Aircraft bumping prediction method, system and equipment based on decoupling type physical constraint

Abstract

The invention relates to the technical field of aviation safety and weather prediction, in particular to an aircraft bumping prediction method, system and equipment based on decoupling type physical constraint. Aligning pre-collected multi-time-sequence weather forecast data and airplane EDR data on the same space-time grid, and converting the data into probabilistic features including a mean value and a standard deviation; for sparse airplane EDR data, introducing an atmospheric physical model correction term based on a wind shear and turbulence energy dissipation rate relationship for filling, and obtaining a probabilistic input tensor; inputting the probabilistic input tensor into a pre-constructed and trained self-correlation attention model fused with physical constraints; and extracting a contribution degree matrix from the attention weight of the trained model, quantifying the contribution ratio of each forecast time point to a prediction result, and outputting a probabilistic prediction result of the airplane bumping and the corresponding contribution degree matrix as a final prediction result. A reliable basis is provided for flight decision making, and the accuracy and robustness of airplane bumping prediction are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of aviation safety and meteorological prediction, and particularly relates to a method, system and device for predicting aircraft turbulence based on decoupled physical constraints. BACKGROUND

[0002] Aircraft turbulence is mainly caused by atmospheric turbulence, which poses a significant threat to flight safety, passenger comfort and operational efficiency. Traditional turbulence prediction methods mainly rely on numerical weather prediction (NWP) models and diagnostic models based on empirical rules. However, these methods have the following limitations: (1) NWP models require large computational resources and are difficult to provide real-time predictions with high temporal and spatial resolution; (2) empirical models lack accurate description of complex nonlinear atmospheric processes; (3) existing methods usually provide deterministic predictions and cannot quantify the uncertainty of predictions, making it difficult to support risk refinement and control decisions.

[0003] In recent years, with the development of artificial intelligence technology, especially the successful application of deep learning in time series prediction, a new approach to turbulence prediction has been provided. For example, models based on recurrent neural networks (RNN), long short-term memory networks (LSTM) and Transformers have been tried for weather prediction. Among them, the Autoformer model effectively captures long-term dependencies and periodic patterns in time series through self-correlation attention mechanisms, and performs well in time series prediction tasks.

[0004] However, direct application of purely data-driven AI models to aviation turbulence prediction faces great challenges. First, the EDR (turbulence energy dissipation rate) data reported by aircraft is extremely sparse in space and time, and simple data interpolation will introduce large errors and violate physical laws. Second, atmospheric motion follows strict physical laws (such as the Navier-Stokes equation), and purely data-driven models may learn false correlations that violate physical laws, resulting in physically unreliable predictions, especially poor generalization in areas with scarce training data. Finally, the aviation industry has extremely high requirements for safety and explainability, and the single prediction value provided by black box models cannot meet the needs of flight decision-making for risk confidence.

[0005] Therefore, there is an urgent need for a new aircraft turbulence prediction method that can integrate physical prior knowledge, handle data uncertainty, and provide explainable prediction results. SUMMARY

[0006] In order to overcome the deficiencies in the prior art, the present application provides a kind of based on decoupling physical constraint aircraft jolt prediction method, system and equipment. By embedding atmospheric physical law in decoupling way into deep learning model, the effective combination of physical guidance and data-driven learning is realized, can output probabilistic prediction results and explainable contribution degree analysis, significantly improve the accuracy of prediction, physical rationality and decision support capability.

[0007] The above invention aims to adopt the following technical solutions: In a first aspect, the present application provides a kind of based on decoupling physical constraint aircraft jolt prediction method, the method comprises: Aligning the pre-acquired multi-time series weather forecast data and aircraft EDR data on the same space-time grid, and converting into probabilistic features containing mean and standard deviation; For sparse aircraft EDR data, introduce atmospheric physical model correction term based on wind shear and turbulence energy dissipation rate relationship to fill, obtain probabilistic input tensor; The probabilistic input tensor is input into the pre-established self-correlation attention model fused with physical constraints to carry out aircraft jolt probabilistic prediction, and Gaussian negative log likelihood is used as loss function to train the model; Wherein, the self-correlation attention model fused with physical constraints includes encoder, attention layer and decoder; The encoder converts the probabilistic input tensor into one or more of query matrix, key matrix and value matrix through linear mapping for self-correlation mapping; The attention layer calculates the cross-correlation function based on the query matrix and the key matrix, constructs the attention score matrix, and introduces decoupling physical constraint, the decoupling physical constraint includes global physical attenuation matrix calculated according to time difference and spherical distance and trainable local fine-tuning matrix for capturing local weather deviation; The decoder takes the physical constraint self-correlation feature sequence as input, and outputs the probabilistic prediction results of future multiple time steps containing EDR mean and standard deviation; Contribution degree matrix is extracted from the attention weight of the trained model, the contribution proportion of each prediction time point to the prediction result is quantified, and the probabilistic prediction results and the corresponding contribution degree matrix of the aircraft jolt are output as the final result of prediction.

