A Flight Trajectory Prediction Method in the Case of Airborne Sensor Failure

Abstract

The present invention discloses a flight trajectory prediction method in the case of airborne sensor failure, comprising the following steps: S1, screening out the data for trajectory prediction from the publicly available ADS-B aircraft terminal and ground station terminal datasets; S2, optimizing the data in S1 by using a Student-t distribution variational Bayesian adaptive Kalman filter model; S3, making a spatio-temporal feature dataset for aircraft trajectory prediction, and the feature dataset includes a training set, a validation set and a test set; S4, establishing a hybrid network prediction model, and training the hybrid network prediction model by using the feature dataset in S3; S5, predicting the flight trajectory by using the hybrid network prediction model. The prediction model of the present invention using multi-channel spatio-temporal hybrid feature extraction fully considers the influence of target multi-dimensional state information on target trajectory prediction, and can also reflect the internal connection between the temporal information of the target state, and can ensure the reliability and accuracy of model prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of trajectory prediction, and particularly to a flight trajectory prediction method in the case of airborne sensor failure. Background Art

[0002] In recent years, the Automatic Dependent Surveillance-Broadcast (ADS-B) technology has developed vigorously and become a promising technology in the next-generation air traffic control. Compared with traditional radars, the ADS-B system does not require manual operation or query, and has the advantages of high positioning accuracy, large data update rate, low maintenance cost, etc. However, since the ADS-B data is derived from the GPS system rather than the key navigation information such as position, altitude, and speed independently calculated by the aircraft flight control system, and the GPS is very vulnerable to intentional or unintentional illegal interference, once the airborne sensor fails and causes the data at the aircraft end to become invalid, it will greatly affect aviation safety. Therefore, reliable short-term and long-term prediction of the aircraft trajectory can improve the flight safety in busy airspace.

[0003] The research and development of flight trajectory prediction methods can provide decision-making support for air traffic controllers and avoid accidents safely and efficiently. Currently, advanced flight trajectory prediction methods are mainly divided into two types: prediction based on physical models and prediction based on data-driven models. Among them, modeling based on physical models requires clear input of modeling flight performance and real-time aircraft states, which has become an obstacle to modeling based on physical models. In recent years, with the rapid development of artificial intelligence, scholars have gradually used deep neural network models to learn and predict flight trajectories, and have also obtained relatively high prediction accuracy. However, previous studies often did not consider comprehensively enough. Either a single neural network model was used to predict flight trajectories, resulting in relatively low prediction accuracy; or a combined neural network was used, but the influence of target multi-dimensional state information on target trajectory prediction was not considered, and the internal connection between the sequential information reflecting the target state was not explored. Summary of the Invention

[0004] The purpose of the present invention is to provide a flight trajectory prediction method in the case of airborne sensor failure, which solves the above technical problems and has more accurate prediction results.

[0005] To achieve the above purpose, the present invention provides the following technical solution: A flight trajectory prediction method in the case of airborne sensor failure, comprising the following steps:

[0006] S1. Screen out the data for trajectory prediction from the publicly available ADS-B aircraft-end and ground-station-end data sets;

[0007] S2. Optimize the data in S1 using the Student-t distribution variational Bayesian adaptive Kalman filter model;

[0008] S3. Create a spatio-temporal feature dataset for aircraft trajectory prediction, where the feature dataset includes a training set, a validation set, and a test set;

[0009] S4. Establish an N-Inception-LSTM-ATT hybrid network prediction model, and use the feature dataset in S3 to train the N-Inception-LSTM-ATT hybrid network prediction model;

[0010] S5. Use the N-Inception-LSTM-ATT hybrid network prediction model to predict the flight trajectory.

[0011] In some embodiments, S1 includes the following steps:

[0012] S11. Process the internationally public ADS-B aircraft terminal dataset, visualize the aircraft flight trajectory, manually judge the validity of the trajectory, eliminate invalid and discontinuous trajectories, and screen out continuous and complete data;

[0013] S12. Conduct a visual analysis of the ADS-B ground station terminal dataset from two perspectives: the clock of the receiver and the received signal strength. Identify and eliminate the receiver data with damaged clocks and received signal strength faults, and retain other data;

[0014] S13. Integrate the final aircraft terminal dataset and ground station terminal dataset in S12 into a new dataset to ensure the reliability of the data and facilitate the production of subsequent training sets, validation sets, and test sets.

