Runway foreign object identification system and method based on millimeter wave radar orthogonal sampling

Abstract

The invention relates to the technical field of airport runway foreign object detection, and discloses a runway foreign object recognition system and method based on millimeter wave radar orthogonal sampling, and the recognition system comprises a radar signal collection module which is used for receiving echo signals reflected from a runway; the signal processing module is used for carrying out denoising, phase recovery and multi-path effect removal processing on the echo signal; the AI target identification module is used for carrying out target identification on the processed signals and identifying foreign objects on the runway; and the real-time processing and feedback module is used for feeding back the target identification result to the management system for subsequent processing. Through combination of millimeter wave radar orthogonal sampling and deep learning, high-precision foreign object identification is provided, echo signal phase information is recovered, and the material and form distinguishing capability is significantly improved. And meanwhile, real-time feedback is realized by utilizing GPU acceleration and a high-speed communication protocol, so that the response speed is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of airport runway foreign object detection, and in particular to a runway foreign object recognition system and method based on millimeter wave radar orthogonal sampling. Background Art

[0002] The main purpose of the airport runway foreign object detection system is to ensure the safety of aircraft during takeoff and landing. Foreign objects (FOD) on the runway, such as rocks, garbage or tools, can easily cause damage to aircraft and even cause accidents in serious cases. Therefore, accurate and efficient detection and timely removal of foreign objects on the runway are important links in ensuring aviation safety.

[0003] Traditional radar detection systems rely on the amplitude information of echo signals to identify targets. However, amplitude information is often insufficient to provide comprehensive characteristics of the target, especially in complex scenes, and cannot effectively distinguish foreign objects of different materials, shapes or sizes. The limitations of this method make it difficult for the system to identify targets with complex shapes and similar reflective characteristics, especially in changing weather and environmental conditions, and the recognition accuracy is greatly reduced.

[0004] The information feedback link in traditional systems usually has a lag. After the target is detected, the signal transmission and processing process is relatively slow, and key information cannot be transmitted to the airport management system in time. This leads to a slow response speed for clearing foreign objects on the runway in an emergency, which may affect the normal operation of flights.

[0005] Many existing target recognition technologies only focus on the static features of the target and lack the analysis of the dynamic changes of the target. On the runway, foreign objects may move with the wind or mechanical equipment. Existing systems cannot effectively cope with these dynamic changes and cannot recognize foreign objects in motion in real time, thus reducing the overall recognition ability of the system.

[0006] Traditional detection systems rely on manual intervention to perform cleaning tasks. Manual operation is not only inefficient but also prone to errors. Especially in complex runway environments, the real-time and accuracy of manual intervention is difficult to guarantee, affecting the operating efficiency and safety of the entire airport. Summary of the invention

[0007] In view of the shortcomings of the prior art, the present invention provides a runway foreign object recognition system and method based on millimeter-wave radar orthogonal sampling, which solves the problems of low recognition accuracy, delayed response and insufficient cleaning efficiency in the existing runway foreign object detection system due to the lack of effective phase information extraction and dynamic target recognition capabilities.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: A runway foreign object recognition system based on millimeter wave radar orthogonal sampling, comprising: a radar signal acquisition module for receiving an echo signal reflected from a runway; a signal processing module for performing denoising, phase recovery and multipath effect removal processing on the echo signal; AI target recognition module, used to perform target recognition on the processed signals and identify foreign objects on the runway; The real-time processing and feedback module is used to feed back the target recognition results to the management system for subsequent processing.

[0009] Preferably, the signal processing module includes: Denoising module, used to remove noise from the signal through the Wiener filtering algorithm; A phase recovery module, used to recover the phase information of the echo signal according to the ratio of the in-phase component to the orthogonal component; The multipath effect removal module is used to remove the multipath effect in the signal by using the Kalman filter algorithm.

[0010] Preferably, the phase recovery module recovers the phase of the echo signal by calculating the ratio between the in-phase component and the orthogonal component.

[0011] Preferably, the AI ​​target recognition module includes: Convolutional neural network module, used to extract spatial features from radar signals; The recurrent neural network module is used to analyze time series information and identify the target's motion trajectory.

[0012] Preferably, the AI ​​target recognition module also includes a data enhancement and transfer learning module, which is used to enhance the training data and accelerate the training process of the FOD recognition model by using a deep learning model pre-trained in other fields.

[0013] Preferably, the real-time processing and feedback module processes the target recognition results in real time through GPU acceleration technology, and feeds back the processing results to the airport management system.

[0014] A runway foreign object recognition method based on millimeter-wave radar orthogonal sampling includes: S1. Receive the echo signal reflected from the runway; S2. De-noising, phase recovery and multipath removal of the echo signal; S3. Perform target recognition based on the processed signal to identify foreign objects on the runway; S4. Feedback the target recognition results to the management system for subsequent processing.

[0015] Preferably, the denoising step removes noise in the echo signal by using a Wiener filtering algorithm.

[0016] Preferably, the phase recovery step includes recovering the phase information of the echo signal by calculating the ratio of the in-phase component to the orthogonal component.

[0017] Preferably, the multipath effect removal step removes the multipath effect in the echo signal by using a Kalman filter algorithm.

[0018] The present invention provides a runway foreign object recognition system and method based on millimeter wave radar orthogonal sampling. It has the following beneficial effects: 1. The present invention adopts a technical solution combining orthogonal sampling of millimeter-wave radar with deep learning, achieving the technical effect of high-precision recognition of foreign objects on the runway. Compared with the traditional radar recognition method that simply relies on amplitude information in the prior art, the present invention restores the phase information in the echo signal, allowing the system to more accurately distinguish foreign objects of different materials and shapes, solving the problem that traditional methods cannot effectively identify complex foreign objects.

[0019] 2. The present invention uses real-time processing and feedback technology that combines GPU acceleration with high-speed communication protocols to ensure high-speed transmission and real-time response from target identification to management system feedback. Compared with the system with a relatively lagging feedback mechanism in the prior art, the present invention significantly improves the efficiency of information transmission, ensures that the management system can respond in a very short time, and solves the runway safety risks caused by response delays in the existing solutions.

[0020] 3. The present invention combines convolutional neural networks with recurrent neural networks to achieve multi-dimensional analysis of the static and dynamic characteristics of the target, achieving a more comprehensive FOD recognition effect. Compared with the traditional static image recognition method, the present invention can simultaneously process the spatial characteristics and time series characteristics of the target, making the FOD recognition in motion more accurate, solving the limitation that the traditional method can only recognize static targets.

