Multi-area-array phased array radar target trajectory prediction system and method
By integrating data acquisition, preprocessing, deep learning prediction, multi-source fusion and intelligent optimization modules in the multi-faceted array phased array radar target trajectory prediction system, the problem of insufficient accuracy and reliability of target trajectory prediction in the existing technology is solved, and more efficient and accurate target trajectory prediction is achieved.
Patent Information
- Application Number
- CN202411520502.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-29
- Publication Date
- 2025-03-11
AI Technical Summary
The prior art has problems of insufficient accuracy and reliability in target trajectory prediction, especially in the complexity of target motion and multi-source data fusion.
A multi-faceted array phased array radar target trajectory prediction system is adopted, which includes a data acquisition module, a data preprocessing module, a trajectory prediction module, a result output module, an intelligent optimization module and a multi-source fusion module. The system achieves more accurate and reliable target trajectory prediction through real-time data acquisition, advanced data cleaning and preprocessing, trajectory prediction of deep learning models, uncertainty estimation, multi-source data fusion and dynamic system optimization.
It improves the accuracy and reliability of target trajectory prediction, can effectively process complex target motion and multi-source data, provide more comprehensive decision-making basis, and adapt to different scenarios through intelligent optimization and dynamic adjustment.
Smart Images
Figure CN119596298A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of radar target trajectory prediction systems, and in particular to a multi-array phased array radar target trajectory prediction system and method. Background Art
[0002] In modern military, aerospace, navigation and other fields, accurate prediction of target trajectory is of vital importance. As an advanced detection device, multi-faceted phased array radar plays an important role in target monitoring and tracking due to its high precision, high resolution and fast response. However, there are still many challenges in relying solely on multi-faceted phased array radar for target trajectory prediction.
[0003] Traditional target trajectory prediction methods are mainly based on simple mathematical models and empirical judgments, and their accuracy and reliability are often limited. For example, prediction models based on the assumption of uniform linear motion or uniformly accelerated linear motion are difficult to accurately predict trajectories when the target motion is more complex. Moreover, traditional methods can usually only use a single data source, that is, the data detected by the multi-array phased array radar itself, lack the fusion of multi-source information, and cannot give full play to the advantages of different data sources.
[0004] With the continuous advancement of science and technology, the movement patterns of targets are becoming more and more complex and changeable. Targets in modern warfare may take various flexible tactical actions, such as high-speed maneuvers and sudden turns, which puts higher requirements on target trajectory prediction. At the same time, in the civilian field, such as air traffic control and maritime ship monitoring, more accurate target trajectory prediction is also needed to ensure safe and efficient operation.
[0005] In terms of data processing, the data detected by multi-array phased array radars often contain noise and outliers, which require effective cleaning and preprocessing before they can be used for trajectory prediction. Traditional data processing methods may not be able to fully remove noise interference, affecting the accuracy of prediction results. In addition, how to extract effective features from a large amount of radar data is also a problem that needs to be solved urgently.
[0006] The development of deep learning technology has brought new opportunities for target trajectory prediction. Deep learning models can automatically learn complex patterns and regularities in data and have powerful feature extraction and prediction capabilities. However, the application of deep learning to multi-array phased array radar target trajectory prediction still faces some technical challenges. For example, how to choose a suitable deep learning model architecture, how to use a large amount of historical data for effective model training, and how to deal with the uncertainty of model prediction results.
[0007] At the same time, multi-sensor fusion technology has gradually attracted attention. By integrating data from multi-array phased array radars and other sensors or data sources, more comprehensive and accurate target information can be provided, improving the accuracy of trajectory prediction. However, the fusion of multi-source data also faces problems such as inconsistent data formats and data synchronization.
[0008] In summary, developing an innovative multi-array phased array radar target trajectory prediction method and system to solve the shortcomings of the existing technology has important practical significance and application value. Summary of the invention
[0009] The multi-array phased array radar target trajectory prediction system and method proposed in the present invention are designed to solve the problems mentioned in the above-mentioned prior art.
