Aircraft airborne component heating prediction method and device, and storage medium

Abstract

The invention relates to an airplane airborne component heating prediction method and device and a storage medium, and belongs to the technical field of space flight and aviation. The method comprises the following steps: collecting an original historical data set of airborne components; sampling the original historical data set by adopting an improved Latin hypercube sampling method to obtain a training data sample; performing fluid dynamic simulation calculation based on the training data sample to obtain thermal characteristic calculation results of the airborne component under different working conditions; taking attribute information and flight condition information of the airborne component of the airplane as input data, taking the thermal characteristic calculation result as output data, training a convolutional neural network model, and generating a thermal prediction neural network model; the thermal prediction neural network model is used for performing thermal prediction according to the new input data and generating temperature distribution data of the airborne component of the aircraft at the prediction moment. The method improves the sampling uniformity and comprehensiveness, and can achieve the real-time and whole-process prediction of the thermal characteristics of the airborne components of the airplane.

Description

Technical Field

[0001] The present application belongs to the field of aerospace technology, and specifically relates to a method, device and storage medium for predicting heat generation of aircraft onboard components. Background Art

[0002] In many industrial applications, electronic components often face the problem of heating up in harsh environments or under long-term use conditions, which may lead to performance degradation or device damage. Heat dissipation equipment can solve this problem to a certain extent, but at the same time, the volume, mass, power consumption and cost will also increase significantly. Therefore, it is necessary to consider the specific application objects and usage environment of the components and predict the heating process and heating characteristics in a targeted manner. The prediction results can help improve the process during component design, or help design variable frequency, low power consumption, and low-cost heat dissipation supporting equipment.

[0003] However, in the prior art, the heat prediction for electronic components, especially aircraft-borne electronic components, is mainly completed by environmental stress testing. This test method usually only considers extreme conditions or a few typical working conditions, and its main purpose is to ensure that the components are not damaged. However, these tests are usually not sufficient to fully reflect the temperature changes and thermal stresses of components during actual use, and pay less attention to the real-time temperature changes, heat dissipation power consumption and long-term service life of components.

[0004] Therefore, how to more comprehensively reflect the heating conditions during the entire mission cycle is a topic worthy of study in this field. Summary of the invention

[0005] In view of the above analysis, the embodiments of the present invention aim to provide a method, device and storage medium for predicting heating of aircraft onboard components, so as to solve the problem that the prior art cannot reflect the heating conditions in the entire mission cycle.

[0006] In a first aspect of the present application, a method for predicting heating of aircraft onboard components is provided, comprising:

[0007] Collecting original historical data sets of airborne components;

[0008] The original historical data set is sampled using an improved Latin hypercube sampling method to obtain training data samples related to the fever prediction task;

[0009] Perform fluid dynamics simulation calculation based on the training data samples to obtain thermal characteristic calculation results of airborne components under different working conditions;

[0010] Using the attribute information of the aircraft's airborne components and the flight condition information as input data, using the thermal characteristic calculation results as output data, training the convolutional neural network model based on the training data samples, and generating a thermal prediction neural network model;

[0011] The thermal prediction neural network model is used to perform thermal prediction based on new input data, and the process includes:

[0012] Obtain the attribute information of aircraft onboard components;

[0013] Acquiring in real time the flight condition information of the aircraft at the predicted time;

[0014] The attribute information and flight condition information are input into the thermal prediction neural network model to generate temperature distribution data of aircraft onboard components at the prediction time.

[0015] Optionally, the adopting of the improved Latin hypercube sampling method to sample the original historical data set to obtain training data samples related to the fever prediction task includes:

[0016] Assuming that the sampling object is N-dimensional, the sampling times are M, and 0-1 uniform sampling is performed on each dimension, the sampling process can be described as an N×M matrix, denoted as A; the improved Latin hypercube sampling method includes the following steps:

[0017] S1: Divide each dimension M into equal parts, that is, the length of each interval is 1 / M;

[0018] S2: Randomly select N values ​​from the first subinterval [0,1 / M] as the first column of matrix A;

[0019] S3: Randomly select N values ​​from the second subinterval [1 / M, 2 / M] as the second column of matrix A;

[0020] S4: And so on, until the last subinterval is reached and the Nth column of matrix A is formed;

[0021]

[0022] Among them, P 11 ,P 12 …,P 1M ,P 21 …,P NM is the sampling value;

[0023] S5: Shuffle each row of A to obtain matrix B. Finally, each column of matrix B represents a sample point, and there are M sample points in total.