[0008] Optionally, the aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid and converting the multi-temporal weather forecast data and the aircraft EDR data into a probability feature including a mean value and a standard deviation comprises: converting a time stamp of the multi-temporal weather forecast data into a UTC time; taking a recording time of the aircraft EDR data as a reference, matching weather forecast data with a time difference less than a preset threshold for each EDR record point; using spherical linear interpolation on the matched weather forecast data, interpolating a spatial resolution of the weather forecast data to a latitude and longitude position of the EDR record point to obtain aligned original data; and performing probability feature representation on the aligned original data to obtain a probability feature vector on each spatio-temporal grid point. The probability feature vector includes an EDR mean component, an EDR standard deviation component, a weather mean component, and a weather standard deviation component.

[0009] Optionally, the probability feature representation on the aligned original data to obtain a probability feature vector on each spatio-temporal grid point comprises: obtaining the EDR mean component and the EDR standard deviation component by calculating a mean value and a standard deviation of the aircraft EDR data within a preset sliding time window; obtaining the weather mean component and the weather standard deviation component by calculating a mean value and a standard deviation of a weather variable at each forecast time point within the same sliding time window. The EDR mean component, the EDR standard deviation component, the weather mean component, and the weather standard deviation component are spliced to form the probability feature vector.

[0010] Optionally, for sparse aircraft EDR data, a physical model correction term based on the relationship between wind shear and turbulent energy dissipation rate is introduced to fill in to obtain a probability input tensor, which comprises: The probability feature vector of the known point is interpolated to obtain an interpolation mean value and an interpolation standard deviation. The unknown point probability feature vector with missing EDR values is filled in based on the correction term of the atmospheric physical model; wherein the correction term of the atmospheric physical model includes the physical relationship between wind shear and turbulent energy dissipation rate. The Richardson number is calculated according to the wind shear and the temperature gradient, and the Richardson number is mapped to a turbulent energy dissipation rate estimate; the turbulent energy dissipation rate estimate is superimposed on the interpolation mean value, and based on the interpolation mean value and the interpolation standard deviation of the superimposed turbulent energy dissipation rate estimate, a probability input tensor is obtained.

[0011] Optionally, the training of the self-correlation attention model fused with physical constraints comprises: A global physical decay matrix is defined based on the time and spatial exponential decay law of atmospheric vortices. A local fine-tuning matrix for capturing local weather bias is established. An Autoformer is taken as a basic framework, a global physical attenuation matrix and a local fine-tuning matrix are taken as decoupled physical constraints, and a self-correlation attention mechanism is improved to construct a self-correlation attention model fusing physical constraints. In the model training stage, the model parameters are optimized through back propagation and gradient update to obtain the trained self-correlation attention model fusing physical constraints.

[0012] Optionally, the definition of the global physical attenuation matrix based on the time and space exponential attenuation law of atmospheric vortices comprises: reading the time difference and spherical distance between any two points in the space-time grid; dividing the time difference by a preset time attenuation constant, dividing the spherical distance by a preset space attenuation constant, adding them together, and taking the negative exponential to obtain the global physical attenuation matrix element.

[0013] Optionally, the optimization of the model parameters through back propagation and gradient update in the model training stage comprises: initializing the local fine-tuning matrix as a full zero matrix, in each iteration training, calculating the attention weight through forward propagation, and calculating the Gaussian negative log-likelihood loss by combining the real EDR value, calculating the gradient of the Gaussian negative log-likelihood loss on the full zero matrix through back propagation, and applying gradient update to the elements of the full zero matrix.

[0014] Optionally, the extraction of the contribution matrix from the attention weight, the quantification of the contribution proportion of each prediction time point to the prediction result, and the output of the probabilistic prediction result and the contribution matrix of the aircraft jolt comprise: grouping and summing the attention weight according to the prediction time point index to obtain the total attention score of each prediction time point; calculating the proportion of the total attention score of each prediction time point in the total attention score of all features to obtain the normalized contribution proportion; filling the normalized contribution proportion into the corresponding position of the contribution matrix to generate the contribution matrix; each element of the contribution matrix represents the relative contribution proportion of the corresponding prediction time point to the prediction result in the prediction time step; and outputting the contribution matrix and the probabilistic prediction result.

[0015] In a second aspect, the present application provides an aircraft jolt prediction system based on decoupled physical constraints, which comprises: The integration module is used for aligning the pre-acquired multi-time series weather forecast data and aircraft EDR data on the same space-time grid, and converting them into probabilistic features containing mean and standard deviation; The data filling module is used for filling the atmospheric physical model correction term based on the relationship between wind shear and turbulent energy dissipation rate for sparse aircraft EDR data to obtain a probabilistic input tensor; The model training module is used for inputting the probabilistic input tensor into the self-correlation attention model fusing physical constraints which is pre-constructed and trained. The result output module is configured to extract a contribution matrix from the attention weight of the trained model, quantify the contribution proportion of each prediction time point to the prediction result, and output the probabilistic prediction result of the aircraft turbulence and the corresponding contribution matrix as the final prediction result.