[0015] In some embodiments, S12 includes the following steps:

[0016] First, screen the data from the aspect of the receiver clock. Plot the two-dimensional graph of the server timestamp and the receiver timestamp in Matlab, and screen out the data on the straight line with a slope of 1;

[0017] Second, screen the data from the aspect of the received signal strength. Plot the relationship graph between the received distance and the received signal strength. Theoretically, the closer the distance, the stronger the received signal strength, and the received signal strength has a linear relationship with the logarithm of the received distance with base 10. Screen out the data that conforms to this rule;

[0018] Finally, eliminate the receiver data that does not meet the above conditions, and finally retain other data.

[0019] In some embodiments, in S2, the Bayesian adaptive Kalman filter model can be used to recursively jointly solve the state and measurement noise parameters, effectively solving the problem of unknown time-varying measurement noise in the original data and further optimizing the data quality. The Bayesian adaptive Kalman filter model is as follows:

[0020]

[0021] α k,i = α k-1,i + 0.5;

[0022] v k,i = α k,i / β k,i ;

[0023]

[0024] D k = (Z k - H k X k )(Z k - H k X k ) T ;

[0025]

[0026] Taking longitude as an example, in the formula, P k-1 is the posterior estimation state covariance matrix at time k - 1; ρ is the attenuation coefficient, ρ ∈ (0, 1]; Λ is the precision matrix; X k is the posterior state estimation value of longitude at time k; is the prior state estimation value of longitude at time k; F k is the state transition matrix; X k-1 is the posterior state estimation value of longitude at time k - 1; is the transposed result of the state transition matrix; Q k is the process noise variance matrix; is the prior estimation state covariance matrix at time k; is the prior value of the hyperparameter α of the Student - t distribution in the i - dimension at time k; α k-1,i is the posterior value of the hyperparameter α of the Student - t distribution in the i - dimension at time k - 1; is the prior value of the hyperparameter β of the Student - t distribution in the i - dimension at time k; β k-1,i is the posterior value of the hyperparameter β of the Student - t distribution in the i - dimension at time k - 1; α k,i is the posterior value of the hyperparameter α of the Student - t distribution in the i - dimension at time k; v k,i is the degree of freedom in the i - dimension at time k; βk,i is the posterior value of the hyperparameter β of the Student-t distribution in the i-th dimension at time k; M k is an auxiliary matrix for convenient calculation; H k is the measurement matrix; is the transposed result of the measurement matrix; K k is the filtering gain matrix; is the inverse matrix of the auxiliary matrix; Z k is the observation vector; is the transpose of the filtering gain matrix; D k is the posterior estimation covariance matrix at time k; is the trace matrix at time k; λ k is the auxiliary random variable introducing the measurement distribution at time k; d is the dimension of the observation vector; is the value of the i-th dimension of the trace matrix at time k.

[0027] In some embodiments, S3 includes the following steps:

[0028] S31. Normalize the dataset obtained in S2 so that the normalized data is in the interval [0, 1];

[0029] S32. Select the features of a single time node; the input features of the single time node are selected as the time at the ground station end, the received signal strength, longitude, latitude, and altitude; the output features are the longitude, latitude, and altitude of the aircraft;

[0030] S33. Select an appropriate sliding window size;

[0031] S34. Shuffle the dataset to finally obtain an appropriate training set, validation set, and test set.

[0032] In some embodiments, the N-Inception-LSTM-ATT hybrid neural network fully considers the influence of the target multi-dimensional state information on the target trajectory prediction and can also reflect the internal connection between the temporal information of the target state. After training, a more accurate and reliable prediction result can be obtained. S4 includes the following steps:

[0033] S41. Vertically splice the convolutional neural network Inception, the long short-term memory neural network LSTM, and the self-attention mechanism sub-module to construct the N-Inception-LSTM-ATT hybrid neural network, where the optimizer is Adam and the loss function is MSE;

[0034] S42. Set the number of iterations and train the N-Inception-LSTM-ATT hybrid neural network to minimize the loss function MSE;

[0035] S43. Adjust the network parameters to obtain an N-Inception-LSTM-ATT hybrid neural network prediction model.