[0010] In order to achieve the above object, the present invention adopts the following technical solution: a multi-array phased array radar target trajectory prediction system, comprising:
[0011] The data acquisition module is used to receive the target data detected by the multi-array phased array radar. The target data includes the position, velocity, and acceleration information of the target, which can be expressed as a vector T = (P, V, A, ...), where P represents the target position data, V represents the target velocity data, and A represents the target acceleration data. The module can receive radar data in real time and preliminarily organize and store the data. At the same time, the module has an adaptive data acquisition function, which can automatically adjust the acquisition parameters according to the dynamic characteristics of the target and the radar environment to ensure the accuracy and integrity of the data.
[0012] The data preprocessing module cleans and preprocesses the collected target data. It uses advanced data cleaning algorithms to remove outliers and noise interference to ensure the purity of the data. 1 ,x 2 ,...,x n}, the denoising function is h(X) = {x′ 1 ,x′ 2 ,...,x′ n},in The sliding average method is used for denoising, m is the window size, here, h(X) represents the denoising function, X is the original data set, x 1 , x 2 , …, x n is the 1-nth data point in the original data, x′ 1 , x′ 2 , …, x′ n is the 1-nth data point after denoising, and then normalized. Let the normalized data be Y={y 1 ,y 2 , …, y n},in μ is the mean of the denoised data, and σ is the standard deviation of the denoised data. Here, Y is the standardized data set, y i are the normalized data points.
[0013] The trajectory prediction module uses a pre-trained deep learning model to predict the trajectory of the pre-processed target data. The model adopts a recurrent neural network (RNN) architecture, which can effectively process time series data. Assume that the input data is I = (i 1 ,i 2 ,…,i t ), where i 1 ,i 2 ,…,i t represents the target data vector at time 1-t, and the predicted output is O=(o 1 ,o 2 ,…,o t+1 ), where o 1 ,o 2 ,…,o t+1 Indicates the predicted position of the target at time 1-t+1. The model predicts the future position, velocity, and acceleration information of the target by learning from historical data. The training data of the model includes a large amount of historical target data and the corresponding future real trajectory data. In addition, this module introduces an uncertainty estimation mechanism to quantitatively evaluate the uncertainty of the prediction results, providing users with a more comprehensive decision-making basis.
[0014] The result output module outputs the trajectory prediction results to the user in an intuitive way, including graphical display and data report form. At the same time, this module can also provide prediction result displays of different accuracy according to user needs.
[0015] The intelligent optimization module automatically adjusts system parameters and model structure according to user feedback on prediction results and the effects in actual applications to improve the accuracy and practicality of prediction. Assume that user feedback is R and the optimization function is where f j (R) represents the adjustment function of the jth parameter according to user feedback R, α j is the corresponding weight coefficient, satisfying g(R) is the optimization function and k is the total number of parameters.
[0016] Multi-source fusion module, which can integrate the data of multi-array phased array radar and other sensors or data sources to achieve more comprehensive and accurate target trajectory prediction. Suppose the multi-array phased array radar data is D r =(d 1 ,d 2 ,…,d n ), other data sources are Do =(d′ 1 ,d′ 2 ,…,d′ m ), the fused data is D f =(f 1 ,f 2 ,…,f p ), where f i =ω 1 d i +ω 2 d′ i％m+1 ,ω 1 +ω 2 = 1. Fusion weight coefficient where s r Represents the reliability index of multi-array phased array radar data, s o Represents the reliability index of data from other data sources. Here, D r is the multi-array phased array radar data set, D o is a collection of data from other data sources, D f is the fused data set, d n , d′ m 、f p are the elements in the corresponding data set, ω 2 ,ω 2 is the fusion weight coefficient, and p is the total number of data after fusion.
[0017] Furthermore, the data acquisition module can communicate data with the multi-array phased array radar in real time to ensure the timeliness and accuracy of the data.
[0018] Furthermore, the deep learning model is a neural network model trained based on a large amount of historical target data, and the training data of the model includes the position, speed, acceleration and corresponding future trajectory labels of different types of targets.