[0016] In a third aspect, the present application provides an electronic device, comprising: at least one processor; and a memory in communication with the at least one processor; wherein The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the method of any one of the first aspect.

[0017] Compared with the closest prior art, the present application has the beneficial effects of: The aircraft turbulence prediction method, system and device based on decoupled physical constraints proposed by the present application introduce decoupled physical constraints (global fixed matrix + local adjustable matrix), directly embed the physical law of atmospheric vortex decay into the learning process of the model, ensure that the model prediction follows the basic physical law, effectively avoid the physical absurd results that may be produced by the pure data-driven model, and significantly improve the prediction accuracy in the data sparse area and extreme situation.

[0018] The present application uses probabilistic features as input and output, and uses a Gaussian negative log-likelihood loss function for training, so that the model can spontaneously learn and output the prediction uncertainty (standard deviation). This provides more rich risk information for flight decision, supports more refined risk assessment and management.

[0019] The present application uses a filling method based on physical prior to process the sparse EDR data, which can generate more physically realistic data than traditional mathematical interpolation, providing a higher quality training basis for the model, thereby improving the overall performance.

[0020] The present application can clearly quantify the contribution proportion of information of different prediction time to the final prediction result by extracting the contribution matrix from the attention mechanism optimized by the physical constraint. This interpretability enables pilots and forecasters to understand the basis of the model's decision, enhancing the trust in the prediction result.

[0021] In the present application, the decoupled design fixes the prior knowledge in the global matrix, providing a stable optimization direction for the model and alleviating the gradient instability problem in long sequence prediction. At the same time, the local adjustable matrix gives the model flexibility to adapt to complex local phenomena (such as terrain effect, thunderstorm), enhancing the generalization ability of the model. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings required to be used in the specific embodiments or prior art description will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference signs. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0023] Figure 1 is a flow chart of an aircraft bump prediction method based on decoupled physical constraints provided by the present application; Figure 2 is a structural schematic diagram of an aircraft bump prediction system based on decoupled physical constraints provided by the present application; Figure 3 is an internal structure diagram of an electronic device provided by the present application. DETAILED DESCRIPTION

[0024] The embodiments of the technical solutions of the present application will be described in detail below with reference to the drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present application, and therefore only serve as examples, and cannot limit the protection scope of the present application.

[0025] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in the present application should be understood as the usual meanings understood by the skilled in the art to which the present application belongs.

[0026] The present application provides an aircraft bump prediction method, system and device based on decoupled physical constraints. The multi-time sequence weather data and aircraft EDR (flight data recorder) data are integrated, and not only a single value is used, but the data is converted into a probabilistic feature containing the mean and standard deviation. This processing method also takes into account the inherent uncertainty of the data in the model, making the prediction result more robust. In addition, for sparse or missing EDR data, prior knowledge of atmospheric physics is used for filling, rather than simple mathematical interpolation. It is ensured that the input data follows the basic physical laws from the beginning, laying a solid physical foundation for the subsequent model learning.

[0027] Secondly, the autocorrelation attention mechanism of Autoformer is upgraded by introducing a decoupled physical constraint. This constraint consists of two parts: a fixed global physical decay matrix and a trainable local fine-tuning matrix. The global matrix is based on the exponential decay physical law of atmospheric vorticity, ensuring that the model always follows the physical common sense that "the closer the distance, the greater the impact". The local fine-tuning matrix allows the model to adaptively capture physical deviations caused by local terrain or sudden weather events. This decoupled design not only solves the problem of gradient instability in long sequence model training, but also perfectly combines the rigor of physics and the flexibility of data-driven models, significantly improving the accuracy and physical reasonableness of the prediction. Finally, the model output not only includes the predicted value, but also the uncertainty (standard deviation) and the interpretable weight matrix, providing transparent and comprehensive basis for flight decision-making.

[0028] Embodiment 1: Please refer to Figure 1 , Figure 1 A method for predicting aircraft turbulence based on decoupled physical constraints is provided for Embodiment 1 of the present application, which specifically includes the following steps: S101 align the pre-collected multi-time series weather forecast data and aircraft EDR data on the same spatio-temporal grid, and convert them into probabilistic features containing mean and standard deviation; S102 For sparse aircraft EDR data, introduce an atmospheric physical model correction term based on the relationship between wind shear and turbulent energy dissipation rate to fill in, obtain a probabilistic input tensor; S103 input the probabilistic input tensor into the self-correlation attention model with fusion physical constraints pre-established to perform probabilistic prediction of aircraft turbulence, and use Gaussian negative log-likelihood as the loss function to train the model; S104 extract the contribution matrix from the attention weight of the trained model, quantify the contribution proportion of each prediction time point to the prediction result, and output the probabilistic prediction result of the aircraft turbulence and the corresponding contribution matrix as the final result of the prediction.

[0029] In the above embodiment, step S101 specifically includes: Convert the timestamp of the multi-time series weather forecast data to UTC time.