[0024] S6: Using parameter G max Evaluate whether the sample point distribution of matrix B is reasonable. If it is unreasonable, return to resample. max is a parameter used to evaluate the uniform distribution of samples in the multidimensional parameter space.

[0025] Optionally, the parameter G max Calculated according to the sampling dimension N and the number of sampling times M:

[0026] oeLh max =(395.7N 0.0048 -394.1)-(0.24N -0.94 +1.03)lnM.

[0027] Optionally, the convolutional neural network includes a component attribute encoder, a flight condition encoder, and a thermal calculation decoder;

[0028] Among them, the component attribute encoder is used to obtain the attribute information of the aircraft's onboard components; the flight condition encoder is used to obtain the aircraft's flight condition information; the thermal calculation decoder is used to sample the original historical data set using an improved Latin hypercube sampling method to obtain training data samples related to the heat prediction task; based on the training data samples, fluid dynamics simulation calculations are performed to obtain thermal characteristic calculation results of airborne components under different operating conditions.

[0029] Optionally, the attribute information of the aircraft-borne components includes material information, volume information, and shape information of the components; the flight condition information includes flight altitude information, speed information, overload information, and temperature information.

[0030] Optionally, the convolutional neural network adopts a U-Net architecture, and the U-Net includes five channels:

[0031] First channel: used for inputting component attribute information encoded by the component attribute encoder;

[0032] The second channel is used to input the flying height information encoded by the component attribute encoder;

[0033] The third channel is used to input the speed information encoded by the component attribute encoder;

[0034] The fourth channel is used to input the overload information encoded by the component attribute encoder;

[0035] The fifth channel is used to input the air temperature information encoded by the component attribute encoder;

[0036] With the component property encoder and the flight condition encoder as input and the thermal calculation decoder as output, U-Net is trained to generate a thermal prediction neural network model.

[0037] Optionally, performing fluid dynamics simulation calculation based on the training data sample to obtain calculation results of thermal characteristics of airborne components under different working conditions includes:

[0038] Collect information about the operating conditions of onboard components and determine the thermal characteristics to be studied;

[0039] Construct geometric models of airborne components so that the models reflect the shapes and sizes of the actual components;

[0040] Generate computational mesh on the geometric model and set boundary conditions according to actual working conditions and environment;

[0041] The corresponding fluid dynamics and heat transfer models are selected for simulation calculation to obtain the thermal characteristics calculation results of airborne components under different working conditions.

[0042] Optionally, the airborne component is an airborne radio altimeter.

[0043] In a second aspect of the present application, a device for predicting the heating of aircraft onboard components is provided, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, a method for predicting the heating of aircraft onboard components is implemented according to any one of the above-described methods.

[0044] According to a third aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for predicting the heating of aircraft onboard components is implemented according to any one of the above-mentioned methods.

[0045] The method for predicting the heating of aircraft airborne components provided in the present application collects the original historical data set of airborne components; uses the improved Latin hypercube sampling method to sample the original historical data set to obtain training data samples related to the heating prediction task; performs fluid dynamics simulation calculations based on the training data samples to obtain the thermal characteristics calculation results of airborne components under different working conditions; uses the attribute information and flight condition information of the aircraft airborne components as input data, uses the thermal characteristics calculation results as output data, and trains the convolutional neural network model based on the training data samples to generate a thermal prediction neural network model; the thermal prediction neural network model is used to perform thermal prediction based on new input data to generate temperature distribution data of aircraft airborne components at the prediction time. The present application improves the uniformity and comprehensiveness of sampling by using the improved Latin hypercube sampling method, and then extracts the key factors that affect the heating of components. Through the established thermal prediction neural network model, the thermal characteristics of aircraft airborne components can be predicted in real time and throughout the process.