[0036] In some embodiments, S41 includes the following steps:

[0037] First, input the training set into the convolutional neural network Inception to extract the spatial features of the target state information, and input the obtained result into the long short-term memory neural network LSTM with stronger long-term sequence processing ability to mine the temporal features of the data. Then, connect an ATT module before the output layer to fully consider all input vectors, capture the dependencies in the sequence data, and improve the performance of the model in sequence data. The self-attention mechanism uses the information inherent in the data features for attention interaction, and calculates the query vector Q (Query), key vector K (Key), and value vector V (Value) through multiple inputs. The calculation expressions are as follows:

[0038]

[0039] In the formula, d k is the dimension of K, and the purpose is to scale the value after the dot product of the query vector Q and the key vector K.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] In the case where the data at the aircraft end is unavailable due to the abnormality of the airborne sensor, the present invention uses the signal strength of the ground receiver to provide a certain degree of reliable prediction value, reducing the time and labor costs.

[0042] The prediction model of the present invention using multi-channel spatio-temporal hybrid feature extraction fully considers the influence of the target multi-dimensional state information on the target trajectory prediction, and can also reflect the internal connection between the temporal information of the target state, ensuring the reliability and accuracy of the model prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a schematic flowchart of the present invention;

[0044] Figure 2 is a schematic flowchart of Embodiment S1 of the present invention;

[0045] Figure 3 is a schematic flowchart of Embodiment S3 of the present invention;

[0046] Figure 4 is a schematic flowchart of Embodiment S32 of the present invention;

[0047] Figure 5 is a schematic flowchart of Embodiment S33 of the present invention;

[0048] Figure 6 It is a schematic structural diagram of the N-Inception-LSTM-ATT hybrid neural network model of the present invention;

[0049] Figure 7 It is a schematic flowchart of trajectory prediction in an embodiment of the present invention;

[0050] Figure 8 shows the predicted results of longitude, latitude and altitude of different models in an embodiment of the present invention; among them, Figure 8(a) is a comparison diagram of the three-dimensional prediction results of the 2-Inception and 2-Inception-LSTM models; Figure 8(b) is a comparison diagram of the three-dimensional prediction results of the Inception-LSTM and 2-Inception-LSTM; Figure 8(c) is a comparison diagram of the three-dimensional prediction results of the 2-Inception-LSTM and 2-Inception-LSTM-ATT. Detailed implementation manners

[0051] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

[0052] Please refer to Figure 1 , a flight trajectory prediction method in the case of an airborne sensor failure, including the following steps:

[0053] S1. Screen out the data for trajectory prediction from the publicly available ADS-B aircraft terminal and ground station terminal data sets.

[0054] As Figure 2 shown, it specifically includes the following steps:

[0055] S11. Process the internationally publicly available ADS-B aircraft terminal data set, import the longitude, latitude and altitude data of the aircraft terminal into Matlab software, use the point plotting method to draw a three-dimensional trajectory to visualize the aircraft flight trajectory, manually eliminate the trajectories with signal interruptions and obvious errors, so as to judge the validity of the trajectories, eliminate the invalid and discontinuous trajectories, and screen out the continuous and complete high-reliability data.

[0056] S12. Conduct a visual analysis of the ADS-B ground station terminal data set from two perspectives of the receiver's clock and the received signal strength, identify and eliminate the receiver data with clock damage and received signal strength faults, and retain the high-reliability data.

[0057] Specifically, first, filter the data from the perspective of the receiver clock. Plot a two-dimensional graph of the server timestamp and the receiver timestamp in Matlab. The horizontal axis is the server timestamp, and the vertical axis is the receiver timestamp. Theoretically, the graph should be a straight line with a slope of 1. As long as the data with a small deviation meets the requirements, it can be retained. Then, filter the data from the perspective of the received signal strength. Plot a two-dimensional graph of the received distance and the received signal strength. The horizontal axis is the received distance, and the vertical axis is the received strength. Theoretically, the received signal strength is affected by the received distance and the propagation environment. When the received distance decreases, the received strength should increase accordingly. Any data where the received signal strength remains unchanged or increases when the received distance decreases needs to be excluded. The received signal strength has a linear relationship with the logarithm of the received distance to the base 10. Filter out the receiver data that does not meet the above conditions and finally retain the highly reliable data.

[0058] S13. Integrate the final aircraft-side dataset and ground station-side dataset in S12 into a new dataset to ensure data reliability and facilitate the creation of subsequent training sets, validation sets, and test sets.