[0030] Take the recording time of the aircraft EDR data as the benchmark, and match the weather forecast data with a time difference less than the preset threshold for each EDR record point; The matched weather forecast data is interpolated using spherical linear interpolation to interpolate the spatial resolution of the weather forecast data to the latitude and longitude position of the EDR record point, obtaining the aligned original data.

[0031] The aligned original data is subjected to a probabilistic feature representation to obtain a probabilistic feature vector at each space-time grid point; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a meteorological mean component, and a meteorological standard deviation component.

[0032] Further, the probabilistic feature representation is realized by calculating the mean and standard deviation of the aircraft EDR data within a preset sliding time window to obtain the EDR mean component and the EDR standard deviation component; calculating the mean and standard deviation of the meteorological variable at each forecast time point within the same sliding time window to obtain the meteorological mean component and the meteorological standard deviation component; and splicing the above four components to form a probabilistic feature vector.

[0033] Wherein, the meteorological forecast data: for each forecast time point Its meteorological features (such as wind speed, wind shear) are also input in the form of mean and standard deviation .

[0034] In the above step S102, for sparse aircraft EDR data, an atmospheric physics model correction term based on the relationship between wind shear and turbulent energy dissipation rate is introduced for filling to obtain a probabilistic input tensor, which comprises: The probabilistic feature vector of the known point is subjected to interpolation calculation (such as Gaussian process regression or Kriging interpolation) to obtain an interpolation mean and an interpolation standard deviation.

[0035] Based on the correction term of the atmospheric physics model, the probabilistic feature vector of the unknown point with missing EDR value is filled. The correction term is constructed based on the physical relationship between wind shear and turbulent energy dissipation rate.

[0036] Specifically, the Richardson number is calculated according to the wind shear and the temperature gradient, and the Richardson number is mapped to an estimated value of the turbulent energy dissipation rate.

[0037] The estimated value of the turbulent energy dissipation rate is superimposed on the interpolation mean, and the interpolation standard deviation is retained, to jointly constitute the probabilistic feature vector of the unknown point, and finally form a complete probabilistic input tensor: .

[0038] Wherein, B is the batch size (Batch Size); L is the historical sequence length (Sequence Length). is the enhanced feature dimension (Feature Dimension). The input vector of each time step t contains EDR historical data: mean and standard deviation .

[0039] In step S103, the probabilistic input tensor is input into the pre-established self-correlation attention model with physical constraints to perform aircraft jolt probability prediction, and a Gaussian negative log-likelihood is used as a loss function to train the model. The self-correlation attention model with physical constraints includes an encoder, an attention layer, and a decoder. The encoder converts the probabilistic input tensor into one or more of a query matrix, a key matrix, and a value matrix through linear mapping for self-correlation mapping. Optionally, the decoder includes: a first linear layer that maps the physical constraint self-correlation feature sequence into a hidden state. A second linear layer maps the hidden state into an EDR mean prediction vector and an EDR standard deviation prediction vector.

[0040] The attention layer calculates a cross-correlation function based on the query matrix and the key matrix, constructs an attention score matrix, and introduces a decoupled physical constraint, which includes a global physical attenuation matrix calculated according to a time difference and a spherical distance and a trainable local fine-tuning matrix for capturing local meteorological bias. The decoder takes the physical constraint self-correlation feature sequence as input and outputs probabilistic prediction results containing EDR mean and standard deviation at multiple future time steps.

[0041] In the above embodiment, determining the cross-correlation function of the query matrix and the key matrix includes: Performing a fast Fourier transform on each column of the query matrix; Performing a fast Fourier transform on each column of the key matrix and taking a conjugate; Multiplying the transformed elements of the query matrix and the key matrix, and performing an inverse fast Fourier transform to obtain the cross-correlation function.

[0042] In one embodiment, the encoder converts the probabilistic input tensor into a query matrix Q, a key matrix K, and a value matrix V through linear mapping.

[0043] In one embodiment, the attention layer is the core innovation of the model, which introduces a decoupled physical constraint including a global physical attenuation matrix and a local fine-tuning matrix Its calculation process includes: Based on the query matrix Q and the key matrix K, the cross-correlation function is calculated through fast Fourier transform (FFT) and inverse transform (IFFT) to obtain an initial attention score matrix.

[0044] The initial score matrix is added to the scaled global physical attenuation matrix and the local fine-tuning matrix ​

[0045] The added matrix is normalized by SoftMax to obtain the final attention weight. The value matrix V is weighted and summed using the attention weight to obtain the physical constraint autocorrelation feature sequence.

[0046] Global physical attenuation matrix Based on the time and space exponential attenuation law of atmospheric vortex, the calculation formula is: ; wherein, is the time difference between t i and t j ; is the spherical distance between the corresponding geographical positions; is the time attenuation constant, reflecting the average life cycle of atmospheric vortex in time, which can be initialized according to empirical values or atmospheric physical models; is the spatial attenuation constant, reflecting the average influence range of atmospheric vortex in space, which is also initialized by physical priori.