[0046] In addition, the present application also provides aircraft onboard component heating prediction equipment and storage media having the above-mentioned technical effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of this specification. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0048] Figure 1 A flowchart of a specific implementation of the method for predicting heating of aircraft airborne components provided in this application;

[0049] Figure 2 This is a schematic diagram of the convolutional neural network model structure for this application;

[0050] Figure 3 This is a schematic diagram of the U-Net model structure for this application;

[0051] Figure 4 A schematic diagram of the process of performing thermal prediction based on new input data by the thermal prediction neural network model of this application;

[0052] Figure 5 This is a structural block diagram of the aircraft onboard component heating prediction device provided in this application. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. It should be noted that, in the absence of conflict, the embodiments in the present disclosure and the features in the embodiments can be combined, separated, interchanged and / or rearranged with each other. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0054] The terms used here are for the purpose of describing specific embodiments, and are not intended to be restrictive. As used here, unless the context clearly indicates otherwise, the singular forms "one (kind, person)" and "said (the)" are also intended to include plural forms. In addition, when the terms "comprise" and / or "include" and their variations are used in this specification, it is explained that there are stated features, integral bodies, steps, operations, parts, assemblies and / or their groups, but it is not excluded that there are or add one or more other features, integral bodies, steps, operations, parts, assemblies and / or their groups. It should also be noted that, as used here, the terms "substantially", "approximately" and other similar terms are used as approximate terms and not as degree terms, so that they are used to explain the inherent deviations of the measured values, calculated values ​​and / or the values ​​provided that will be recognized by those of ordinary skill in the art.

[0055] A flowchart of a specific implementation of the method for predicting the heating of aircraft airborne components provided in this application is as follows: Figure 1 As shown, the method includes:

[0056] S101: Collect the original historical data set of airborne components.

[0057] The airborne components may be airborne radio altimeters, or other aircraft airborne components, which are not limited here. The original historical data set is various performance and environmental data under actual operating conditions. These data include but are not limited to the following types:

[0058] (1) Operation condition data:

[0059] Flight altitude: working data of components at different altitudes;

[0060] Flight speed: the impact of different speeds on component performance.

[0061] Ambient temperature: The working data of components under various external temperature conditions.

[0062] (2) Component performance data

[0063] Temperature record: temperature data of components during operation.

[0064] Power usage: This includes current and voltage logging, which can help assess power consumption and heat generation.

[0065] Fault record: Records the performance degradation or failure of components under specific conditions.

[0066] (3) Physical and mechanical data:

[0067] Overload data: records the physical stress or overload conditions to which components are subjected during flight.

[0068] Vibration data: The level of vibration to which components are subjected during flight, which may affect their heating and performance.

[0069] (4) Environmental data:

[0070] Air pressure data: Air pressure data at different flight altitudes affects the heat dissipation performance of components.

[0071] Humidity data: The possible impact of ambient humidity on component performance.

[0072] It is understandable that the above data can be collected through airborne sensors of different types and dimensions. The collection of this data can provide a comprehensive understanding of the behavior patterns and performance of components under various actual working conditions, preparing for the next step of sampling.

[0073] S102: Using an improved Latin hypercube sampling method to sample the original historical data set to obtain training data samples related to the fever prediction task.

[0074] The heating process of the airborne altimeter is related to many factors such as flight altitude, speed and even installation location. This step aims to select representative samples from the extensive data set obtained by S101 that are useful for the heating prediction task. Find out the most critical factors, ignore the minor factors, and then establish a thermal prediction neural network. One specific implementation method is to conduct sensitivity analysis, the premise of which is sampling and experimentation.

[0075] Latin Hypercube Sampling (LHS) is a low-cost sampling method and is the premise for analyzing the sensitivity of model output to input. Traditional LHS cannot evenly fill the entire problem space, so this application proposes an improved LHS method.

[0076] LHS is a stratified sampling technique that divides the sampling space into different layers according to certain characteristics or rules, and then independently and randomly selects samples from different layers.