[0019] Furthermore, it also includes a dynamic adjustment module:
[0020] This module dynamically adjusts the system parameters according to the target's motion state and environmental changes. Suppose the target motion state parameter is S = {s 1 ,s 2 ,…,s l}, the environmental parameter is E = {e 1 ,e 2 ,…,e m}, adjust the function to f adjust (S,E) = {a 1 (S,E),a 2 (S,E),…,a n (S,E)}, where a1 (S,E),a 2 (S,E),…,a n (S, E) represents the adjustment of the 1st to nth system parameters according to the target motion state parameters and environmental parameters. S is the set of target motion state parameters, s l is an element in the set, E is the set of environmental parameters, e 1 ,e 2 ,...,e m is an element in the set, f adjust (S,E) is the adjustment function, l is the total number of target motion state parameters, m is the total number of environmental parameters, and n is the total number of system parameters.
[0021] Further, including visual analysis module:
[0022] This module visualizes the target trajectory prediction results. Let the prediction result be R and the visualization function be g vis (R) = {V 1 (R), V 2 (R),…,V k (R)}, where V k (R) represents the kth visualization method of the prediction result R, such as a three-dimensional trajectory diagram and a speed curve. Through the visualization analysis module, users can understand the target trajectory prediction results more intuitively.
[0023] Furthermore, it has an anomaly detection module:
[0024] This module performs anomaly detection on the input data. Assume that the input data is D = {d 1 , d 2 , ..., d n}, the anomaly detection function is h detect (D) = {b 1 , b 2 , …, b n},in If abnormal data is detected, the module will issue an alarm and take corresponding measures.
[0025] Furthermore, it contains a data storage optimization module:
[0026] This module is responsible for optimizing the data storage method, improving storage efficiency and data retrieval speed. Suppose the amount of stored data is S, and the optimization function is g store (S) = {S 1 , S 2 , …, S r}, where S rIndicates an optimized storage method. For example, you can use compression storage technology to reduce storage space, or create indexes to increase data retrieval speed.
[0027] Furthermore, a multi-array phased array radar target trajectory prediction method comprises the following steps:
[0028] S1. Receive target data detected by the multi-array phased array radar in real time through the data acquisition module. These data include the position, speed and acceleration information of the target;
[0029] S2. Use the data preprocessing module to clean and standardize the collected data, remove outliers and noise interference, and ensure the purity and standardization of the data;
[0030] S3, the trajectory prediction module calls the pre-trained deep learning model to perform trajectory prediction on the processed data, which is used to process time series data and predict the future position, speed and acceleration information of the target;
[0031] S4. After completing the trajectory prediction, the prediction results are output to the user in an intuitive manner through the result output module, such as graphical display and data report form, to meet the user's requirements for different precisions;
[0032] S5, intelligent optimization module automatically adjusts system parameters and model structure according to user feedback on prediction results and actual application effects, so as to improve the accuracy and practicality of prediction;
[0033] S6, the multi-source fusion module integrates data from multi-array phased array radars and other data sources to achieve more comprehensive and accurate target trajectory prediction;
[0034] S7, the dynamic adjustment module will dynamically adjust the system parameters according to the target's motion state and environmental changes to adapt to different situations;
[0035] S8, the visualization analysis module performs visualization analysis on the prediction results, such as generating a three-dimensional trajectory diagram and a speed curve, so that users can understand the target trajectory prediction results more intuitively;
[0036] S9, the anomaly detection module performs anomaly detection on the input data, and once abnormal data is found, an alarm is issued and corresponding processing is performed;
[0037] S10. The data storage optimization module optimizes the data storage method to improve storage efficiency and data retrieval speed.
[0038] Compared with the prior art, the present invention has the following beneficial effects:
[0039] The present invention receives target data detected by a multi-array phased array radar in real time through a data acquisition module, and has an adaptive data acquisition function, which can automatically adjust acquisition parameters according to target dynamic characteristics and radar environment, thereby ensuring the accuracy and integrity of the data.
[0040] The data preprocessing module uses advanced data cleaning algorithms and standardization processing to remove outliers and noise interference, providing a high-quality data foundation for subsequent trajectory prediction.
