Wavelet transform-based state spacecraft trajectory prediction method and system
The frequency domain features of the aircraft trajectory are extracted through wavelet transformation and the Transformer-Mamba model is constructed, which solves the problem of frequency domain information being ignored and long trajectory prediction capabilities in the prior art, and improves the accuracy of trajectory prediction and long sequence prediction effect.
Patent Information
- Application Number
- CN202411081496.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-08-08
AI Technical Summary
The prior art ignores frequency domain information in aircraft trajectory prediction, and has weak input characterization ability to long trajectories, resulting in prediction misalignment, especially in longer sequence trajectory prediction tasks.
The frequency domain features of the aircraft trajectory are extracted by a wavelet transform method, and a state space prediction model composed of the Transformer encoder, the Mamba state space module and the Transformer decoder are constructed. The trajectory feature extraction and prediction are performed through discrete wavelet transformation and inverse wavelet transformation. .
It improves the accuracy of aircraft trajectory prediction and long-sequence prediction capabilities, enhances the model's ability to capture the overall trend and detailed characteristics of trajectory movement, and reduces prediction misalignment.
Smart Images

Figure CN119089261A_ABST
Abstract
Description
Technical Field
[0001] The present specification relates to the field of trajectory prediction, and in particular to a state-space aircraft trajectory prediction method and system based on wavelet transform. Background Art
[0002] Spatial trajectory prediction refers to analyzing the three-dimensional trajectory of the target aircraft in the past period of time, and then predicting its flight trajectory in the future. In some special flight scenarios, the trajectory prediction of the aircraft can effectively perform key tasks such as threat assessment, target tracking, and maneuvering decisions. The accurate prediction of the target aircraft's movement plays an important role in the application of some scenarios.
[0003] In the early days, the trajectory prediction method of aircraft relied on aerodynamic models and state estimation algorithms. The aerodynamic model models the aircraft motion to achieve its trajectory prediction by affecting the performance, state, environment, intention and other aspects of the aircraft; the state estimation algorithm estimates the state of the aircraft at the next moment by constructing state transitions. However, these methods have a large number of modeling parameters and limited prediction length, making them difficult to apply to real environments. With the rise of artificial intelligence technology, deep neural networks have begun to be applied to target trajectory prediction. In the early stages of this direction, various sequence models such as LSTM, RNN, GRU, etc. were used for trajectory prediction. Shi et al. used LSTM neural networks to construct a 4D trajectory prediction model, which achieved obvious advantages over traditional Markov models. Sekhon et al. used LSTM networks with attention mechanisms for autonomous ship prediction. The model combines spatial and temporal attention mechanisms and spatial and temporal weights so that different areas of information can be focused. With the higher requirements for trajectory prediction length and accuracy, research began to use Transformer models for long-sequence trajectory prediction. Nguyen et al. proposed the TrAISformer model, which obtains data representation that is more conducive to model learning by gridding trajectory data and uses Transformer networks for iterative ship trajectory prediction. Guo et al. proposed the FlightBERT model for civil aviation trajectory prediction tasks. The model proposed a new attention block called attribute correlation attention to explicitly capture the impact of speed attributes on the corresponding position dimension, thereby improving the Transformer model's ability to mine trajectory attribute information.
[0004] However, existing technologies only consider the time series information of trajectory data and ignore the useful frequency domain information in the data. At the same time, these models have weak input representation capabilities for long trajectories and poor output prediction effects. In longer sequence trajectory prediction tasks, large prediction inaccuracies still occur. Summary of the invention
[0005] One of the embodiments of this specification provides a state space aircraft trajectory prediction method based on wavelet transform. The state space aircraft trajectory prediction method based on wavelet transform includes:
[0006] S1: Obtain the historical observation sequence of the original trajectory of the target aircraft;
[0007] S2: extracting the frequency domain features of the historical observation sequence by discrete wavelet transform;
[0008] S3: construct a state space prediction model and train it. The state space prediction model consists of a Transformer encoder, a Mamba state space module, and a Transformer decoder. The frequency domain features of the historical observation sequence are input into the trained state space trajectory prediction model to generate the frequency domain features of the future prediction trajectory.