[0059] S2. Optimize the data in S1 using the Student-t distributed variational Bayesian adaptive Kalman filter model;

[0060] Since the original dataset is affected by the characteristics of the sensors themselves and the surrounding complex environment, it is necessary to use the Student-t distributed variational Bayesian adaptive Kalman filter model to optimize it to solve the problem of unknown time-varying noise and the existence of outliers.

[0061] Model longitude, latitude, and altitude respectively. For longitude, the state variable and the observation vector can be selected as the longitude values of three different sensors. The same applies to the modeling of latitude and altitude. The Student-t distributed variational Bayesian adaptive Kalman filter model mainly consists of two aspects: prediction and update. By changing the values of α and β to change v and λ, the purpose of adaptive estimation is finally achieved. The prediction and update formulas are as follows:

[0062]

[0063] α k,i = α k-1,i + 0.5;

[0064] v k,i = α k,i / β k,i ;

[0065]

[0066] D k =(Z k - H kX k )(Z k -H k X k ) T ;

[0067]

[0068] Taking longitude as an example, in the formula, P k-1 is the posterior estimation state covariance matrix at time k - 1; ρ is the attenuation coefficient, ρ ∈ (0, 1]; Λ is the precision matrix; X k is the posterior state estimation value of longitude at time k; is the prior state estimation value of longitude at time k; F k is the state transition matrix; X k-1 is the posterior state estimation value of longitude at time k - 1; is the transposed result of the state transition matrix; Q k is the process noise variance matrix; is the prior estimation state covariance matrix at time k; is the prior value of the hyperparameter α of the Student - t distribution in the i - dimension at time k; α k-1,i is the posterior value of the hyperparameter α of the Student - t distribution in the i - dimension at time k - 1; is the prior value of the hyperparameter β of the Student - t distribution in the i - dimension at time k; β k-1,i is the posterior value of the hyperparameter β of the Student - t distribution in the i - dimension at time k - 1; α k,i is the posterior value of the hyperparameter α of the Student - t distribution in the i - dimension at time k; v k,i is the degree of freedom in the i - dimension at time k; β k,i is the posterior value of the hyperparameter β of the Student - t distribution in the i - dimension at time k; M k is the auxiliary matrix for convenient calculation; H k is the measurement matrix; is the transposed result of the measurement matrix; K k is the filter gain matrix; is the inverse matrix of the auxiliary matrix; Z k is the observation vector; is the transpose of the filter gain matrix; D k is the posterior estimation covariance matrix at time k; is the trace matrix at time k; λ k is the auxiliary random variable introducing the measurement distribution at time k; d is the dimension of the observation vector; is the value of the i - th dimension of the trace matrix at time k.

[0069] S3. Create a spatio-temporal feature dataset for aircraft trajectory prediction. The feature dataset includes a training set, a validation set, and a test set. As shown in Figure 3 the following steps are included:

[0070] S31. Normalize the dataset obtained in S2 so that the normalized data is in the range [0, 1]. The normalization process is to perform min-max normalization on each field in the dataset separately:

[0071]

[0072] In the formula, X is the data of a certain field, min is the minimum value in this field, is the maximum value, and N is the result after normalization.

[0073] S32. Select the features at a single time node. As shown in Figure 4 the input features at the single time node are selected as 3 ground station data, each of which includes time, receiver signal strength, longitude, latitude, and altitude, for a total of 15 input features; the output features are the longitude, latitude, and altitude of the aircraft, for a total of 3 output features.

[0074] S33. Select an appropriate sliding window size. As shown in Figure 5 in order to make full use of the time dimension features of the dataset, the time window is set to 64. Combining with the input features selected in the previous step, the slider is 64×15.

[0075] S34. Shuffle the dataset to finally obtain appropriate training set, validation set, and test set.

[0076] S4. Establish an N-Inception-LSTM-ATT hybrid network prediction model and use the feature dataset in S3 to train the N-Inception-LSTM-ATT hybrid network prediction model. As shown in Figure 6 the following steps are included:

[0077] S41. Vertically splice the convolutional neural network Inception, the long short-term memory neural network LSTM, and the self-attention mechanism sub-module to construct an N-Inception-LSTM-ATT hybrid neural network, where the optimizer is Adam and the loss function is MSE.