[0041] The trajectory prediction module uses a pre-trained deep learning model to effectively process time series data and accurately predict the target's future position, speed, acceleration, etc. At the same time, it introduces an uncertainty estimation mechanism to provide users with a more comprehensive decision-making basis.
[0042] The result output module outputs the prediction results to the user in an intuitive way, meeting different accuracy requirements and improving the user experience.
[0043] The intelligent optimization module automatically adjusts system parameters and model structure according to user feedback and actual results, continuously improving the accuracy and practicality of predictions.
[0044] The multi-source fusion module integrates data from multi-faceted phased array radars and other data sources, giving full play to the advantages of different data sources to achieve more comprehensive and accurate target trajectory prediction.
[0045] The dynamic adjustment module dynamically adjusts system parameters according to the target motion state and environmental changes, so that the system can better adapt to different situations.
[0046] The visualization analysis module performs visualization analysis on the prediction results, allowing users to understand the target trajectory prediction results more intuitively, facilitating decision-making and analysis.
[0047] The anomaly detection module performs anomaly detection on input data, discovers and handles abnormal situations in a timely manner, and ensures the stability and reliability of the system.
[0048] The data storage optimization module optimizes the data storage mode, improves storage efficiency and data retrieval speed, and provides support for the efficient operation of the system. In summary, the multi-array phased array radar target trajectory prediction method and system of the present invention have brought new breakthroughs and developments to the field of target trajectory prediction. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] Figure 1 A schematic block diagram of a multi-array phased array radar target trajectory prediction system proposed by the present invention;
[0050] Figure 2 This is a schematic block diagram of the multi-array phased array radar target trajectory prediction method proposed in the present invention. DETAILED DESCRIPTION
[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0052] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise" and "counterclockwise" indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.
[0053] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, and it can be the internal connection of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. The present invention will be further described in detail below in conjunction with the accompanying drawings.
[0054] Reference Figure 1-2 :A multi-array phased array radar target trajectory prediction system, comprising:
[0055] The data acquisition module is used to receive the target data detected by the multi-array phased array radar. The target data includes the position, velocity, and acceleration information of the target, which can be expressed as a vector T = (P, V, A, ...), where P represents the target position data, V represents the target velocity data, and A represents the target acceleration data. The module can receive radar data in real time and preliminarily organize and store the data. At the same time, the module has an adaptive data acquisition function, which can automatically adjust the acquisition parameters according to the dynamic characteristics of the target and the radar environment to ensure the accuracy and integrity of the data.
[0056] The data preprocessing module cleans and preprocesses the collected target data. It uses advanced data cleaning algorithms to remove outliers and noise interference to ensure the purity of the data. 1 , x 2 , …, x n}, the denoising function is h(X) = {x′ 1 , x′ 2 , …, x′ n},in The sliding average method is used for denoising, m is the window size, here, h(X) represents the denoising function, X is the original data set, x 1 , x 2 , …, x n is the 1-nth data point in the original data, x′ 1 , x′ 2 , …, x′ n is the 1-nth data point after denoising, and then normalized. Let the normalized data be Y={y 1 ,y 2 , …, y n},in μ is the mean of the denoised data, and σ is the standard deviation of the denoised data. Here, Y is the standardized data set, y i are the normalized data points.
[0057] The trajectory prediction module uses a pre-trained deep learning model to predict the trajectory of the pre-processed target data. The model adopts a recurrent neural network (RNN) architecture, which can effectively process time series data. Assume that the input data is I = (i 1 ,i 2 ,…,i t ), where i 1 ,i 2 ,…,i t represents the target data vector at time 1-t, and the predicted output is O=(o 1 , o 2 ,…,o t+1 ), where o 1 , o 2 ,…,o t+1 Indicates the predicted position of the target at time 1-t+1. The model predicts the future position, velocity, and acceleration information of the target by learning from historical data. The training data of the model includes a large amount of historical target data and the corresponding future real trajectory data. In addition, this module introduces an uncertainty estimation mechanism to quantitatively evaluate the uncertainty of the prediction results, providing users with a more comprehensive decision-making basis.