[0009] S4: Restore the trajectory sequence of the aircraft by using the frequency domain features of the future predicted trajectory through inverse wavelet transform to determine the prediction result.
[0010] One of the embodiments of this specification provides a state space aircraft trajectory prediction system based on wavelet transform, comprising:
[0011] An acquisition module is used to obtain the historical observation sequence of the original trajectory of the target aircraft;
[0012] A feature extraction module, used for extracting frequency domain features of the historical observation sequence through discrete wavelet transform;
[0013] A trajectory prediction module is used to construct and train a state space prediction model, wherein the state space prediction model is composed of a Transformer encoder, a Mamba state space module, and a Transformer decoder. The frequency domain features of the historical observation sequence are input into the trained state space trajectory prediction model to generate the frequency domain features of the future predicted trajectory.
[0014] The determination module is used to restore the trajectory sequence of the aircraft by using the frequency domain characteristics of the future predicted trajectory through inverse wavelet transform to determine the prediction result.
[0015] One of the embodiments of the present specification provides a state-space aircraft trajectory prediction device based on wavelet transform, comprising a processor, wherein the processor is used to execute a state-space aircraft trajectory prediction method based on wavelet transform.
[0016] One of the embodiments of the present specification provides a computer-readable storage medium, wherein the storage medium stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes a state-space aircraft trajectory prediction method based on wavelet transform. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:
[0018] Figure 1 is an exemplary flow chart of state-space aircraft trajectory prediction based on wavelet transform according to some embodiments of this specification;
[0019] Figure 2 is an exemplary flow chart of state space prediction model training according to some embodiments of this specification;
[0020] Figure 3 A comparison diagram of the results of predicting the flight trajectory of an aircraft using a state space trajectory prediction model (WTTr-M) and other models shown in some embodiments of this specification;
[0021] Figure 4 is an example diagram of the training effect of the state space prediction model shown in some embodiments of this specification;
[0022] Figure 5 It is an exemplary module diagram of a state-space aircraft trajectory prediction system based on wavelet transform according to some embodiments of the present specification. DETAILED DESCRIPTION
[0023] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.
[0024] It should be understood that the "system", "device", "unit" and / or "module" used herein are a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.
[0025] As shown in this specification and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.
[0026] Flowcharts are used in this specification to illustrate the operations performed by the system according to the embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.
[0027] Figure 1 FIG. 1 is an exemplary flow chart of a state space aircraft trajectory prediction method based on wavelet transform according to some embodiments of this specification. Figure 1 As shown, the process includes the following steps.
[0028] S1: Get the historical observation sequence of the original trajectory of the target aircraft.
[0029] The target aircraft may include various types of aircraft, drones, etc. that are performing flight missions. The original trajectory of the target aircraft includes multiple data contents about the aircraft. Specifically, the data of the original trajectory is obtained through the ADS-B system and saved in the form of a sqlite database. The file usually contains the aircraft number, data acquisition time, longitude, latitude, altitude, and speed information during the operation of the aircraft.
[0030] In some embodiments, the data of the original trajectory is preprocessed, specifically including extracting trajectory information by removing abnormal points and resampling time. The historical observation sequence includes a 4-dimensional multivariate time series in the data format of [longitude, latitude, altitude, speed].
[0031] S2: Extracting the frequency domain features of the historical observation sequence through discrete wavelet transform.
[0032] In some embodiments, the frequency domain feature X of the historical observation sequence obtains its frequency domain information through discrete wavelet transform, specifically including: S21: decomposing the historical observation sequence into approximate coefficients and detail coefficients through discrete wavelet transform, wherein the approximate coefficients are the low-frequency part of the transformed signal, and the detail coefficients are the high-frequency part of the transformed signal; S22: combining the approximate coefficients and the detail coefficients by splicing in the variable dimension to extract the frequency domain features of the historical observation sequence.
[0033] Historical observation sequence X∈R C×L , perform discrete wavelet transform on each variable. Through discrete wavelet transform (DWT), the input trajectory sequence can be decomposed into approximate coefficients and detail coefficients. The approximate coefficients are:
[0034] A[n]=∑ k x[2n+k]·g[k],
[0035] Where x[n] is the input signal, g[k] is the low-pass filter coefficient, and the detail coefficient is:
[0036] D[n]=∑ k x[2n+k]·h[k],
[0037] Where h[k] is the high-pass filter coefficient.