[0078] First, the training set is input into the convolutional neural network Inception to extract the spatial features of the target state information, and the obtained results are input into the long short-term memory neural network LSTM, which has stronger processing capabilities for long time series, to mine the temporal features of the data; then, the ATT module is connected before the output layer to fully consider all input vectors, capture the dependencies in the sequence data, and improve the performance of the model in the sequence data; the long short-term memory neural network LSTM consists of a total of 128 memory units, followed by the Dropout layer with a parameter of 0.5 (indicating that 50% of the neurons are discarded). The Dropout layer reduces the dependencies between neurons by randomly discarding a part of the neurons during the training process, thereby reducing the overfitting of the model to the training data and forcing the neural network to learn more robust features, thereby improving the generalization ability of the model.

[0079] The self-attention mechanism uses the inherent information in the data features to perform attention interaction, and calculates the query vector Q (Query), key vector K (Key) and value vector V (Value) through multiple inputs. The calculation expression is:

[0080]

[0081] Where, d k is the dimension of K, and its purpose is to scale the value after the dot product of the query vector Q and the key vector K.

[0082] S42, set the number of iterations, train the N-Inception-LSTM-ATT hybrid neural network to minimize the loss function MSE; Figure 7 As shown, the following steps are included:

[0083] Input the training set data into the N-Inception-LSTM-ATT hybrid neural network, observe the Loss value of the training set and the Loss value of the test set, and save it after judging whether the hybrid neural network is fitted and meets the accuracy. When judging whether the hybrid neural network is fitted, the optimizer is Adam and the loss function is MSE. The loss function MSE is used as the evaluation indicator, and the formula is as follows:

[0084]

[0085] In the formula, y i is the true value of the ith data sample, is the corresponding predicted value, n is the number of samples, the smaller the value of the loss function MSE, the closer the model's prediction result is to the true value, and the better the model's performance is. Set an appropriate number of iterations and train the network to minimize the loss function MSE, and the loss curve shows a normal downward trend.

[0086] S43. Adjust the network parameters to obtain the N-Inception-LSTM-ATT hybrid neural network prediction model.

[0087] When judging whether the model meets the accuracy, it is necessary to adjust the network parameters, such as the learning rate lr, the batch size batch_size, and the number of iterations, etc. Among them, lr is set to 0.001, batch_size is set to 4000, and the number of training rounds is set to 200. Finally, the N-Inception-LSTM-ATT hybrid neural network prediction model with the highest accuracy is obtained.

[0088] S5. Use the N-Inception-LSTM-ATT hybrid network prediction model to predict the flight trajectory.

[0089] In a specific embodiment, the comparison between the present application and the models in the prior art is as follows:

[0090] Table 1 Parameter setting table of the experimental group

[0091]

[0092] As shown in Figure 8, Figure 8(a) shows the comparison diagram of the three-dimensional prediction results of the 2-Inception and 2-Inception-LSTM models; Figure 8(b) is the comparison diagram of the three-dimensional prediction results of Inception-LSTM and 2-Inception-LSTM; Figure 8(c) is the comparison diagram of the three-dimensional prediction results of 2-Inception-LSTM and 2-Inception-LSTM-ATT. Through the comparison of multiple different models, it can be intuitively seen that the N-Inception-LSTM-ATT model has higher prediction accuracy and is closer to the real trajectory.

[0093] Considering the situation where the data at the aircraft end is unavailable due to the failure of the on-board sensor, features such as the received signal strength of the ground receiving end are selected for prediction. Using the N-Inception-LSTM-ATT model can provide reliable and accurate prediction values within a certain range, reducing time and labor costs.