[0021] 4. The present invention achieves the technical effect of automated and intelligent management by feeding back the foreign object identification results to the airport management system in real time and automatically scheduling with the cleaning equipment. Compared with the runway maintenance scheme that relies on manual intervention in the prior art, the present invention greatly improves the operation efficiency and accuracy, reduces manual errors, and ensures the timeliness and accuracy of runway cleaning work. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a schematic diagram of the system architecture of the present invention; Figure 2 It is a schematic diagram of the signal processing flow of the present invention; Figure 3It is a schematic diagram of the target identification method flow of the present invention; Figure 4 It is a schematic diagram of the real-time feedback process of the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the specification 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.

[0024] Please refer to the attached Figure 1 The embodiment of the present invention provides a runway foreign object recognition system based on millimeter wave radar orthogonal sampling, including a radar signal acquisition module, a signal processing module, an AI target recognition module, and a real-time processing and feedback module. Each module works together to ensure that the system can quickly and accurately detect and identify foreign objects on the runway.

[0025] The radar signal acquisition module of the present invention is used to receive the echo signal on the runway and serve as the input of the entire system. The signal acquisition module is closely connected with the signal processing module to play the role of signal acquisition and transmission. The module captures the reflected signal from the runway foreign object (FOD) in real time through the millimeter wave radar system, providing basic data for subsequent signal processing and target identification.

[0026] In the above system structure, the radar signal acquisition module is the starting point of the signal processing process. It receives electromagnetic wave signals reflected by different objects (including FOD) on the runway, converts these signals into digital form and transmits them to the signal processing module. The quality of signal acquisition directly affects the accuracy of subsequent signal processing, phase recovery and target identification, so the accuracy and sensitivity of this module are the key to system performance.

[0027] In this embodiment, the radar signal acquisition module includes a receiving antenna and a related signal processing unit. The main function of the receiving antenna is to detect objects in the area through millimeter wave signals and transmit the reflected signals to the back-end processing module. The received echo signal usually contains multiple information dimensions, including the amplitude, phase, distance and speed of the target object.

[0028] Generally, the radar system operates in a specific frequency range to ensure that the reflected signal of the target can be accurately detected. In this embodiment, the millimeter wave radar usually operates in the frequency range of 30 GHz to 300 GHz, which is suitable for high-resolution and long-range target detection. Since foreign objects on the runway are usually small in size, high-frequency millimeter wave radar can provide sufficient spatial resolution.

[0029] As an option, the receiving antenna can achieve higher directivity and coverage by designing multiple antenna arrays. The design of the antenna array can control the beam direction through electronic scanning, improve the sensitivity of the detection area, and thus enhance the signal reception effect. Depending on different implementation requirements, the antenna array can adopt a linear array, a planar array, or a circular array structure.

[0030] Specifically, the signal acquisition module receives radar waves through the receiving antenna and converts them into electrical signals, and then performs preliminary amplification and conversion on these electrical signals through the front-end signal processing unit. The front-end signal processing unit usually includes a low noise amplifier (LNA), a mixer, and an analog-to-digital converter (ADC). In the mixer, the received signal is mixed with the local oscillator signal to generate a baseband signal and a modulated copy of its spectrum. Through the analog-to-digital converter, the signal is converted into a digital signal for further processing.

[0031] In a possible implementation, the received signal is processed in the frequency domain by fast Fourier transform (FFT) to obtain spectrum information. Specifically, assuming that the received signal is a complex signal: s(t)=A(t)e jθ(t) Among them, A(t) is the amplitude information of the signal, which indicates the echo intensity of the target, and θ(t) is the phase information of the signal, which indicates the relative position and reflection characteristics of the target.

[0032] After conversion, the amplitude information and phase information of the signal are stored separately to provide raw data for subsequent signal processing modules.

[0033] In the process of radar signal acquisition, orthogonal sampling technology is applied to signal processing. The goal of orthogonal sampling is to decompose the echo signal into an in-phase component (I) and a quadrature component (Q), so that the phase information of the signal can be retained. Specifically, by performing orthogonal sampling on the received signal after it is received, the amplitude and phase information of the signal can be extracted.

[0034] In this implementation, the signal can be expressed by the following formula: s(t)=A(t)e jθ(t) =A(t)[cos(θ(t))+jsin(θ(t))+jsin(θ(t))] The received signal is decomposed into the in-phase component I(t) and the quadrature component Q(t): I(t)=A(t)cos(θ(t)), Q(t)=A(t)sin(θ(t)) These in-phase components and orthogonal components can be obtained through different circuit designs, usually using digital mixing and sampling technology, and converting the signal into a digital signal using a high-frequency sampling circuit.

[0035] The output of the signal acquisition module is based on the amplitude and phase information of the echo signal. The collected signals will undergo preliminary analog processing, then converted into digital signals through ADC, and then passed to the signal processing module. In the signal processing module, these signals will be processed for denoising, phase recovery, and multipath effect removal.

[0036] The accuracy of signal acquisition is closely related to the design of the system. In order to ensure the integrity and accuracy of the signal, the design of the signal acquisition module needs to meet the requirements of high sensitivity and high sampling rate. In general, the sampling rate should be at least twice the bandwidth of the target signal to avoid information loss. For foreign objects moving at high speed, the sampling rate and processing rate must be high enough to accurately capture their dynamic characteristics.

[0037] In this embodiment, the amplitude A(t) and phase θ(t) of the signal are obtained through orthogonal sampling. Specifically, the amplitude A(t) is the strength of the echo signal, which reflects the reflection characteristics of the target, while the phase θ(t) contains key information related to the relative position and movement of the target. By decomposing the signal into in-phase and orthogonal components, the system can better extract this information and provide basic data for subsequent phase recovery and target recognition.

[0038] Signal processing relies on precise sampling and conversion techniques to ensure that the radar system can capture the detailed characteristics of the target in a complex environment. In different implementations, parameters such as sampling frequency, sensor sensitivity, antenna design, etc. can be adjusted as needed to further improve system performance.

[0039] Through the implementation of the above-mentioned embodiment, the radar signal acquisition module can work stably in different environments and accurately capture the echo signal from foreign objects on the runway. Through orthogonal sampling and signal decomposition, the system can extract richer signal features and provide accurate data support for subsequent signal processing and target identification. The design and implementation of this module ensures the high accuracy and efficiency of the runway foreign object identification system of the present invention.