[0047] In the above embodiment, the global physical attenuation matrix defined based on the time and space exponential attenuation law of atmospheric vortex includes: Read the time difference and spherical distance between any two points in the space-time grid; Divide the time difference by the preset time attenuation constant and the spherical distance by the preset spatial attenuation constant, add them and take the negative exponential to obtain the global physical attenuation matrix element.

[0048] In the above embodiment, the local fine-tuning matrix is initialized as a zero matrix, and is updated by gradient descent algorithm during model training to capture local deviations other than global physical laws.

[0049] The decoder takes the physical constraint autocorrelation feature sequence output by the attention layer as input, and outputs the prediction result of the future O time steps through forward propagation. The result includes the EDR mean and standard deviation .

[0050] Further, the training of the autocorrelation attention model fused with physical constraints in step S103 includes: Based on the time and space exponential attenuation law of atmospheric vortex, the global physical attenuation matrix is defined; Establish a local fine-tuning matrix for capturing local meteorological deviations; Take Autoformer as the basic architecture, and take the global physical attenuation matrix and the local fine-tuning matrix as the decoupled physical constraints to improve the autocorrelation attention mechanism and build the autocorrelation attention model fused with physical constraints; In the model training stage, the model parameters are optimized by back propagation and gradient update, and a trained self-correlation attention model with physical constraints is obtained.

[0051] Further, in the model training stage, optimizing the model parameters by back propagation and gradient update includes: initializing the local fine-tuning matrix as a full zero matrix, in each iteration training, calculating the attention weight by forward propagation, and calculating the Gaussian negative log-likelihood loss by combining the real EDR value, calculating the gradient of the Gaussian negative log-likelihood loss on the full zero matrix by back propagation, and applying gradient update to the elements of the full zero matrix.

[0052] The model is trained using Gaussian negative log-likelihood as the loss function L:

[0053] Wherein, is the real EDR value. This loss function drives the model to optimize both the accuracy of the prediction and the uncertainty estimation.

[0054] In the step S104, the contribution matrix is extracted from the attention weight, the contribution proportion of each prediction time point to the prediction result is quantified, and the probabilistic prediction result of the aircraft jolt and the contribution matrix are outputted, including: grouping and summing the attention weight according to the prediction time point index to obtain the total attention score of each prediction time point; calculating the proportion of the total attention score of each prediction time point in the total attention score of all features to obtain the normalized contribution proportion; filling the normalized contribution proportion into the corresponding position of the contribution matrix to generate the contribution matrix; each element of the contribution matrix represents the relative contribution proportion of the corresponding prediction time point to the prediction result in the prediction time step; the contribution matrix and the probabilistic prediction result are outputted.

[0055] Specifically, the contribution matrix is extracted :

[0056] In the formula, is the attention matrix optimized by physical constraints; indices j contains all the indexes of the features related to the j j th prediction time point in the input; intuitively represents the relative contribution of the j th prediction time point when predicting the i j th time step.

[0057] The final attention weight matrix is grouped according to the prediction time point index corresponding to the input feature. The attention weights on all feature indexes k belonging to the same prediction time point j are summed:

[0058] In the above formula, λ1 and λ2 are adjustable hyperparameters for controlling the weights of the global physical constraint term and the local fine-tuning term in the attention mechanism, The matrix directly participates in the calculation of the attention weight as an additive term; dividing the sum by the sum of the attention weights on all feature indexes, the normalized contribution ratio is obtained. The contribution matrix intuitively shows the weight of the prediction information at different prediction time points (such as 3 hours, 12 hours in advance) when predicting the i-th time step, providing interpretability for the prediction results.

[0059] Embodiment 2: The second embodiment of the present application is based on the above method, and proposes a method of a probabilistic autocorrelation attention model based on decoupled physical constraints in aircraft turbulence prediction. The method mainly includes data preprocessing, model construction, model training and prediction output aspects.

[0060] 1. In the aspect of data preprocessing, the embodiment is executed through the following 3 steps: Data collection and alignment: Collecting multi-time series meteorological data of a certain route in the past year, including wind speed, wind shear, temperature gradient, etc., with a data time interval of 1 hour; At the same time, collect the EDR data of the corresponding flights of the route, with a data time interval of 1 hour. Align the meteorological data and EDR data accurately according to the time stamp and geographical location, and ensure that each time step of meteorological data matches the corresponding EDR data.

[0061] Probabilistic feature construction: For EDR historical data, calculate the mean and standard deviation of each time step. For example, for time step t=1, the mean and standard deviation of the EDR data of the 10 adjacent time steps before and after this time step are calculated. For meteorological data, for each prediction time point (such as 3 hours, 6 hours, 12 hours, 24 hours in advance), calculate the mean and standard deviation of its meteorological data to form an input tensor; Where, batch size B=32, history sequence length L=24 (i.e. using the data of the past 24 hours), enhanced feature dimension D'=16 (including the mean and standard deviation of EDR, and the mean and standard deviation of wind speed, wind shear, and temperature gradient of 4 prediction time points).