Claims

1. A flight trajectory prediction method in the case of an airborne sensor failure, characterized in that It includes the following steps: S1. Screen out the data for trajectory prediction from the publicly available ADS-B aircraft-end and ground-station-end datasets; S2. Optimize the data in S1 using the Student-t distribution variational Bayesian adaptive Kalman filter model; the Bayesian adaptive Kalman filter model is: ； ； ； ； ； ； ； ； ； ； ； ； Taking longitude as an example, in the formula, is the posterior estimation state covariance matrix at time is the attenuation coefficient, ; is the precision matrix; is the posterior state estimation value of longitude at time is the prior state estimation value of longitude at time is the state transition matrix; is the posterior state estimation value of longitude at time is the transpose result of the state transition matrix; is the process noise variance matrix; is the prior estimation state covariance matrix at time is the hyperparameter prior value of the Student-t distribution of dimension is the hyperparameter posterior value of the Student-t distribution of dimension is the hyperparameter prior value of the Student-t distribution of dimension is the hyperparameter posterior value of the Student-t distribution of dimension is the hyperparameter posterior value of the Student-t distribution of dimension is the degree of freedom of the Student-t distribution of dimension is the hyperparameter posterior value of the Student-t distribution of dimension is the auxiliary matrix for convenient calculation; is the measurement matrix; is the transpose result of the measurement matrix; is the filtering gain matrix; is the inverse matrix of the auxiliary matrix; is the observation vector; is the transpose of the filtering gain matrix; is the posterior estimation covariance matrix at time is the trace matrix at time is the auxiliary random variable introducing the measurement distribution at time the dimension of the observation vector; is the value of the -th dimension of the trace matrix at time S3. Make a spatio-temporal feature dataset for aircraft trajectory prediction, and the feature dataset includes a training set, a validation set, and a test set; S4. Establish an N-Inception-LSTM-ATT hybrid network prediction model, and use the feature dataset in S3 to train the N-Inception-LSTM-ATT hybrid network prediction model; it includes the following steps: S41. Vertically splice the convolutional neural network Inception, the long short-term memory neural network LSTM, and the self-attention mechanism sub-module to construct an N-Inception-LSTM-ATT hybrid neural network, where the optimizer is Adam and the loss function is MSE; specifically, First, input the training set into the convolutional neural network Inception to extract the spatial features of the target state information, and input the obtained result into the long short-term memory neural network LSTM to mine the temporal features of the data; Then, connect the ATT module before the output layer, fully consider all input vectors, capture the dependencies in the sequence data, and improve the performance of the model in the sequence data; the self-attention mechanism uses the information inherent in the data features for attention interaction, and calculates the query vector Q, the key vector K, and the value vector V through multiple inputs, and the calculation expression is: ； In the formula, is 's dimension, aiming to scale the value after the dot product of the query vector and the key vector ; S42. Set the number of iterations and train the N-Inception-LSTM-ATT hybrid neural network to minimize the loss function MSE; S43. Adjust the network parameters to obtain the N-Inception-LSTM-ATT hybrid neural network prediction model; S5. Use the N-Inception-LSTM-ATT hybrid network prediction model to predict the flight trajectory.

2. The flight trajectory prediction method in case of an airborne sensor failure according to claim 1, characterized in that, S1 includes the following steps: S11. Process the internationally publicly available ADS-B aircraft-end dataset, visualize the aircraft flight trajectory, manually judge the validity of the trajectory, eliminate invalid and discontinuous trajectories, and screen out continuous and complete data; S12. Conduct a visual analysis of the ADS-B ground-station-end dataset from two perspectives: the receiver clock and the received signal strength, identify and eliminate the receiver data with clock damage and received signal strength faults, and retain other data; S13. Integrate the final aircraft-end dataset and ground-station-end dataset in S12 into a new dataset to ensure the reliability of the data and facilitate the production of the subsequent training set, validation set, and test set.

3. The flight trajectory prediction method in case of an airborne sensor failure according to claim 2, characterized in that, S12 It includes the following steps: First, screen the data from the aspect of the receiver clock, draw a two-dimensional graph of the server timestamp and the receiver timestamp in Matlab, and screen out the data on the straight line with a slope of 1; Secondly, we screened the data from the perspective of received signal strength and plotted the relationship between the receiving distance and the received signal strength. The closer the distance, the stronger the received signal strength. The received signal strength and the logarithm of the receiving distance with the base 10 are linearly related. We screened out the data that met this rule. Finally, the receiver data that does not meet the above conditions is eliminated, and the other data is finally retained.

4. A flight trajectory prediction method in the case of an airborne sensor failure according to claim 1, characterized in that, S3 includes the following steps: S31, normalize the data set obtained in S2 so that the normalized data is between the interval [0,1]; S32, selecting features of a single time node; The input features of the single time node are selected from the time, receiver signal strength, longitude, latitude and altitude of the ground station; The output features are the longitude, latitude, and altitude of the aircraft; S33, selecting a suitable sliding window size; S34. Shuffle the data set to finally obtain the training set, validation set and test set.

Images