[0038] The beneficial effect of using discrete wavelet transform is that its approximate coefficient is the low-frequency part of the transformed signal, which can reflect the overall trend of the trajectory movement in the frequency domain space. Understanding this part of the data will help the model to characterize the general direction of the trajectory; the detail coefficient is the high-frequency part of the transformed signal, which can reflect the movement details of the trajectory movement in the frequency domain space. Strengthening the learning of this part of the data will help the model to characterize the detailed characteristics of the trajectory in motion.
[0039] In some embodiments, the approximate coefficients are combined with the detail coefficients by splicing in the variable dimension to extract the frequency domain features of the historical observation sequence. The length of the approximate coefficient and the detail coefficient is half the length of the original trajectory, that is, The frequency domain features of the extracted historical observation sequence are For example, the input data dimension is (4, N), where N is the length of the input sequence. After discrete wavelet transform and concatenation with the low-frequency and high-frequency parts, frequency domain data with (8, N / 2) dimensions is formed. Thus, x wt The variable dimension doubles and the sequence dimension decreases by half. Through this dimensional change, the model can implement the "space for time" training strategy during training, that is, the length of the sequence that the model needs to remember becomes shorter, but the sequence dimension increases.
[0040] S3: Build and train a state space prediction model, which consists of a Transformer encoder, a Mamba state space module, and a Transformer decoder. The frequency domain features of the historical observation sequence are input into the trained state space trajectory prediction model to generate the frequency domain features of the future prediction trajectory. For details on the state space prediction model training process, please refer to Figure 2 The specific steps for generating frequency domain features of future prediction trajectories also include the following:
[0041] S31: The frequency domain features of the historical observation sequence are input into the Transformer encoder, and the contextual relationship in the frequency dimension is captured through the self-attention mechanism.
[0042] In some embodiments, the frequency domain data x of the historical observation sequence wtFuture trajectory prediction will be performed through an encoder-decoder. In this neural network, the data vector will represent the time-frequency features present in the input trajectory through high-dimensional embedding and the Transformer encoder, and the contextual relationship in the frequency dimension will be captured through the self-attention mechanism:
[0043] H = Encoder(InputEmbedding(x wt )).
[0044] S32: Based on the state space model of Mamba, the features passing through the Transformer encoder are selectively processed to obtain selective trajectory features.
[0045] In some embodiments, a state space model (SSM) is a model used to describe these state representations and predict what its next state might be based on some input, and can represent any cyclic process with potential states. The SSM uses a first-order differential equation to represent the state evolution inside the model:
[0046] h(t)′=Ah(t)+Bx(t),
[0047] y(t)=Ch(t),
[0048] Among them, A, B, and C are learnable matrices, x(t) is the sequence input to the Mamba module, specifically the sequence data after the Transformer encoder, y(t) is the output, h(t) is the hidden state representation, and h(t) is the hidden state representation. ′ is the derivative of the hidden state, indicating how the state is updated.
[0049] For continuous time series, the SSM model discretizes the sequence by step size. The discrete SSM model is expressed as:
[0050]
[0051] y t =Ch t ,
[0052] in, is a discretization of its continuous form.
[0053] The Mamba state-space model functions the matrices B, C, and Δ, allowing the model to adaptively adjust its behavior based on the input content, achieving selective attention and filtering of information. The selective state-space model can improve efficiency and performance in long-sequence tasks, and has the potential to enhance long-term trajectory prediction performance.
[0054] S33: Add the selective trajectory features to the features passed through the Transformer encoder to generate frequency domain features of the future predicted trajectory.
[0055] In some embodiments, the present invention embeds the Mamba model into the main network structure in the form of a module, receives the trajectory embedding H as input, and adds the output to the original embedding to capture all information of the trajectory and selective key information at the same time:
[0056] H ′ =MambaBlock(H)+H.