[0062] Sparse data filling: For part of the missing EDR data, fill it in by the formula:

[0063] Where, ​Advanced statistical methods such as Gaussian process regression (GPR) or Kriging interpolation are used to interpolate based on known EDR data; F is based on the correction term of atmospheric physical models (such as Richardson number), and the physical estimate of turbulent energy dissipation rate is calculated based on wind shear and temperature gradient to correct the interpolation results.

[0064] 2. In terms of model construction, a global physical attenuation matrix is ​​constructed according to the formula. Calculate, where the time decay constant is... Based on the empirical value of the atmospheric eddy current time lifespan being set at 6 hours, the spatial decay constant is... Based on the empirical value of the spatial influence range of atmospheric eddies, it is set to 100 kilometers, and Δtᵢⱼ is the time difference (in hours) between time steps i and j. Δlocᵢⱼ represents the spherical distance (in kilometers) between the corresponding geographical locations. The local fine-tuning matrix is ​​initialized as a 24×24 zero matrix.

[0065] Autocorrelated attention is calculated using the following formula: .

[0066] Where Q, K, and V are derived from the probabilistic input tensor X enc The query matrix Q, key matrix K, and value matrix V obtained after linear transformation, and the model hidden dimension d. model =128, hyperparameters λ1=0.5, λ2=0.3.

[0067] Auto-correlation Q, K The calculation of ) is achieved through FFT transformation. First, FFT transformations are performed on Q and K respectively. The conjugate of the FFT result of K is then multiplied with the Hadamard product of the FFT result of Q, and finally, the inverse FFT transformation is performed to obtain: .

[0068] In this embodiment, the Autoformer model is used as the model network, which contains a linear layer (input dimension 2, output dimension 16) and a ReLU activation function. The mean μ and the logarithm log(σ) of the standard deviation σ of each feature are taken as input, and after linear transformation and activation function, the fused vector representation h is obtained. t .

[0069] in, This indicates that taking the logarithm can avoid numerical instability and ensure that the standard deviation is positive.

[0070] In the above embodiment, the model decoder includes two independent linear layers, each with an input dimension of 128 and an output dimension of 1, for predicting the mean and standard deviation of the EDR, forming an output tensor with a predicted sequence length O = 12 (i.e., predicting the heave situation 12 hours into the future).

[0071] 3. In terms of model training, the following steps are included: The local fine-tuning matrix is initialized as a 24x24 zero matrix, and the other parameters of the model are initialized using the Xavier initialization method.

[0072] Forward propagation: In each training iteration, a batch (32 samples) of probabilistic input tensors X enc is input into the model, and the fused feature vectors are obtained through the probabilistic data fusion module, then the attention weight is calculated, and finally the prediction result is output through the decoder.

[0073] Loss calculation: According to the predicted mean , standard deviation and real EDR value output by the model, the loss is calculated using the Gaussian negative log-likelihood loss function, with the formula: .

[0074] Backpropagation and gradient update: Using the Adam optimizer with a learning rate η = 0.001, the loss function is backpropagated, the gradient of the loss with respect to all parameters of the model (including the local fine-tuning matrix) is calculated, and the parameters are updated according to the gradient. The number of training iterations is set to 1000, and the validation set loss is calculated after each iteration. When the validation set loss does not decrease for 10 consecutive iterations, the training is stopped.

[0075] 4. In terms of prediction output, the preprocessed test data (batch size B = 32, historical sequence length L = 24) is input into the trained model, and the model outputs the predicted mean and standard deviation of EDR for the next 12 hours, such as "the mean heave index for the next 1 hour is 4.8, and the standard deviation is 0.3; the mean heave index for the next 2 hours is 5.1, and the standard deviation is 0.4", etc. At the same time, an interpretable contribution matrix is generated:

[0076] where is the attention matrix optimized by physical constraints; contains the indices of all features related to the jth prediction time point in the input; Intuitively, it is shown that the relative contribution of the jth prediction time point when predicting the ith time step.

[0077] For example, "when predicting the 1-hour future jolt, the 3-hour-ahead prediction weight is 65%, the 6-hour-ahead prediction weight is 20%, the 12-hour-ahead prediction weight is 10%, and the 24-hour-ahead prediction weight is 5%", which provides a reference for flight decision-making.

[0078] Based on the same inventive concept, the embodiments of the present application also provide a decoupled physical constraint-based aircraft jolt prediction system for implementing the above-mentioned decoupled physical constraint-based aircraft jolt prediction method. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme described in the above-mentioned embodiment method, so the specific limitations in one or more decoupled physical constraint-based aircraft jolt prediction system embodiments provided below can be referred to the limitations of the decoupled physical constraint-based aircraft jolt prediction method described above, which will not be repeated here.