[0040] Please refer to the attached Figure 2 , This signal processing module is a key component of the system of the present invention. It is closely connected with the aforementioned radar signal acquisition module and is responsible for a series of processing of the collected original echo signal, including denoising, phase recovery, and multipath effect removal, etc., to ensure the accuracy of subsequent target recognition. The main task of this module is to convert the signal received from the radar signal acquisition module into high-quality data that can be used for analysis, and provide clear and reliable input signals for the AI ​​target recognition module.

[0041] In the aforementioned technologies, the signals collected by the radar signal acquisition module usually contain amplitude information and phase information, which may be affected by noise interference, phase distortion and multipath effects if not processed. Therefore, the main function of the signal processing module is to pre-process these signals to make them suitable for subsequent target recognition and classification.

[0042] In this embodiment, the signal processing module includes a denoising module, a phase recovery module and a multipath effect removal module. Each module is responsible for different processing tasks, but they work together to improve the quality of the signal to meet the needs of subsequent target recognition.

[0043] Denoising module: In practical applications, radar echo signals are often interfered by environmental noise, especially in complex runway environments. The main task of the signal denoising module is to eliminate these noises and ensure that the effective information in the signal can be retained. In some embodiments, the Wiener filtering algorithm is used for signal denoising. The Wiener filtering is an adaptive filtering method based on the minimum mean square error (MSE) criterion, and the parameters of the filter can be dynamically adjusted according to the known statistical characteristics of the signal and noise.

[0044] Specifically, the Wiener filtering algorithm denoises the signal using the following formula: in, is the filtered signal, x(t) is the original signal, and H(t) is the transfer function of the Wiener filter, which is defined as: Among them, S xx (t) is the power spectrum density of the signal, S vv (t) is the power spectral density of the noise. In practical applications, the noise and effective information of the signal usually have different spectral characteristics. Through Wiener filtering, the noise component is effectively removed, thereby retaining a clearer signal.

[0045] Phase recovery module: After signal denoising, the next step is phase recovery. The radar echo signal carries the phase information of the target, which is crucial for accurate target identification. In traditional radar systems, usually only the amplitude information of the echo signal is concerned, while the phase information is ignored. However, phase information is particularly important for distinguishing FOD of different materials. The purpose of the phase recovery module is to recover the lost phase information from the radar signal, thereby providing richer features for subsequent target identification.

[0046] In this embodiment, the phase recovery module recovers the phase of the echo signal by calculating the ratio between the in-phase component (I) and the quadrature component (Q). Assume that the sampled signal is a complex signal, which is expressed as: s(t)=A(t)e jθ(t) Where A(t) is the amplitude of the signal and θ(t) is the phase of the signal. We decompose the signal into the in-phase component I(t) and the quadrature component Q(t), and recover the phase using the following formula: I(t)=A(t)cos(θ(t)),Q(t)=A(t)sin(θ(t)) By calculating: The precise phase information of the target can be obtained. This process will provide the necessary phase data for subsequent target recognition and help the system better distinguish different foreign objects (FOD).

[0047] Multipath effect removal module: The multipath effect is a phenomenon in which radar signals are reflected or refracted due to multiple reflection sources or obstacles during propagation, resulting in interference in echo signals. This effect can affect signal quality, especially in complex runway environments. In order to eliminate the interference caused by the multipath effect, this embodiment uses a Kalman filter algorithm to optimize signal processing.

[0048] Kalman filtering is a recursive algorithm suitable for state estimation of dynamic systems. The algorithm is based on the motion model of the target and uses the current state and noise information of the system to predict the next state of the signal and make corrections. The basic formula of Kalman filtering is as follows: x k =Ax k-1 +Bu k +w k z k =Hx k +v k in: x k is the system state; A is the state transfer matrix; B is the control input matrix; u k is the control input; w k is the process noise; z k is the observation value; H is the observation matrix; v k is the observation noise.

[0049] Through Kalman filtering, the system can effectively eliminate signal deviations caused by multipath effects, optimize echo signals, and make phase information and target signals more accurate.

[0050] The signal processing module is closely connected with the radar signal acquisition module and the AI ​​target recognition module. The radar signal acquisition module transmits the collected raw echo signal to the signal processing module, which converts the signal into high-quality data suitable for target recognition based on operations such as denoising, phase recovery and multipath effect removal. The processed signal will be transmitted to the AI ​​target recognition module for further feature extraction and target classification.

[0051] The entire processing flow is optimized step by step, and each step of the signal processing module provides more accurate information for subsequent target identification. The denoising operation eliminates interference and ensures signal clarity; the phase recovery module provides important phase information for target identification; and the Kalman filter eliminates signal distortion caused by environmental factors and ensures data accuracy.

[0052] The signal processing module plays a crucial role in the present invention. Through the steps of denoising, phase recovery and multipath effect removal, the signal processing module can provide high-quality input data for subsequent target recognition. Through these processes, the system can accurately identify foreign objects on the runway, thereby ensuring flight safety. The implementation of this module relies on advanced signal processing technology and algorithms to ensure the efficiency and accuracy of the entire recognition process.

[0053] Please refer to the attached Figure 3 , this AI target recognition module is a crucial part of the present invention. Its main function is to analyze and process the signal transmitted by the signal processing module, and finally complete the task of identifying the runway foreign object (FOD). This module is closely connected with the aforementioned signal processing module. It receives the signal after denoising, phase recovery and multipath effect removal, and then extracts the features in the signal through deep learning technology, analyzes the shape, size, material and other attributes of the target, so as to accurately determine whether the target is FOD. The core of this module is the deep learning model, which includes the combination of convolutional neural network (CNN) and recurrent neural network (RNN).

[0054] The signal processing module provides high-quality input data for the AI ​​target recognition module. The signal after denoising, phase recovery and multipath effect removal contains the spatial characteristics and time series characteristics of the target. These characteristics are the basis for the AI ​​model to perform target recognition. The function of the AI ​​target recognition module is to convert these signals into specific target categories, and then determine whether there is FOD on the runway.

[0055] In this embodiment, the AI ​​target recognition module uses a combination of convolutional neural network (CNN) and recurrent neural network (RNN). CNN is used to extract spatial features in signal images to help the system identify static features such as the shape and size of FOD. RNN is mainly used to process time series features to help the system analyze the motion state and change trend of FOD. By combining these two networks, the present invention can simultaneously process the static and dynamic information of the target, significantly improving the accuracy of FOD recognition.