[0058] This module has a powerful graphic display function. For graphic display, advanced visualization technology is used to generate clear and accurate two-dimensional and three-dimensional trajectory diagrams. In the two-dimensional trajectory diagram, the target's motion trajectory is displayed on a coordinate plane. The target positions at different time points are represented by markers of different colors and connected by lines to intuitively present the target's moving path. At the same time, key geographical location information, obstacle positions, etc. can also be marked on the diagram to help users better understand the target's environment. In the three-dimensional trajectory diagram, the target's motion trajectory is displayed in a three-dimensional space. Users can observe the target's motion from different angles and have a more comprehensive understanding of the target's position changes. In addition, according to user needs, the target's trajectory change process can be dynamically displayed, and the user can feel the target's motion trend more intuitively through animation effects.
[0059] In addition to graphical display, the result output module can also output prediction results in the form of data reports. The data report lists in detail the specific values of the target's position, speed, acceleration, etc. at different time points, and classifies and statistically analyzes them. For example, the target's position data is arranged in chronological order, and statistical indicators such as the average speed and maximum acceleration of the target in different time periods are calculated to provide users with a more comprehensive data analysis. At the same time, the data report can also be customized according to user needs. Users can select specific time periods and specific target attributes for query and analysis to meet the personalized needs of different users.
[0060] At the same time, the module is highly flexible and can provide prediction results with different precisions according to user needs. For users who need high-precision prediction results, the module can provide detailed trajectory prediction data, including the position change and speed change of the target in each tiny time interval. These high-precision data can be used to accurately track and analyze the target, and are suitable for scenarios with high prediction accuracy requirements, such as military target tracking, aerospace, etc. For users who only need to have a general understanding of the target's movement trend, the module can provide a simplified version of the prediction results, such as the target's approximate movement direction, possible areas to reach, etc. This way of displaying prediction results with different precisions not only meets the needs of professional users for high-precision data, but also facilitates ordinary users to quickly understand the movement of the target, improving the practicality of the system and user experience.
[0061] The intelligent optimization module automatically adjusts system parameters and model structure according to user feedback on prediction results and the effects in actual applications to improve the accuracy and practicality of prediction. Assume that user feedback is R and the optimization function is where f j (R) represents the adjustment function of the jth parameter according to user feedback R, α j is the corresponding weight coefficient, satisfying g(R) is the optimization function and k is the total number of parameters.
[0062] Multi-source fusion module, which can integrate the data of multi-array phased array radar and other sensors or data sources to achieve more comprehensive and accurate target trajectory prediction. Suppose the multi-array phased array radar data is D r =(d 1 , d 2 , …, d n ), other data sources are D o =(d′ 1 , d′ 2 ,…,d′ m ), the fused data is D f =(f 1 , f 2 , ..., f p ), where f i =ω 1 d i +ω 2 d′ i％m+1 ,ω 1 +ω 2 = 1. Fusion weight coefficient where s r Represents the reliability index of multi-array phased array radar data, s o Represents the reliability index of data from other data sources. Here, D r is the multi-array phased array radar data set, D o is a collection of data from other data sources, D f is the fused data set, d n , d′ m 、f p are the elements in the corresponding data set, ω 1 ,ω 2 is the fusion weight coefficient, and p is the total number of data after fusion.
[0063] In the present invention, the data acquisition module plays a key role in ensuring data quality. The module can communicate with multi-array phased array radar in real time, and achieve efficient data transmission through advanced communication protocols and technical means. For example, a high-speed data bus and a dedicated data interface are used to ensure the stability and reliability of data during transmission. At the same time, the data acquisition module is also equipped with data caching and preprocessing functions, which can preliminarily organize and screen the received data, remove invalid data and noise interference, and improve the quality and availability of data.