[0057] S4: The frequency domain features of the future predicted trajectory are restored to the time domain trajectory sequence features through inverse wavelet transform to determine the prediction result. For example, the prediction is input into the state space Transformer model for prediction, and the prediction result is output with a dimension of (8, L / 2). The time domain trajectory is restored through IDWT with a dimension of (4, L), where L is the length of the predicted trajectory.
[0058] Figure 2 is an exemplary flowchart of state space prediction model training according to some embodiments of this specification.
[0059] A trajectory training data set is obtained, where the trajectory training data set includes multiple groups of historical trajectory training sequences and future trajectory training sequences.
[0060] The data of the trajectory training data set can be obtained through the ADS-B system. The data is saved in the form of a sqlite database. The file usually contains the aircraft number, data acquisition time, longitude, latitude, altitude, and speed information during the operation of the aircraft.
[0061] In some embodiments, the data of the original trajectory is preprocessed, specifically including extracting trajectory information by removing abnormal points and resampling time. For each trajectory data, the trajectory can be divided into sliding window trajectories of equal length by sliding window processing, and each sliding window trajectory can be divided into two parts: a historical observation sequence and a future trajectory training sequence, where the historical observation sequence is used as the model input and the future trajectory training sequence is used as the model training label.
[0062] The historical trajectory training sequence is used as the input of the state-space trajectory prediction model, and the future prediction features of the historical trajectory training sequence are output; the future trajectory training sequence is used as a label and trained with the future prediction features of the historical trajectory training sequence using the MSE loss function.
[0063] Based on the training results of the loss function, the parameters of the state-space trajectory prediction model are adjusted to obtain the optimal state-space trajectory prediction model.
[0064] Figure 3 A comparison chart of the results of the state space trajectory prediction model (WTTr-M) disclosed in the present invention, the general Transformer network model and the TrAISformer model in predicting the flight trajectory of an aircraft; Figure 4 It is an example diagram of the training effect of the state-space prediction model, where the input shown in the figure is the input historical trajectory training sequence, ground truth is the future trajectory training sequence, WTTr-M is the future prediction feature of the state-space trajectory prediction model trained as described in the embodiment in the specification, and Transformer and TrAISformer represent the future prediction feature results of their corresponding models.
[0065] Figure 5 It is an exemplary module diagram of a state-space aircraft trajectory prediction system based on wavelet transform according to some embodiments of the present specification.
[0066] In some embodiments, the present invention discloses a state-space aircraft trajectory prediction system 500 based on wavelet transform, comprising:
[0067] An acquisition module 510 is used to acquire a historical observation sequence of the original trajectory of the target aircraft;
[0068] A feature extraction module 520, used to extract frequency domain features of the historical observation sequence by discrete wavelet transform;
[0069] The trajectory prediction module 530 is used to construct and train a state space prediction model, wherein the state space prediction model is composed of a Transformer encoder, a Mamba state space module, and a Transformer decoder, and the frequency domain features of the historical observation sequence are input into the trained state space trajectory prediction model to generate the frequency domain features of the future predicted trajectory;
[0070] The determination module 540 is used to restore the trajectory sequence of the aircraft by using the inverse wavelet transform to restore the frequency domain characteristics of the future predicted trajectory and determine the prediction result.
[0071] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of this specification. Although not explicitly stated here, those skilled in the art may make various modifications, improvements and corrections to this specification. Such modifications, improvements and corrections are suggested in this specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.
[0072] At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of this specification can be appropriately combined.
[0073] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this specification, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.
[0074] Similarly, it should be noted that in order to simplify the description disclosed in this specification and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this specification, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the subject matter of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.
[0075] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may change according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining digits. Although the numerical domains and parameters used to confirm the breadth of their range in some embodiments of this specification are approximate values, in specific embodiments, the setting of such numerical values is as accurate as possible within the feasible range.
[0076] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, documents, etc., cited in this specification are hereby incorporated by reference in their entirety. Except for application history documents that are inconsistent with or conflicting with the contents of this specification, documents that limit the broadest scope of the claims of this specification (currently or later attached to this specification) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the materials attached to this specification are inconsistent or conflicting with the contents described in this specification, the descriptions, definitions, and / or use of terms in this specification shall prevail.