[0056] In this embodiment, a convolutional neural network (CNN) is used to extract the spatial features of the target. Specifically, the convolution layer of the CNN performs convolution operations on the signal image through multiple convolution kernels to extract local features in the image, such as the edge, corner, texture and other information of the target. The pooling layer further reduces the dimension of the feature map, making the features more prominent and reducing the amount of calculation.

[0057] The mathematical representation of the convolutional layer is: O(t)=I(t)*W(t) Among them, O(t) is the output feature map, I(t) is the input signal (i.e., the image representation of the radar echo signal), W(t) is the convolution kernel, and * represents the convolution operation.

[0058] The function of the convolution layer is to extract local features in the signal through multiple convolution kernels. Each convolution kernel focuses on different areas in the image, thereby extracting different features. After the convolution operation, the pooling layer downsamples the feature map to reduce the data dimension but retain the most important feature information. This process can effectively improve the model's ability to recognize target features.

[0059] For FOD recognition, the spatial features extracted by CNN help the system identify basic information such as the shape, size, and texture of the object. This information is the basis for determining whether FOD belongs to a certain type of substance.

[0060] Recurrent neural networks (RNNs) are mainly used to process time series features in signals. FOD may move or change on the runway, so RNNs can effectively capture the motion state of the target by processing the time series of signals. For example, when FOD moves quickly or interacts with other objects, RNNs can analyze the target's changing trend and make accurate predictions.

[0061] The recursive formula of RNN is: h t =σ(W hh h t-1 +W hx x t +b h ) y t =Why h t +b y Among them, h t is the hidden state at time t, x t is the input at time t, W hh ,W hx ,W hy is the weight matrix, b h ,b y is the bias term, and σ is the activation function (such as tanh or ReLU).

[0062] The role of RNN is to capture dynamic changes in time series, which is crucial for tracking FOD. By analyzing signal data at multiple moments, RNN can determine the motion state of FOD at different time points and help the system identify whether it is a foreign object on the runway.

[0063] In order to further improve the recognition accuracy of the model, the AI ​​target recognition module also introduces data enhancement and transfer learning technologies.

[0064] Generally speaking, when training deep learning models, data diversity is crucial to improving the generalization ability of the model. Data augmentation technology generates multiple transformed data by rotating, flipping, scaling, and other operations on the original training data, thereby enhancing the diversity of training data. For example, the original radar image can be rotated, translated, and other operations to generate multiple new training samples, so that the model can better adapt to FOD recognition tasks in different environments.

[0065] As an option, transfer learning technology is also applied to the AI ​​target recognition module. Transfer learning uses deep learning models trained in other fields (such as security monitoring), transfers their weight parameters and fine-tunes them to adapt to the task of identifying runway foreign objects. The advantage of transfer learning is that it can accelerate the convergence of the model, reduce training time, and improve the performance of the model on new tasks.

[0066] The core idea of ​​transfer learning is to reduce the time required for training from scratch by using model parameters obtained through training on existing data, and to make full use of knowledge in existing fields to improve the recognition ability of new tasks. Especially when the amount of data is limited, transfer learning can significantly improve the recognition effect of the model.

[0067] After being processed by CNN and RNN, the AI ​​target recognition module will output the classification result of the target. For example, it will determine whether a target is a foreign object (FOD) on the runway. This result will be passed to the subsequent real-time processing and feedback module. This module is responsible for feeding back the recognition results to the airport management system so that managers can make further operations based on the recognition results, such as clearing foreign objects on the runway.

[0068] The AI ​​target recognition module achieves accurate recognition of runway foreign objects (FOD) by combining convolutional neural networks (CNN) and recurrent neural networks (RNN). CNN extracts spatial features from the signal, and RNN processes time series features to ensure comprehensive recognition of the target. At the same time, through data enhancement and transfer learning technology, this module can adapt to different recognition environments and data situations, improving the accuracy and robustness of FOD recognition.

[0069] Please refer to the attached Figure 4 The real-time processing and feedback module of the present invention is one of the cores of the entire runway foreign object recognition system. It is responsible for real-time processing and feedback of the target recognition results obtained from the AI ​​target recognition module, ensuring that the airport management system can obtain relevant information about foreign objects on the runway in a timely manner and take corresponding measures. This module is closely connected with the aforementioned target recognition module and airport management system to ensure the real-time and high efficiency of the entire system.

[0070] The aforementioned AI target recognition module processes radar echo signals through deep learning algorithms to identify foreign objects on the runway. After target recognition, the recognition results need to be further processed by the real-time processing and feedback module to ensure the rapid transmission of recognition information and decision execution.

[0071] In this embodiment, the real-time processing and feedback module processes the target recognition results in real time through GPU acceleration technology, and transmits the results to the airport management system in real time through a high-speed communication interface. The core task of this module is to process a large amount of data in a very short time and ensure that the system can still operate efficiently in a complex environment.

[0072] Generally, the output of the target recognition module may include the type, location, size and corresponding time information of FOD, which needs to be quickly fed back to the management system so that further actions can be taken. The main function of the real-time processing and feedback module is to receive the output data from the target recognition module, format it and transmit it to the management system.

[0073] Specifically, the real-time processing module first receives the recognition results output by the AI ​​target recognition module, which are generally the classification information and location information of the target. For each target, the system needs to identify whether it is a foreign object, where it is located on the runway, and record the corresponding time point. All this information will be passed to the airport management system in a standard format for use by managers. Data transmission is achieved through high-speed data bus or wireless communication technology, ensuring real-time performance.

[0074] As an option, the real-time processing and feedback module can encrypt and compress the target recognition results. Encryption ensures the security of data during transmission and prevents information leakage or tampering; while data compression can reduce the data volume and improve data transmission efficiency, especially when the network bandwidth is limited, compression can significantly improve the transmission speed and stability.

[0075] In order to meet the system's requirements for real-time and high efficiency, the real-time processing and feedback module uses GPU acceleration technology. GPU (Graphics Processing Unit) can process large amounts of data in parallel, significantly increasing the speed of data processing. Especially in deep learning and image processing, the advantages of GPU are very obvious. This module uses GPU acceleration to quickly calculate and process target recognition results, ensuring that data processing and feedback are completed in a very short time.

