximum number of network layers is 10, the minimum number of neurons in the hidden layer is 5, the maximum number of neurons in the hidden layer is 10, the minimum order of magnitude of the regularization parameter is 2, the maximum order of magnitude of the regularization parameter is 5, the learning rate is 0.001, and the total number of iterations is 8000.

Images