[0077] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.
Claims
1. A state-space aircraft trajectory prediction method based on wavelet transform, It is characterized in that include: S1: Obtain the historical observation sequence of the original trajectory of the target aircraft; S2: extracting the frequency domain features of the historical observation sequence by discrete wavelet transform; S3: construct a state space prediction model and train it. The state space prediction model consists of a Transformer encoder, a Mamba state space module, and a Transformer decoder. The frequency domain features of the historical observation sequence are input into the trained state space trajectory prediction model to generate the frequency domain features of the future prediction trajectory. S4: Restore the trajectory sequence of the aircraft by using the frequency domain features of the future predicted trajectory through inverse wavelet transform to determine the prediction result.
2. The state-space aircraft trajectory prediction method based on wavelet transform as claimed in claim 1, It is characterized in that The S2 extracts the frequency domain features of the historical observation sequence by discrete wavelet transform, and further includes: S21: decomposing the historical observation sequence into approximate coefficients and detail coefficients by discrete wavelet transform, wherein the approximate coefficients are the low-frequency parts of the transformed signal, and the detail coefficients are the high-frequency parts of the transformed signal; S22: The approximate coefficients are combined with the detail coefficients by splicing in the variable dimension to extract the frequency domain features of the historical observation sequence.
3. The state-space aircraft trajectory prediction method based on wavelet transform as claimed in claim 1, It is characterized in that The specific steps of generating the frequency domain features of the future prediction trajectory include: S31: Input the frequency domain features of the historical observation sequence into the Transformer encoder, and capture the contextual relationship in the frequency dimension through the self-attention mechanism; S32: selectively processing the features passing through the Transformer encoder based on the state space model of Mamba to obtain selective trajectory features; S33: Add the selective trajectory features to the features passed through the Transformer encoder to generate frequency domain features of the future predicted trajectory.
4. The state-space aircraft trajectory prediction method based on wavelet transform as claimed in claim 1, It is characterized in that The training of the state-space prediction model includes: Acquire a trajectory training data set, where the trajectory training data set includes multiple groups of historical trajectory training sequences and future trajectory training sequences; Using the historical trajectory training sequence as an input of a state-space trajectory prediction model, and outputting future prediction features of the historical trajectory training sequence; The future trajectory training sequence is used as a label and the future prediction features of the historical trajectory training sequence are trained using the MSE loss function; Get the trained state-space trajectory prediction model.
5. A state-space aircraft trajectory prediction system based on wavelet transform, It is characterized in that include: An acquisition module is used to obtain the historical observation sequence of the original trajectory of the target aircraft; A feature extraction module, used for extracting frequency domain features of the historical observation sequence through discrete wavelet transform; A trajectory prediction module is used to construct and train a state space prediction model, wherein the state space prediction model is composed of a Transformer encoder, a Mamba state space module, and a Transformer decoder. The frequency domain features of the historical observation sequence are input into the trained state space trajectory prediction model to generate the frequency domain features of the future predicted trajectory. S4: Restore the trajectory sequence of the aircraft by using the frequency domain features of the future predicted trajectory through inverse wavelet transform to determine the prediction result.
6. A state-space aircraft trajectory prediction device based on wavelet transform, comprising a processor, wherein the processor is used to execute the state-space aircraft trajectory prediction method based on wavelet transform according to any one of claims 1 to 4.
7. A computer-readable storage medium storing computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the state-space aircraft trajectory prediction method based on wavelet transform according to any one of claims 1 to 4.
Citation Information
Patent Citations
Lightweight time sequence prediction method based on discrete wavelet transform
CN114219027A
Dynamic space-time diagram attention method for ship trajectory prediction
CN117112710A
Regional ionosphere delay prediction method based on LSTM-Transform model
CN117973440A
Cited By
Sound velocity distribution forecasting method based on Mama neural network
CN119918593A
Robot action prediction method and device, equipment and storage medium
CN120002675A
Single-target tracking method and device based on state space model and attention and medium
CN120235912A
Unmanned aerial vehicle control method combining sequence modeling and predictive control
CN120686888A
FT-Mama architecture-based ship attitude prediction method and system
CN121167659A