[0064] In order to ensure the timeliness of data, the data acquisition module uses real-time data acquisition technology, which can obtain the target data detected by the radar in the shortest time. For example, by setting the trigger conditions and frequency of data acquisition, the time interval of data acquisition is automatically adjusted according to the target's motion state and the radar's scanning cycle, ensuring that the latest position, speed, acceleration and other information of the target can be obtained in time. In addition, the data acquisition module also has a data synchronization function, which can synchronize the time and align the data collected by different radar sites or different sensors in space to ensure the consistency and accuracy of the data.
[0065] In the present invention, the deep learning model is the core component for realizing target trajectory prediction. The model is a neural network model trained based on a large amount of historical target data. By learning from historical data, it can automatically extract the characteristics and laws of target movement and realize accurate trajectory prediction.
[0066] During the model training process, a large amount of historical target data was used, including the position, velocity, acceleration and corresponding future trajectory labels of different types of targets. These data were carefully sorted and labeled to ensure the accuracy and reliability of the data. In order to improve the generalization ability of the model, data enhancement techniques such as random rotation, translation, and scaling were also used to increase the diversity and complexity of the data.
[0067] The deep learning model uses advanced neural network architectures, such as recurrent neural networks (RNNs) and long short-term memory networks (LSTMs), which can effectively process time series data and capture the dynamic characteristics of target motion. At the same time, optimization algorithms and loss functions, such as stochastic gradient descent (SGD) and mean square error loss functions, are also used to continuously adjust the model parameters and improve the model's prediction accuracy. In addition, in order to prevent the model from overfitting, regularization techniques, such as L1 and L2 regularization and Dropout, are also used to increase the stability and generalization ability of the model.
[0068] The present invention also includes a dynamic adjustment module:
[0069] This module dynamically adjusts the system parameters according to the target's motion state and environmental changes. Suppose the target motion state parameter is S = {s 1 ,s 2 ,…,s l}, the environmental parameter is E = {e 1 ,e 2 ,…,e m}, adjust the function to f adjust (S,E) = {a 1 (S,E),a 2 (S,E),…,a n (S,E)}, where a 1(S,E),a 2 (S,E),...,a n (S, E) represents the adjustment of the 1st to nth system parameters according to the target motion state parameters and environmental parameters. S is the set of target motion state parameters, s l is an element in the set, E is the set of environmental parameters, e 1 , e 2 ,…,e m is an element in the set, f adjust (S, E) is the adjustment function, l is the total number of target motion state parameters, m is the total number of environmental parameters, and n is the total number of system parameters.
[0070] The present invention also includes a visual analysis module:
[0071] This module visualizes the target trajectory prediction results. Let the prediction result be R and the visualization function be g vis (R) = {V 1 (R), V 2 (R),…,V k (R)}, where V k (R) represents the kth visualization method of the prediction result R, such as a three-dimensional trajectory diagram and a speed curve. Through the visualization analysis module, users can understand the target trajectory prediction results more intuitively.
[0072] The present invention also has an anomaly detection module:
[0073] This module performs anomaly detection on the input data. Assume that the input data is D = {d 1 , d 2 , …, d n}, the anomaly detection function is h detect (D) = {b 1 , b 2 , …, b n},in If abnormal data is detected, the module will issue an alarm and take corresponding measures.
[0074] The present invention also includes a data storage optimization module:
[0075] This module is responsible for optimizing the data storage method, improving storage efficiency and data retrieval speed. Suppose the amount of stored data is S, and the optimization function is g store (S) = {S 1 , S 2 , …, S r}, where S r Indicates an optimized storage method. For example, you can use compression storage technology to reduce storage space, or create indexes to increase data retrieval speed.