[0077] Assuming that the sampling object is N-dimensional, the sampling times are M, and 0-1 uniform sampling is performed on each dimension, the sampling process can be described as an N×M matrix, denoted as A; the improved Latin hypercube sampling method includes the following steps:

[0078] S1: Divide each dimension M into equal parts, that is, the length of each interval is 1 / M.

[0079] S2: Randomly select N values ​​from the first subinterval [0,1 / M] as the first column of matrix A.

[0080] S3: Randomly select N values ​​from the second subinterval [1 / M, 2 / M] as the second column of matrix A.

[0081] S4: And so on, until the last subinterval is reached and the Nth column of matrix A is formed;

[0082]

[0083] Among them, P 11 ,P 12 …,P 1M ,P 21 …,P NM is the sampling value.

[0084] S5: Shuffle each row of A to obtain matrix B. Finally, each column of matrix B represents a sample point, and there are M sample points in total.

[0085] S6: Using parameter G max Evaluate whether the sample point distribution of matrix B is reasonable. If not, return and resample.

[0086] Among them, G max G is a parameter used to evaluate the uniform distribution of samples in multidimensional parameter space. It is usually calculated based on sample variance and sample standard deviation. Specifically, the sum of sample variance and three times sample standard deviation can be used as G max The calculation method of .

[0087] Define the function for evaluating sampling effect as G:

[0088] Where q represents a value in the matrix B. In a large number of experiments, the value of G is close to the normal distribution. The smaller G is, the more uniform the sampling is. Therefore, it is necessary to limit the upper limit of G during sampling, thereby limiting the sampling effect. The upper limit of G is G max .

[0089] Here, G is represented by the sum of three times the sample standard deviation. max Assume N = 10, LHS sampling 1000 times, assume that the number of samples M is usually larger than the object dimension N, calculate G under different M values max The values ​​are as follows, with the result rounded to two significant figures:

[0090]

[0091]

[0092] Fit the values ​​of M and N: lnG max =q+plnM

[0093] Where p and q are fitting coefficients. In contrast to the above process, a fixed set of G max and M, analyze the changes of p and q with N, where the result is retained to four significant figures:

[0094] N value p q N value p q 2 -1.154 2.863 12 -1.051 6.304 4 -1.093 4.202 14 -1.046 6.592 6 -1.069 4.954 16 -1.044 6.857 8 -1.062 5.523 18 -1.043 7.092 10 -1.058 5.971 20 -1.040 7.291

[0095] Also fit the values ​​of p and q:

[0096] p=-0.24N -0.94 -1.03

[0097] q=395.7N 0.0048 -394.1

[0098] Therefore, the parameter G max Calculated according to the sampling dimension N and the number of sampling times M:

[0099] oeLh max =(395.7N 0.0048 -394.1)-(0.24N -0.94 +1.03)lnM.

[0100] Finally, G is calculated based on the sampling dimension N and the number of sampling times M max , then according to G max Determine whether the matrix B is reasonable. If it is unreasonable, resampling is required.

[0101] Compute G directly from sample data max The method for evaluating sample quality based on actual data, and the embodiment of the present application uses N and M to calculate G max Based on the statistical analysis of a large number of simulation or experimental data of LHS sampling, the relationship between N, M and G is obtained. max This approach allows estimating the theoretical G before actually generating and evaluating samples. max The value can be used to guide the sampling process to ensure that the generated sample set is sufficiently uniform and representative.

[0102] The improved sampling method provided in this application generates representative samples from a large number of possible flight conditions and attribute information of airborne components. These samples represent a variety of different flight conditions and equipment states, aiming to fully cover possible operating environments. These samples not only cover a wide area of ​​the feature space, but also ensure the diversity of the data, which is very important for neural networks to learn complex patterns.

[0103] The improved Latin hypercube sampling method proposed in this application is based on a large number of previous analyses. The sampling level can be evaluated based on the number of samples and the dimension of the problem. This allows real-time monitoring of the uniformity of the samples, allowing dynamic adjustments during the sampling process to improve the spatial distribution of the samples, which helps improve the accuracy of model training and prediction.