[0079] In one embodiment, the decoupled physical constraint-based aircraft jolt prediction system provided by the embodiments of the present application can be deployed on the server of the air traffic control center in actual application, and can receive meteorological data and aircraft EDR data in real time, input the data after data preprocessing into the model for prediction, and output the prediction results in time to provide protection for flight safety. As shown in Figure 2 The system includes an integration module 210, a data filling module 220, a model training module 230, and a result output module 240, wherein: The integration module 210 is configured to align the pre-acquired multi-time series meteorological prediction data and aircraft EDR data on the same space-time grid, and convert the data into probability features containing mean and standard deviation; The data filling module 220 is configured to fill the sparse aircraft EDR data by introducing an atmospheric physical model correction term based on the relationship between wind shear and turbulent energy dissipation rate to obtain a probability input tensor; The model training module 230 is configured to input the probability input tensor into a self-correlation attention model with fused physical constraints which is pre-constructed and trained; The result output module 240 is configured to extract a contribution matrix from the attention weight of the trained model, quantify the contribution proportion of each prediction time point to the prediction result, and output the probability prediction result of the aircraft jolt and the corresponding contribution matrix as the final prediction result.

[0080] Embodiment 4: The fourth embodiment of the present application is based on the technical principle of the above-mentioned aircraft bump prediction system based on decoupled physical constraints, and provides a system for aircraft bump prediction based on a decoupled physical constraint probabilistic self-correlation attention model, which includes a data preprocessing module, a model construction module, a model training module and a prediction output module.

[0081] Among them, the data preprocessing module adopts Python programming language and carries out data processing based on Pandas library. First, multi-time series meteorological data and aircraft EDR data are obtained through a data interface, and the data is aligned according to the timestamp and geographical position by using the merge function of Pandas. Then, the mean and standard deviation of each data are calculated using the numpy library to construct probabilistic features. For sparse or missing EDR data, the Gaussian process regression model in the Scikit-learn library is called to realize the function of G function, and F function is written based on atmospheric physical formula to realize data filling.

[0082] The model construction module is built based on the PyTorch deep learning framework. In the self-correlation attention calculation module, the nn.Linear layer of PyTorch is used to realize the linear transformation of Q, K and V, and the FFT transformation and inverse transformation are realized by torch.fft module to calculate the autocorrelation. The global physical attenuation matrix is calculated and generated in Python according to the preset parameters and formula, and the local fine-tuning matrix is initialized by nn.Parameter and set requires_grad=True to realize training. The probabilistic data fusion module is composed of nn.Linear layer and nn.ReLU activation function, and the uncertainty quantification module is realized by two independent nn.Linear layers.

[0083] The model training module is also based on the PyTorch framework, and uses torch.optim.Adam optimizer to write training loop code according to the training steps in embodiment 1 to realize forward propagation, loss calculation, back propagation and parameter update. In the training process, TensorBoard is used to record the training loss and validation loss, which is convenient for monitoring the model training process.

[0084] The prediction output module converts the prediction results and weight matrix of the model into a text format easy to understand, builds a Web interface through the Flask framework, and outputs the results to the front-end page for pilots or relevant staff to view. At the same time, the results can be stored in the database for subsequent analysis and query.

[0085] In one embodiment, an electronic device, which can be a terminal, is provided, and an internal structure diagram of the electronic device can be as shown in Figure 3As shown in the figure. The electronic device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the electronic device is used for wired or wireless communication with external terminals. Wireless communication can be achieved through WIFI, mobile cellular network, NFC (near field communication) or other technologies. The computer program is executed by the processor to implement any one of the steps S101 to S104 of the aircraft bump prediction method based on decoupled physical constraints. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen. The input device of the electronic device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the electronic device, or an external keyboard, touchpad or mouse, etc.

[0086] Those skilled in the art can understand that Figure 3 The skilled in the art can understand that

[0087] Those skilled in the art can understand that

[0088] The embodiments of the present application can be provided as a method, system or computer program product. Therefore, the present application can be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code. Figure 1 Figure 1 The embodiments of the present application can be provided as a method, system or computer program product. Therefore, the present application can be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code. The embodiments of the present application can be provided as a method, system or computer program product. Therefore, the present application can be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0089] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flow Figure 1 The flow or flows and / or blocks Figure 1 The function specified in the flow or flows and / or block or blocks.

[0090] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions that execute on the computer or other programmable apparatus provide steps for implementing the flow Figure 1 The flow or flows and / or blocks Figure 1 The function specified in the flow or flows and / or block or blocks.

[0091] The above merely provides an embodiment of the present application, but is not intended to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the scope of the claims of the present application.