[0076] The present invention discloses a multi-array phased array radar target trajectory prediction method, comprising the following steps:
[0077] S1. Receive target data detected by the multi-array phased array radar in real time through the data acquisition module. These data include the position, speed and acceleration information of the target;
[0078] S2. Use the data preprocessing module to clean and standardize the collected data, remove outliers and noise interference, and ensure the purity and standardization of the data;
[0079] S3, the trajectory prediction module calls the pre-trained deep learning model to perform trajectory prediction on the processed data, which is used to process time series data and predict the future position, speed and acceleration information of the target;
[0080] S4. After completing the trajectory prediction, the prediction results are output to the user in an intuitive manner through the result output module, such as graphical display and data report form, to meet the user's requirements for different precisions;
[0081] S5, intelligent optimization module automatically adjusts system parameters and model structure according to user feedback on prediction results and actual application effects, so as to improve the accuracy and practicality of prediction;
[0082] S6, the multi-source fusion module integrates data from multi-array phased array radars and other data sources to achieve more comprehensive and accurate target trajectory prediction;
[0083] S7, the dynamic adjustment module will dynamically adjust the system parameters according to the target's motion state and environmental changes to adapt to different situations;
[0084] S8, the visualization analysis module performs visualization analysis on the prediction results, such as generating a three-dimensional trajectory diagram and a speed curve, so that users can understand the target trajectory prediction results more intuitively;
[0085] S9, the anomaly detection module performs anomaly detection on the input data, and once abnormal data is found, an alarm is issued and corresponding processing is performed;
[0086] S10. The data storage optimization module optimizes the data storage method to improve storage efficiency and data retrieval speed.
[0087] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can replace or change the technical solution and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. A multi-array phased array radar target trajectory prediction system, characterized in that: include: The data acquisition module is used to receive the target data detected by the multi-array phased array radar. The target data includes the position, velocity and acceleration information of the target, which is expressed as a vector T = (P, V, A, ...), where P represents the target position data, V represents the target velocity data, and A represents the target acceleration data. It has an adaptive data acquisition function and automatically adjusts the acquisition parameters according to the dynamic characteristics of the target and the radar environment; The data preprocessing module cleans and preprocesses the collected target data, and uses the data cleaning algorithm to remove outliers and noise interference. Suppose the original data is X = {x1, x2, ..., x n }, the denoising function is h(X)={x′1,x′2,…,x′ n },in The sliding average method is used for denoising, m is the window size, h(X) represents the denoising function, X is the original data set, x1, x2,…, x n is the 1st to nth data point in the original data, x′1,x′2,…,x′ n is the 1st to nth data point after denoising, and then normalizes it. Let the normalized data be Y=(y1,y2,…,y n ),in μ is the mean of the denoised data, σ is the standard deviation of the denoised data, Y is the standardized data set, y i is the normalized data point; The trajectory prediction module uses a pre-trained deep learning model to predict the trajectory of the pre-processed target data. The model adopts a recurrent neural network RNN architecture. The input data is I = (i1, i2, ..., i t ), where i1,i2,…,i t represents the target data vector at time 1-t, and the predicted output is O=(o1,o2,…,o t+1 ), where o1, o2, …, o t+1 Represents the predicted position of the target at time 1-t+1. The model predicts the future position, velocity and acceleration information of the target by learning from historical data. The training data of the model includes historical target data and the corresponding future real trajectory data. The uncertainty estimation mechanism is introduced to quantitatively evaluate the uncertainty of the prediction results. The result output module outputs the trajectory prediction results to the user, including graphical display and data report form, and also provides prediction result displays of different accuracy according to user needs; The intelligent optimization module automatically adjusts system parameters and model structure based on user feedback on prediction results and the effects in actual applications. Let user feedback be R and the optimization function be where f j (R) represents the adjustment function of the jth parameter according to user feedback R, α j is the corresponding weight coefficient, satisfying g(R) is the optimization function, k is the total number of parameters; Multi-source fusion module, which can integrate the data of multi-array phased array radar and other sensors or data sources. Suppose the multi-array phased array radar data is D r =(d1,d2,…,d n ), other data sources are D o =(d′1, d′2, …, d′ m ), the fused data is D f =(f1,f2,…,f p ), where f i =ω1d i +ω2d′ i % m+1 , ω1+ω2=1, fusion weight coefficient where s r Represents the reliability index of multi-array phased array radar data, s o Represents the reliability index of data from other data sources, D r is the multi-array phased array radar data set, D o is a collection of data from other data sources, D f is the fused data set, d n , d′ m 、f p They are the elements in the corresponding data set, ω1 and ω2 are the fusion weight coefficients, and p is the total number of data after fusion.