[0076] Specifically, GPU acceleration technology is used in the following aspects in this module: Real-time processing and integration of target recognition results: For each detected foreign object, the system needs to determine in real time whether it is FOD and record key information such as the target type and location. GPU acceleration can complete the processing of large amounts of data in a relatively short period of time, avoiding the performance bottleneck caused by CPU single-thread processing.

[0077] Real-time data transmission: The processed results need to be transmitted to the management system through a high-speed interface. In practical applications, the system usually processes feedback information from multiple radar echo signals, and the parallel processing capability of GPU can ensure that all data is transmitted quickly.

[0078] In one possible implementation, GPU acceleration can use specially designed parallel processing algorithms, such as parallel computing of deep learning reasoning, to ensure that the system can maintain efficient operation under multi-tasking. The real-time performance of the system is crucial in actual scenarios, especially when detecting foreign objects on the runway. The lower the latency, the stronger the system's responsiveness, thus providing stronger protection for airport safety.

[0079] After the target identification results are processed by the real-time processing and feedback module, they will be quickly transmitted to the airport management system. The system responds in real time based on this information and instructs staff to clear foreign objects from the runway. The feedback information usually includes parameters such as the target location, type, size, and may include recommended disposal methods (such as cleaning paths, etc.). This information helps the management system to clear targets on the runway and ensure the safe takeoff and landing of flights.

[0080] In some embodiments, the feedback information may also include dynamic changes of the target. For example, when the target moves or deforms, the system can help the staff track the moving target through real-time updated feedback information to ensure timely cleaning and safety monitoring of foreign objects.

[0081] In this embodiment, the real-time processing and feedback module is not limited to feeding back the target recognition results to the airport management system. In order to ensure airport safety, the feedback information may also need to be transmitted to other related subsystems (such as flight control systems, safety monitoring systems, etc.). In some cases, managers may need to obtain multi-dimensional information from different systems to coordinate and handle complex situations. The efficient data processing and transmission capabilities of the real-time processing and feedback module ensure that these systems can work in coordination and respond quickly.

[0082] The real-time processing and feedback module ensures that the information obtained from the target recognition module can be quickly and accurately fed back to the management system through GPU acceleration and efficient data transmission mechanism. Through high-speed data bus, encryption and compression technology, and GPU acceleration, the system can provide timely and accurate feedback in complex environments. This module is not only a bridge for information transmission, but also undertakes the key task of ensuring the efficient operation of the entire system. Through this precise real-time feedback, foreign objects on the runway can be quickly cleared to ensure flight safety.

[0083] As part of this application, the present invention also provides a runway foreign object recognition method based on millimeter wave radar orthogonal sampling, comprising: S1. Receive the echo signal reflected from the runway; This step mainly involves receiving the echo signal reflected from the runway through the millimeter wave radar system. This step is the starting step of the entire FOD (foreign object) identification process and is directly related to the subsequent signal processing and target identification. The received echo signal contains information such as the amplitude, phase, distance and speed of the target, which is the basic data for the subsequent signal processing module and AI target recognition module.

[0084] Connected with the aforementioned signal acquisition module, this step realizes the data flow from the radar system to the signal processing module, providing the necessary input signals for subsequent signal processing (such as denoising, phase recovery, multipath effect removal, etc.) and target identification.

[0085] In this embodiment, the radar signal acquisition module receives the electromagnetic wave signals reflected from the runway through the receiving antenna and converts these signals into electrical signals. These signals mainly include two aspects of information: one is the amplitude information of the signal, and the other is the phase information. The amplitude information usually reflects the reflection intensity of the target, while the phase information is closely related to the relative distance, shape, material and other characteristics of the target. In order to ensure that subsequent signal processing can be performed efficiently, these echo signals need to be captured and converted by sophisticated hardware equipment.

[0086] In the present invention, the signal acquisition module operates in the millimeter wave frequency band, usually between 30GHz and 300GHz. The selection of this frequency band enables the radar system to have high resolution and detect smaller foreign objects on the runway. The radar system measures the distance, speed and other characteristics of the target by transmitting millimeter wave signals and receiving the echoes of these signals. When the radar wave encounters the target object, part of the signal will be reflected back and captured by the system through the receiving antenna.

[0087] Specifically, the received signal s(t) can be expressed in complex form: s(t)=A(t)e jθ(t) in: A(t) is the amplitude of the signal, indicating the strength of the echo; θ(t) is the phase of the echo signal, which indicates the relative distance and reflection characteristics of the target.

[0088] These signals are captured by the receiving antenna and transmitted to the signal processing module for subsequent processing.

[0089] In some embodiments, the signal acquisition module first captures the echo signal from the foreign object on the runway through the receiving antenna. Generally, the radar system measures the distance and reflection intensity of the target object by continuously transmitting pulse signals and receiving the echoes of these pulses. Specifically, when the radar transmits a signal, the signal will propagate to the foreign object on the runway and be reflected back to the radar system. At this time, the intensity and phase information of the echo signal will be directly related to the material, shape and relative distance of the target object to the radar.

[0090] As an option, the signal acquisition module can use a multi-channel antenna array to improve the system's detection range and resolution. The antenna array can more accurately control the direction and shape of the radar beam through the combination of multiple unit antennas, allowing the system to efficiently scan a specific area, thereby improving the detection capability of foreign objects.

[0091] In this embodiment, the output signal of the signal acquisition module is transmitted to the signal processing module. First, the echo signal will be processed by the analog front end, including signal amplification and frequency mixing, so as to convert the signal appropriately. This process usually involves a low noise amplifier (LNA) to ensure that weak echo signals can be effectively enhanced. After that, the signal is mixed with the local oscillator signal through a mixer to obtain a baseband signal, which is finally converted into a digital signal through an analog-to-digital converter (ADC).

[0092] Specifically, the signals after mixing and amplification are transmitted to the signal processing module in a digital form. The signal processing module further performs denoising, phase recovery, multipath effect removal and other processing on these digital signals to provide clear and accurate target information for the subsequent AI target recognition module.

[0093] The signal output by the signal acquisition module usually contains two types of information: amplitude information and phase information. In general, the amplitude information reflects the reflection intensity of the target, while the phase information is closely related to the relative position, motion state, and material characteristics of the target. In order to ensure accurate identification of the target, the extraction of phase information is particularly important.