Claims

1. A method for predicting aircraft bump based on decoupled physical constraints, characterized in that, The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; for sparse aircraft EDR data, introducing an atmospheric physics model correction term based on the relationship between wind shear and turbulent energy dissipation rate to fill in, obtaining a probabilistic input tensor; inputting the probabilistic input tensor into a pre-constructed and trained self-correlation attention model with physical constraints; wherein the self-correlation attention model with physical constraints comprises an encoder, an attention layer and a decoder; the encoder converts the probabilistic input tensor into one or more of a query matrix, a key matrix and a value matrix through linear mapping for self-correlation mapping; the attention layer calculates the cross-correlation function based on the query matrix and the key matrix, constructs an attention score matrix, and introduces a decoupled physical constraint, which includes a global physical attenuation matrix calculated according to the time difference and the spherical distance, and a trainable local fine-tuning matrix for capturing local weather bias; the decoder takes the physical constraint self-correlation feature sequence as input, and outputs the probabilistic prediction results of the future multiple time steps containing the EDR mean and standard deviation; extracting a contribution matrix from the attention weight of the trained model, quantifying the contribution proportion of each prediction time point to the prediction result, and outputting the probabilistic prediction result of the aircraft turbulence and the corresponding contribution matrix as the final result of the prediction.

2. The method of claim 1, wherein, The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; 3. The method of claim 2, wherein, wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises:

4. The method of claim 1, wherein, aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre-collected multi-temporal weather forecast data and the aircraft EDR data on the same spatio-temporal grid, and converting into probabilistic features containing mean and standard deviation; wherein the probabilistic feature vector comprises an EDR mean component, an EDR standard deviation component, a weather mean component and a weather standard deviation component. The method comprises: aligning the pre Interpolation is performed on the probabilistic feature vectors of the known points to obtain an interpolation mean and an interpolation standard deviation; The probabilistic feature vectors of the unknown points with missing EDR values are filled based on a correction term of an atmospheric physical model, wherein the correction term of the atmospheric physical model includes a physical relationship between wind shear and turbulent energy dissipation rate; The Richardson number is calculated according to the wind shear and the temperature gradient, and the Richardson number is mapped to a turbulent energy dissipation rate estimate; the turbulent energy dissipation rate estimate is superimposed on the interpolation mean, and based on the interpolation mean and the interpolation standard deviation of the superimposed turbulent energy dissipation rate estimate, a probabilistic input tensor is obtained.

5. The method of claim 1, wherein, The training of the self-correlation attention model fused with physical constraints includes: Based on the time and space exponential decay law of atmospheric vortices, a global physical decay matrix is defined; A local fine-tuning matrix for capturing local meteorological bias is established; The Autoformer is taken as a basic architecture, and the global physical decay matrix and the local fine-tuning matrix are taken as decoupled physical constraints to improve the self-correlation attention mechanism, thereby constructing a self-correlation attention model fused with physical constraints; In the model training stage, the model parameters are optimized through back propagation and gradient update to obtain the trained self-correlation attention model fused with physical constraints.

6. The method of claim 5, wherein, The definition of the global physical decay matrix based on the time and space exponential decay law of atmospheric vortices includes: The time difference and the spherical distance between any two points in the space-time grid are read; The time difference is divided by a preset time decay constant, the spherical distance is divided by a preset space decay constant, and the negative exponential of the sum is obtained to obtain the elements of the global physical decay matrix.

7. The method of claim 5, wherein, In the model training stage, the model parameters are optimized through back propagation and gradient update, including: initializing the local fine-tuning matrix as a full zero matrix, in each iteration training, the attention weight is calculated through forward propagation, and the Gaussian negative log-likelihood loss is calculated combined with the real EDR value, the gradient of the Gaussian negative log-likelihood loss to the full zero matrix is calculated through back propagation, and the gradient update is applied to the elements of the full zero matrix.

8. The method of claim 7, wherein, The contribution matrix is extracted from the attention weight of the trained model, the contribution proportion of each prediction time point to the prediction result is quantified, and the probabilistic prediction result of the aircraft turbulence and the corresponding contribution matrix are output as the final prediction result, including: grouping and summing the attention weight according to the prediction time point index to obtain the total attention score of each prediction time point; calculate the proportion of the total attention score of each prediction time point in the total attention score of all features to obtain the normalized contribution proportion; fill the normalized contribution proportion into the corresponding position of the contribution matrix to generate the contribution matrix; each element of the contribution matrix represents the relative contribution proportion of the corresponding prediction time point to the prediction result in the prediction time step; and the contribution matrix and the probabilistic prediction result are output.

9. A decoupled physical constraint based aircraft bump prediction system, characterized in that, The system includes: The integration module is used to align the pre-acquired multi-time series meteorological prediction data and aircraft EDR data on the same space-time grid, and convert them into probabilistic features containing mean and standard deviation; The data filling module is configured to introduce an atmospheric physical model correction term based on a relationship between wind shear and turbulent energy dissipation rate to fill the sparse aircraft EDR data, so as to obtain a probabilistic input tensor; The model training module is configured to input the probabilistic input tensor into a self-correlation attention model with fused physical constraints which is pre-constructed and trained; The result output module is configured to extract a contribution matrix from attention weights of the trained model, quantify a contribution proportion of each prediction time point to a prediction result, and output the probabilistic prediction result of the aircraft jolt and the corresponding contribution matrix as a final prediction result.

10. An electronic device, comprising: The electronic device comprises: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the method in any one of claims 1-8.

Images