2. The multi-array phased array radar target trajectory prediction system according to claim 1, characterized in that: The data acquisition module can communicate with the multi-array phased array radar in real time to ensure the timeliness and accuracy of the data.
3. The multi-array phased array radar target trajectory prediction system according to claim 1, characterized in that: The deep learning model is a neural network model trained based on a large amount of historical target data. The training data of the model includes the position, speed, acceleration and corresponding future trajectory labels of different types of targets.
4. The multi-array phased array radar target trajectory prediction system according to claim 1, characterized in that: Also includes dynamic adjustment modules: This module dynamically adjusts the system parameters according to the target's motion state and environmental changes. Suppose the target motion state parameters are S = {s1, s2, ..., s l }, the environmental parameters are E = {e1, e2, …, e m }, adjust the function to f adjust (S,E)={a1(S,E),a2(S,E),…,a n (S,E)}, where a1(S,E),a2(S,E),...,a n (S, E) represents the adjustment of the 1st to nth system parameters according to the target motion state parameters and environmental parameters, S is the set of target motion state parameters, s l are elements in the set, E is the set of environmental parameters, e1, e2, ..., e m is an element in the set, f adjust (S,E) is the adjustment function, l is the total number of target motion state parameters, m is the total number of environmental parameters, and n is the total number of system parameters.
5. The multi-array phased array radar target trajectory prediction system according to claim 1, characterized in that: Includes visual analysis modules: This module visualizes the target trajectory prediction results. Let the prediction result be R and the visualization function be g. vis (R)={V1(R), V2(R),…,V k (R)}, where V k (R) represents the kth visualization method of the prediction result R, including the three-dimensional trajectory diagram and the speed curve. Through the visualization analysis module.
6. The multi-array phased array radar target trajectory prediction system according to claim 1, characterized in that: Equipped with anomaly detection module: This module performs anomaly detection on the input data. Suppose the input data is D={d1,d2,...,d n }, the anomaly detection function is h detect (D) = {b1,b2,...,b n },in If abnormal data is detected, the module will issue an alarm and take corresponding measures.
7. The multi-array phased array radar target trajectory prediction system according to claim 1, characterized in that: Contains data storage optimization module: This module is responsible for optimizing the data storage method to improve storage efficiency and data retrieval speed. Suppose the amount of stored data is S and the optimization function is g store (S) = {S1, S2, ..., Sr}, where S r Indicates an optimized storage method that uses compression storage technology to reduce storage space or creates indexes to increase data retrieval speed.
8. A method for using a multi-array phased array radar target trajectory prediction system according to any one of claims 1 to 7, characterized in that: The following steps are involved: S1. Receive target data detected by the multi-array phased array radar in real time through the data acquisition module. These data include the position, speed and acceleration information of the target; S2. Use the data preprocessing module to clean and standardize the collected data, remove outliers and noise interference, and ensure the purity and standardization of the data; S3, the trajectory prediction module calls the pre-trained deep learning model to perform trajectory prediction on the processed data, which is used to process time series data and predict the future position, speed and acceleration information of the target; S4. After completing the trajectory prediction, the prediction results are output to the user in an intuitive manner through the result output module, such as graphical display and data report form, to meet the user's requirements for different precisions; S5, intelligent optimization module automatically adjusts system parameters and model structure according to user feedback on prediction results and actual application effects, so as to improve the accuracy and practicality of prediction; S6, the multi-source fusion module integrates data from multi-array phased array radars and other data sources to achieve more comprehensive and accurate target trajectory prediction; S7, the dynamic adjustment module will dynamically adjust the system parameters according to the target's motion state and environmental changes to adapt to different situations; S8, the visualization analysis module performs visualization analysis on the prediction results, such as generating a three-dimensional trajectory diagram and a speed curve, so that users can understand the target trajectory prediction results more intuitively; S9, the anomaly detection module performs anomaly detection on the input data, and once abnormal data is found, an alarm is issued and corresponding processing is performed; S10. The data storage optimization module optimizes the data storage method to improve storage efficiency and data retrieval speed.