[0094] Specifically, during the signal processing, the amplitude A(t) and phase θ(t) of the signal will be extracted and processed one by one. For example, after orthogonal sampling, the in-phase component I(t) and the orthogonal component Q(t) of the signal will be used to restore the phase information of the signal: I(t)=A(t)cos(θ(t)),Q(t)=A(t)sin(θ(t)) Through these components, we can further recover the phase information θ(t) of the target, thereby providing rich signal features for target recognition.

[0095] This step is the signal acquisition step, which is directly related to the accuracy and efficiency of subsequent signal processing. Through the signal acquisition module of the millimeter wave radar system, the echo signal from the runway can be captured in real time, the amplitude and phase information of the target can be extracted, and this information can be passed to the signal processing module. This process is the basis of the entire system, and its accuracy and real-time performance are crucial for the accurate identification of FOD. Through precise signal acquisition, effective signal transmission and processing, the present invention can provide clear data support for target identification, thereby realizing efficient and accurate runway foreign object identification.

[0096] S2. De-noising, phase recovery and multipath removal of the echo signal; In the runway foreign object (FOD) identification method of the present invention, this step is a key step, which involves further processing the echo signal obtained by the radar signal acquisition module in the previous step. In the previous step, the signal acquisition module has completed the initial capture of the signal and passed it to the signal processing module, and the collected signal at this time may still contain noise, multipath effect interference and other problems. If these problems are not handled, they will affect the subsequent target recognition accuracy. Therefore, the main task of this step is to remove these problems to ensure that the quality of the signal is high enough to provide accurate input data for target recognition.

[0097] Specifically, this step includes three main processing tasks: denoising, phase recovery, and multipath removal. Each task is optimized for different signal problems, thereby improving the accuracy of subsequent target recognition.

[0098] Generally, radar signals are interfered by environmental noise during transmission, which can affect the quality of the signal and reduce the accuracy of target recognition. In order to eliminate these noises, the denoising module uses the Wiener filtering algorithm in this embodiment. The Wiener filtering is a filtering method based on the minimum mean square error (MSE) criterion, which can optimize the signal according to the statistical characteristics of the signal and the noise.

[0099] In the Wiener filter algorithm, assume that we have an original signal x(t), which consists of a valid signal and noise. The Wiener filter processes the signal through the following formula: in, is the denoised signal, x(t) is the original signal, H(t) is the transfer function of the filter, and H(t) is defined as: Among them, S xx (t) is the power spectrum density of the signal, S vv (t) is the power spectral density of the noise. Through this formula, the Wiener filter can effectively suppress the noise while retaining the effective components in the signal.

[0100] Through Wiener filtering, the system can reduce the impact of environmental noise on the signal and ensure the clarity of the signal when it enters phase recovery and subsequent processing steps.

[0101] The signal after signal denoising still needs to be processed by phase recovery. Phase information is crucial in radar signals because it reflects the precise position, shape, and relative distance of the target to the radar. Without accurate phase information, the system will not be able to accurately distinguish foreign objects of different materials. Therefore, phase recovery is another key task in this step.

[0102] In this embodiment, the phase recovery module recovers the phase information of the signal by processing the in-phase component (I) and the quadrature component (Q). The echo signal s(t) can be expressed as a complex signal: s(t)=A(t)e jθ(t) Among them, A(t) is the amplitude of the signal, which indicates the reflection intensity of the target, and θ(t) is the phase of the signal, which reflects the relative position and reflection characteristics of the target.

[0103] To extract the phase information, the signal is first decomposed into the in-phase component I(t) and the quadrature component Q(t): I(t)=A(t)cos(θ(t)), Q(t)=A(t)sin(θ(t)) Then, the phase information is recovered using the following formula: Through this process, the system is able to recover precise phase information from the echo signal, which is crucial for subsequent target identification.

[0104] In practical applications, radar signals may be affected by multipath effects, especially in complex environments, such as buildings or other obstacles near the runway that may cause signal reflection or refraction, thereby affecting the quality of the echo signal. In order to eliminate the interference caused by the multipath effect, this embodiment uses a Kalman filter algorithm.

[0105] Kalman filtering is a recursive algorithm suitable for state estimation of dynamic systems. In the present invention, the Kalman filtering algorithm is used to correct the radar echo signal and eliminate the error caused by the multipath effect.

[0106] The basic state update formula of Kalman filtering is as follows: x k =Ax k-1 +Bu k +w k z k =Hx k +v k Among them, x k is the system state, indicating the real state of the signal; A is the state transition matrix, indicating the state transition from the previous moment to the current moment; B is the control input matrix, indicating the influence of external control on the system state; u k is the control input; w k is the process noise; z k is the observed value, which represents the collected signal; H is the observation matrix, which represents the relationship between the observations; v k is the observation noise.

[0107] Through Kalman filtering, the system can continuously predict and correct the signal deviation caused by the multipath effect, eliminate the interference caused by the multipath effect, and thus restore a more accurate signal.

[0108] The signal processed in this step will be passed to the subsequent AI target recognition module. Signal denoising and phase recovery ensure the accuracy of the signal, while Kalman filtering eliminates the interference of multipath effects on the signal. Through this series of processing, the signal quality has been significantly improved, which can provide accurate data support for the AI ​​target recognition module and help accurately identify foreign objects on the runway in the future.

[0109] The denoising, phase recovery and multipath effect removal processing of the signal in this step ensures the high quality of the signal, so that subsequent target identification can be carried out accurately. By removing noise through Wiener filtering, obtaining accurate target information using phase recovery technology, and eliminating multipath interference through Kalman filtering, the present invention can provide a powerful signal processing process. This process provides accurate and reliable signal input for identifying foreign objects on the runway, ensuring the overall efficiency and accuracy of the system.

[0110] S3. Perform target recognition based on the processed signal to identify foreign objects on the runway; In this step, the main task is to identify FOD (foreign object) targets through the signals processed by the previous steps (i.e., signal denoising, phase recovery, and multipath effect removal). As mentioned earlier, after preprocessing by the signal acquisition module and the signal processing module, clear and accurate signal information has been obtained, which is passed to the target identification module in this step. Step 3 relies on the high-quality signal data obtained from the previous steps and uses advanced artificial intelligence algorithms to classify and identify targets, thereby determining whether the target on the runway is a foreign object.

[0111] The target recognition module is the core part of the whole system. It not only relies on the static characteristics of the signal (such as the shape and size of the target), but also needs to analyze the time series characteristics in the signal (such as the movement trajectory of the target). This process is achieved with the help of deep learning algorithms, especially the combination of convolutional neural networks (CNN) and recurrent neural networks (RNN).