[0104] The improved sampling method can ensure that the selected samples have good coverage and uniform distribution in all dimensions (such as temperature, height, speed, etc.), so that the trained prediction model can more accurately reflect the thermal behavior of components under various conditions. This is very important for designing more effective heat dissipation systems and predicting potential failure points in advance.

[0105] S103: Perform fluid dynamics simulation calculation based on the training data samples to obtain thermal characteristic calculation results of airborne components under different working conditions.

[0106] In this step, the specific process of performing fluid dynamics simulation calculation based on the training data samples to obtain the thermal characteristics calculation results of airborne components under different working conditions includes:

[0107] Collect information about the operating conditions of airborne components and determine the thermal characteristics to be studied. Collect information about the operating conditions of airborne components, including flight altitude, speed, ambient temperature, air pressure, etc. Identify the thermal characteristics that need to be studied, such as temperature distribution, heat flow, cooling efficiency, etc.

[0108] The geometric model of the airborne components is constructed so that the model reflects the shape and size of the actual components. Optionally, the geometric model can also be simplified as necessary to remove details that have little effect on the flow field and thermal field to reduce the computational complexity.

[0109] Generate computational mesh on the geometric model and set boundary conditions according to actual working conditions and environment.

[0110] The corresponding fluid dynamics and heat transfer models are selected for simulation calculation to obtain the thermal characteristics calculation results of airborne components under different working conditions.

[0111] S104: Using the attribute information of the aircraft's onboard components and the flight condition information as input data, using the thermal characteristic calculation results as output data, training the convolutional neural network model based on the training data samples, and generating a thermal prediction neural network model.

[0112] The structure of Convolutional Neural Networks (CNN) includes convolutional layer, excitation layer, pooling layer, and fully connected layer. Among them, the most important part is the convolutional layer, whose main function is to use the convolution kernel to calculate the matrix. For example, it can turn a 7×7 matrix convolution layer into a 5×5 matrix, thereby speeding up the operation, but at the same time it also reduces the output dimension.

[0113] In order to maintain the size of the output matrix, this application designs a thermal prediction neural network model based on the encoder-decoder architecture of the traditional CNN and the U-Net model, a variant of the fully convolutional neural network.

[0114] like Figure 2 As shown in the structural diagram of the convolutional neural network model, the convolutional neural network includes a component attribute encoder, a flight condition encoder, and a thermal calculation decoder.

[0115] Among them, the component attribute encoder is used to obtain the attribute information of the aircraft's onboard components; the flight condition encoder is used to obtain the aircraft's flight condition information; the thermal calculation decoder is used to sample the original historical data set using an improved Latin hypercube sampling method to obtain training data samples related to the heat prediction task; based on the training data samples, fluid dynamics simulation calculations are performed to obtain thermal characteristic calculation results of airborne components under different operating conditions.

[0116] The attribute information of the aircraft's onboard components includes the material information, volume information, and shape information of the components; the flight condition information includes the flight altitude information, speed information, overload information, and temperature information.

[0117] Unlike CNN, U-Net replaces the fully connected layer with a convolutional layer, so that the entire neural network is composed entirely of convolutional layers, pooling layers, and upsampling layers, so that the data or matrix only changes resolution during the neural network operation and transmission process. Since U-Net is a fully convolutional neural network, it is impossible to use a fully connected neural network to process the flight condition vector composed of flight altitude, speed, overload, temperature, etc. like CNN. Consider combining the row condition vector into a five-channel matrix as the input of the entire neural network.

[0118] like Figure 3 As shown in the schematic diagram of the U-Net model structure, U-Net includes five channels:

[0119] First channel: component attribute information encoded by the component attribute encoder;

[0120] Second channel: flying height information encoded by the component attribute encoder;

[0121] The third channel: speed information encoded by the component attribute encoder;

[0122] The fourth channel: overload information encoded by the component attribute encoder;

[0123] The fifth channel: the air temperature information encoded by the component attribute encoder;

[0124] With the component property encoder and the flight condition encoder as input and the thermal calculation decoder as output, U-Net is trained to generate a thermal prediction neural network model.