[0112] In this embodiment, the target recognition method is completed through the following steps: The input signal data is processed by the feature extraction network to extract the spatial features and time series features of the target; Based on the extracted features, the deep learning model inputs these features into the convolutional neural network (CNN) for image feature extraction; Subsequently, the extracted spatial features are fed into a recurrent neural network (RNN) for time series analysis to capture the dynamic characteristics of the target; Finally, the output of the deep learning model is used to classify the target and determine whether it is a foreign object.

[0113] Feature extraction and object classification In general, target recognition requires extracting a large number of features from the signal, including static features (such as the shape and size of the object) and dynamic features (such as the trajectory of motion). In order to achieve this goal, it is first necessary to extract features from the signal. In this embodiment, a convolutional neural network (CNN) is used to extract spatial features. CNN can effectively extract local features in the image through the convolution layer, such as edges, corners, textures, etc. These features are important information for identifying the shape, material and size of the target.

[0114] Specifically, the convolution operation of CNN is implemented by the following formula: O(t)=I(t)*W(t) Among them, O(t) represents the output after convolution processing, I(t) is the input radar signal image (that is, the data representation after signal acquisition and processing), W(t) is the convolution kernel, and * represents the convolution operation. The function of the convolution layer is to extract the spatial features in the signal through multiple convolution kernels. Each position in the feature map corresponds to the feature of a local area.

[0115] Specifically, CNN consists of multiple convolutional layers and pooling layers to form the feature extraction part. The convolution operation extracts different spatial features by applying different convolution kernels, and the pooling layer reduces the dimension of the feature map to focus on the most important feature information. In radar images, the features extracted by CNN include the edge, shape, texture, etc. of the target, which help to distinguish different types of FOD.

[0116] As an option, the CNN module can also be combined with a residual network or an attention mechanism to further enhance the model's feature extraction capabilities. By increasing the network depth and focusing on information in specific areas, the system can extract more refined features and enhance the robustness and accuracy of the model.

[0117] In some embodiments, radar signals not only reflect the static characteristics of the target, but also contain the dynamic characteristics of the target. In particular, when the target moves or changes, these dynamic information is particularly important. In order to process these dynamic characteristics, the system introduces a recurrent neural network (RNN) specifically for analyzing time series data.

[0118] Through its feedback structure, RNN can memorize information from the previous moment and combine it with the current input to better understand the target's motion state. For foreign objects on the runway, RNN can analyze the target's changing trends at different time points and then identify the target's motion characteristics.

[0119] The working process of RNN can be described by the following recursive formula: h t =σ(W hh h t-1 +W hx x t +b h ) y t =W hy h t +b y Among them, h t is the hidden state at time t, x t is the input signal at time t, W hh and W hx is the weight matrix, b h and b y is the bias term, σ is the activation function (such as tanh or ReLU), y t is the output of the model. Through this process, RNN can capture the time series characteristics of the signal, analyze the movement trend of the target, and provide support for target classification.

[0120] In one possible implementation, the long short-term memory network (LSTM) or the gated recurrent unit (GRU) is combined to enhance the memory capacity of the RNN. Especially for target dynamics over a long time span, LSTM and GRU can more effectively maintain important historical information and avoid the problem of gradient disappearance or explosion.

[0121] After extracting spatial features through convolutional neural networks (CNN) and analyzing time series features with recurrent neural networks (RNN), the deep learning model will generate classification results for the target. The results indicate whether there are foreign objects on the runway, and based on the classification of the target, the system can further determine the type of foreign objects (such as rocks, garbage, etc.).

[0122] The output of a deep learning model is usually a probability distribution, which indicates the probability that a certain target is a foreign object. For example, the system can output the probability value of each detected target belonging to FOD, and set a threshold for classification based on this value. If the probability exceeds a certain preset threshold, the target will be judged as a foreign object.

[0123] In some embodiments, the system can also combine the outputs of multiple deep learning models to further improve classification accuracy through model integration techniques (such as voting method, weighted average method, etc.). This integration method helps to reduce the overfitting problem that may occur in a single model and improve the robustness of the system.

[0124] After being processed by the target recognition module, the recognition results will be passed to the subsequent real-time processing and feedback module. This module is responsible for quickly transmitting the processing results to the airport management system to ensure that foreign objects on the runway are cleared in a timely manner.

[0125] In this step, the target recognition module effectively extracts static and dynamic features from the radar signal by combining convolutional neural networks (CNN) and recurrent neural networks (RNN). By analyzing the extracted features through a deep learning model, the system can accurately identify foreign objects on the runway. This process not only improves recognition accuracy, but also enables the system to handle complex target recognition tasks, such as the detection of dynamic targets. Finally, the recognition results are passed to the real-time processing and feedback module, ensuring the real-time and high efficiency of the system.

[0126] S4. Feedback the target recognition results to the management system for subsequent processing.

[0127] This step is the last important link of the runway foreign object (FOD) identification method of the present invention, which mainly includes feeding back the target identification results obtained in the previous steps to the airport management system. This link ensures that the system can respond quickly according to the identification results and take corresponding cleaning measures to ensure the safety of the airport runway and the normal take-off and landing of the aircraft. The core task of this step is to pass the identified foreign object information to the management personnel to help them clean up the foreign objects on the runway in time.

[0128] The steps of signal acquisition, processing, and target recognition have been discussed above, which generate key information about foreign objects on the runway. This step, as the last link, connects the system's recognition and decision execution. Through efficient data transmission and real-time feedback mechanisms, the management system can respond quickly and formulate processing plans based on the target recognition results. The signal processing module, AI target recognition module, and real-time processing and feedback module work closely together to ensure smooth and efficient data flow.

[0129] In this embodiment, the target recognition results are processed in real time through GPU acceleration technology, and the recognition results are transmitted to the airport management system through a high-speed communication interface. The management system analyzes and responds to the received FOD information, so as to arrange cleaning personnel or take corresponding measures in time.

[0130] In this embodiment, the real-time feedback mechanism is a key part to ensure that the target identification information can be transmitted in time and effectively processed. Generally, the detection of FOD needs to be completed in the shortest possible time to ensure flight safety. Therefore, this step realizes the efficient transmission of the identification results through the real-time processing and feedback module.