[0125] After the thermal prediction neural network model is generated, the thermal prediction neural network model is used to perform thermal prediction based on new input data, such as Figure 4 The schematic diagram of the process of thermal prediction neural network model performing thermal prediction based on new input data is shown in the figure. The process specifically includes:

[0126] S401: Obtaining attribute information of aircraft onboard components.

[0127] S402: Acquire the flight condition information of the aircraft at the predicted time in real time.

[0128] Obtain flight condition information, such as flight altitude, speed, overload and temperature. Specifically, it can combine data from trajectory planning, airspace planning and maneuver planning by integrating information from different sources, including scheduled flight plans, real-time flight data and meteorological data.

[0129] Among them, trajectory planning involves determining the best path for the aircraft from its starting point to its destination. During the trajectory planning process, the ideal flight altitude and speed for different stages (such as climb, cruise and descent) will be determined. These data are usually clearly stated in the flight plan and optimized according to the requirements of flight performance, fuel efficiency and air traffic control.

[0130] Airspace planning involves managing and optimizing the movement of aircraft within a specific area of ​​space to ensure safety and efficiency. Based on the volume and restrictions of the airspace, flight altitude and speed may need to be adjusted. Maneuvers such as sharp turns, climbs, or descents may result in momentary overloads. Congestion or avoidance measures in the airspace may require aircraft to perform these maneuvers.

[0131] Maneuver planning is usually related to the performance of the aircraft and special situations that may be encountered during flight. Specific maneuvers (such as obstacle avoidance or emergency descent) will directly affect the g-load data. The pilot or the automatic flight system executes these maneuvers as needed.

[0132] S403: Input the attribute information and flight condition information into the thermal prediction neural network model to generate temperature distribution data of aircraft onboard components at the prediction time.

[0133] Specifically, the U-Net model can be used to estimate in real time the heat rise of the altimeter during flight based on the aircraft's trajectory planning, airspace planning, and maneuvering planning, as well as the material, volume, and shape of the altimeter, thereby providing a reference for the overall design of the altimeter and even the aircraft.

[0134] Convolutional neural networks change the matrix size and affect the resolution of the data. To this end, this application proposes an encoder-decoder model full convolution U-Net model, which can integrate component properties, flight conditions and thermal simulation calculation results, and can be used to improve the design of airborne altimeters in a targeted manner.

[0135] In addition, the present application also provides a device for predicting the heating of aircraft-borne components, such as Figure 5 The present application provides a structural block diagram of an aircraft onboard component heating prediction device, which specifically includes a memory 51 and a processor 52. The memory 51 stores a computer program, and when the computer program is executed by the processor 52, it implements the aircraft onboard component heating prediction method according to any one of the above-mentioned methods.

[0136] In addition, the present application also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for predicting the heating of aircraft onboard components according to any one of the above-mentioned methods is implemented.

[0137] Computer readable storage media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0138] The professionals should also be further aware that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0139] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0140] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only the specific implementation method of the present application and is not intended to limit the scope of protection of the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A method for predicting heating of aircraft onboard components, characterized in that: include: Collecting original historical data sets of airborne components; The original historical data set is sampled using an improved Latin hypercube sampling method to obtain training data samples related to the fever prediction task; Perform fluid dynamics simulation calculation based on the training data samples to obtain thermal characteristic calculation results of airborne components under different working conditions; Using the attribute information of the aircraft's airborne components and the flight condition information as input data, using the thermal characteristic calculation results as output data, training the convolutional neural network model based on the training data samples, and generating a thermal prediction neural network model; The thermal prediction neural network model is used to perform thermal prediction based on new input data, and the process includes: Obtain the attribute information of aircraft onboard components; Acquiring in real time the flight condition information of the aircraft at the predicted time; The attribute information and flight condition information are input into the thermal prediction neural network model to generate temperature distribution data of aircraft onboard components at the prediction time.

2. The method for predicting heating of aircraft onboard components according to claim 1, characterized in that: The improved Latin hypercube sampling method is used to sample the original historical data set to obtain training data samples related to the fever prediction task, including: Assuming that the sampling object is N-dimensional, the sampling times are M, and 0-1 uniform sampling is performed on each dimension, the sampling process can be described as an N×M matrix, denoted as A; the improved Latin hypercube sampling method includes the following steps: S1: Divide each dimension M into equal parts, that is, the length of each interval is 1 / M; S2: Randomly select N values ​​from the first subinterval [0,1 / M] as the first column of matrix A; S3: Randomly select N values ​​from the second subinterval [1 / M, 2 / M] as the second column of matrix A; S4: And so on, until the last subinterval is reached and the Nth column of matrix A is formed; Among them, P 11 ,P 12 …,P 1M ,P 21 …,P NM is the sampling value; S5: Shuffle each row of A to obtain matrix B. Finally, each column of matrix B represents a sample point, and there are M sample points in total. S6: Using parameter G max Evaluate whether the sample point distribution of matrix B is reasonable. If it is unreasonable, return to resample. max is a parameter used to evaluate the uniform distribution of samples in the multidimensional parameter space.

3. The method for predicting heating of aircraft onboard components according to claim 2, characterized in that: The parameter G max Calculated according to the sampling dimension N and the number of sampling times M: <h2 style=";text-align:left;direction:ltr">lnG<h2 style=";text-align:left;direction:ltr"> max <h2 style=";text-align:left;direction:ltr"> (395.7N)<h2 style=";text-align:left;direction:ltr"> 0.0048 <h2 style=";text-align:left;direction:ltr"> -394.1)-(0.24N<h2 style=";text-align:left;direction:ltr"> -0.94 <h2 style=";text-align:left;direction:ltr"> +1.03)lnM。 4. The method for predicting heating of aircraft onboard components according to any one of claims 1 to 3, characterized in that: The convolutional neural network includes a component attribute encoder, a flight condition encoder, and a thermal calculation decoder; Among them, the component attribute encoder is used to obtain the attribute information of the aircraft's onboard components; the flight condition encoder is used to obtain the aircraft's flight condition information; the thermal calculation decoder is used to sample the original historical data set using an improved Latin hypercube sampling method to obtain training data samples related to the heat prediction task; based on the training data samples, fluid dynamics simulation calculations are performed to obtain thermal characteristic calculation results of airborne components under different operating conditions.

5. The method for predicting heating of aircraft onboard components according to claim 4, characterized in that: The attribute information of the aircraft-borne components includes material information, volume information, and shape information of the components; the flight condition information includes flight altitude information, speed information, overload information, and temperature information.

6. The method for predicting heating of aircraft onboard components according to claim 5, characterized in that: The convolutional neural network adopts the form of U-Net architecture, which includes five channels: First channel: used for inputting component attribute information encoded by the component attribute encoder; The second channel is used to input the flying height information encoded by the component attribute encoder; The third channel is used to input the speed information encoded by the component attribute encoder; The fourth channel is used to input the overload information encoded by the component attribute encoder; The fifth channel is used to input the air temperature information encoded by the component attribute encoder; With the component property encoder and the flight condition encoder as input and the thermal calculation decoder as output, U-Net is trained to generate a thermal prediction neural network model.

7. The method for predicting heating of aircraft onboard components according to any one of claims 1 to 3, characterized in that: The fluid dynamics simulation calculation based on the training data sample is performed to obtain the thermal characteristics calculation results of the airborne components under different working conditions, including: Collect information about the operating conditions of onboard components and determine the thermal characteristics to be studied; Construct geometric models of airborne components so that the models reflect the shapes and sizes of the actual components; Generate computational mesh on the geometric model and set boundary conditions according to actual working conditions and environment; The corresponding fluid dynamics and heat transfer models are selected for simulation calculation to obtain the thermal characteristics calculation results of airborne components under different working conditions.

8. The method for predicting heating of aircraft onboard components according to claim 4, characterized in that: The airborne component is an airborne radio altimeter.

9. An aircraft onboard component heating prediction device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the method for predicting the heating of aircraft onboard components according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the method for predicting the heating of aircraft onboard components according to any one of claims 1-8 is implemented.

Images