[0131] Specifically, the real-time processing and feedback module is responsible for receiving the recognition results from the target recognition module. These results usually include information such as the type, location, size, and movement trajectory of the FOD. The feedback process transmits this information to the airport management system through an efficient data bus or wireless communication. In order to improve the response speed and stability of the system, encryption and compression technology are used in data transmission to ensure the security and efficiency of the transmission process.

[0132] In one possible implementation, the data transmission process can be achieved through a cloud platform. By uploading information to the cloud, the management system can view the status of foreign objects on the runway in real time and send the information to specific operators. By storing and processing information on the cloud platform, the management system can achieve remote monitoring and cross-regional coordination, improving the level of intelligent airport management.

[0133] In order to ensure that the target recognition results can be efficiently and accurately transmitted to the management system, the real-time processing and feedback module formats the recognition results. In some embodiments, the recognition results are standardized before transmission to ensure that data can be effectively shared between different systems and modules.

[0134] Typically, target recognition results include the following key information: Target type: Identify the target as a foreign object and determine its specific type (such as stone, garbage, equipment, etc.); Target location: Identify the specific location of the target, including its coordinates on the runway; Target size: Estimate the size of the target based on the intensity of the echo signal and the morphological characteristics of the target; Time information: The specific time when the target is identified, used for tracing and coordination.

[0135] All this information will be transmitted to the management system in a preset data format. Standardization of data formats helps improve the efficiency of information processing and reduce errors caused by inconsistent data formats.

[0136] In order to ensure that data can be transmitted to the management system in real time and accurately, the real-time processing and feedback module in this embodiment adopts high-speed communication protocols. The design of these protocols ensures that there is no obvious delay in the signal transmission process, especially in a multi-tasking environment, ensuring timely feedback of information.

[0137] Generally, the system uses local area network (LAN) or wireless local area network (WLAN) protocol to transmit data to ensure the reliability and real-time performance of the transmission. In some implementations, the communication protocol can also use 5G communication technology to further improve the data transmission speed and system responsiveness.

[0138] In some embodiments, data packet encryption and identity authentication mechanisms can also be used to ensure the security of feedback information during transmission. Encryption processing can prevent sensitive information from being leaked or tampered with, ensuring that the system and management personnel can receive accurate target information.

[0139] Once the identification results are transmitted to the management system, the system will analyze the data and take appropriate actions. Based on the transmitted target identification results, the management system generates corresponding work tasks and assigns them to runway maintenance personnel. Specifically, the tasks will include: Determine the cleanup priorities for the targets; Arrange cleanup personnel and equipment; Update runway status to ensure that flights are not affected during the cleaning process.

[0140] As an option, the management system can automatically dispatch cleaning equipment and make dynamic adjustments based on information such as the type, location, and movement trajectory of the target. This automated processing method helps improve efficiency and reduce human intervention.

[0141] In some embodiments, the management system can also generate real-time reports and send the reports to airport managers through mobile terminals and other devices. Through this function, managers can grasp the situation of foreign objects on the runway in real time and take corresponding measures according to actual needs.

[0142] This step uses the real-time processing and feedback module to quickly and accurately transmit the results of the target recognition module to the airport management system, realizing the timely identification and cleaning of foreign objects on the runway. The application of technologies such as efficient data transmission, standardized data format, encryption technology, and high-speed communication protocol ensures rapid feedback of information and system response. Through the work of this module, foreign objects on the runway can be quickly handled, ensuring the safety and efficiency of airport operations.

[0143] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A runway foreign object recognition system based on millimeter-wave radar orthogonal sampling, characterized in that: include, A radar signal acquisition module is used to receive the echo signal reflected from the runway; A signal processing module, used for performing denoising, phase recovery and multipath effect removal processing on the echo signal; AI target recognition module, used to perform target recognition on the processed signals and identify foreign objects on the runway; The real-time processing and feedback module is used to feed back the target recognition results to the management system for subsequent processing.

2. The runway foreign object recognition system based on millimeter wave radar orthogonal sampling according to claim 1 is characterized in that: The signal processing module comprises: Denoising module, used to remove noise from the signal through the Wiener filtering algorithm; A phase recovery module, used to recover the phase information of the echo signal according to the ratio of the in-phase component to the orthogonal component; The multipath effect removal module is used to remove the multipath effect in the signal by using the Kalman filter algorithm.

3. The runway foreign object recognition system based on millimeter wave radar orthogonal sampling according to claim 1 is characterized in that: The phase recovery module recovers the phase of the echo signal by calculating the ratio between the in-phase component and the orthogonal component.

4. The runway foreign object recognition system based on millimeter wave radar orthogonal sampling according to claim 1 is characterized in that: The AI ​​target recognition module includes: Convolutional neural network module, used to extract spatial features from radar signals; The recurrent neural network module is used to analyze time series information and identify the target's motion trajectory.

5. The runway foreign object recognition system based on millimeter wave radar orthogonal sampling according to claim 1 is characterized in that: The AI ​​target recognition module also includes a data enhancement and transfer learning module, which is used to enhance the training data and use the deep learning model pre-trained in other fields to accelerate the training process of the FOD recognition model.

6. The runway foreign object recognition system based on millimeter wave radar orthogonal sampling according to claim 1 is characterized in that: The real-time processing and feedback module processes the target recognition results in real time through GPU acceleration technology, and feeds back the processing results to the airport management system.

7. A runway foreign object recognition method based on millimeter wave radar orthogonal sampling, characterized in that: A runway foreign object recognition system based on millimeter wave radar orthogonal sampling according to any one of claims 1 to 6, comprising: S1. Receive the echo signal reflected from the runway; S2. De-noising, phase recovery and multipath removal of the echo signal; S3. Perform target recognition based on the processed signal to identify foreign objects on the runway; S4. Feedback the target recognition results to the management system for subsequent processing.

8. The runway foreign object recognition method based on millimeter wave radar orthogonal sampling according to claim 7 is characterized in that: The denoising step removes noise in the echo signal by using a Wiener filtering algorithm.

9. The runway foreign object recognition method based on millimeter wave radar orthogonal sampling according to claim 7 is characterized in that: The phase recovery step includes recovering the phase information of the echo signal by calculating the ratio of the in-phase component to the orthogonal component.

10. The runway foreign object recognition method based on millimeter wave radar orthogonal sampling according to claim 7, characterized in that: The multipath effect removal step removes the multipath effect in the echo signal by using a Kalman filter algorithm.

Images