Airport FOD target detection method and system based on three-dimensional radar

Abstract

The invention provides an airport FOD target detection method and system based on a three-dimensional radar. The method comprises the steps that S1, an airport high-precision radar carries out preprocessing through radar original echoes; s2, performing an airport FOD detection analysis and target screening algorithm based on kernel density estimation; and S3, carrying out system global optimization and parameter adjustment based on performance inspection. According to the method, a rapid stabilization algorithm of the airport FOD target is realized, and rapid, stable and accurate target detection of the FOD is successfully realized by means of a kernel density FOD detection and target screening algorithm and by using field actual data. In continuous engineering application practices, continuous technology updating and method iteration form a set of complete method, and the engineering application shows that the scheme is effective and practical; the method solves a pain point in the airport aspect, realizes rapid, stable and reliable FOD target detection, and has very important application theory and engineering significance for eliminating the potential safety hazard of the FOD target in the airport.

Description

Technical Field

[0001] The present invention relates to the technical field of radar data processing, and particularly to a detection method and system for airport FOD targets based on three-dimensional radar. Background Art

[0002] FOD is the abbreviation of foreign object debris, generally referring to some foreign objects that may damage aircraft or systems, often called runway foreign objects. The monitoring of runway FOD refers to comprehensively inspecting the runway surface through manual or automated means, and specifically removing FOD to prevent harm to aircraft. Timely discovery and detection are the top priorities for airport safety.

[0003] Sustainable safety is a permanent theme in the development of civil aviation, and issues threatening the safe operation of aircraft in the flight area, especially FOD, have attracted increasing attention. It is reported that various unsafe events caused by runway FOD reach as high as 4 times per 10,000 takeoffs and landings, and the direct loss of each airport is about 20 million US dollars per year on average. Runway FOD often consists of some "small" volume objects such as pavement fragments, plastic products, animals and plants, metal devices, etc. However, due to the high-speed taxiing of aircraft on the runway, such objects may cause big troubles, such as scratching the tires of aircraft, being inhaled into the engine or breaking the fuel tank, etc. Currently, the detection of runway FOD at the vast majority of airports is still carried out manually. The field service team drives to conduct 4 inspections of the runway surface every day, with obvious limitations such as low efficiency, easy omission, and high cost.

[0004] Currently, the monitoring technologies for runway FOD are mainly based on two technologies: one is based on radar technology, and the other is based on video image recognition technology. Among them, the radar-based monitoring technology has strong environmental adaptability, and the video image recognition technology is more sensitive to object color and light contrast.

[0005] According to the installation method, FOD detectors are mainly divided into tower-mounted and edge-light-mounted types. Among them, the tower-mounted type is far from the runway center line and needs to meet the side clearance requirements of the flight area, with few installation points; the edge-light-mounted type is installed near the runway edge lights or taxiway edge lights, with less detection interference but greater later maintenance workload.

[0006] All edge-light-mounted detectors combine radar technology and image recognition technology. Tower-mounted detectors mainly use image recognition technology, and radar is an optional installation. Considering the respective advantages and disadvantages of edge-light-mounted detectors and tower-mounted detectors, as well as the general layout and long-term plan of Xiamen New Airport, taking into account the airport clearance conditions and studying the technical parameters of existing manufacturers' equipment, the final design plan is that both the North Runway 1 and the South Runway 1 adopt the edge-light-mounted monitoring method, and the edge-light-mounted monitoring method is radar + optical monitoring.

[0007] There are mainly the following problems:

[0008] 1) The problem of algorithm generality. Currently, researchers at home and abroad have developed various runway foreign object detection algorithms. However, most of these algorithms are used to construct background models in fixed areas. When the background area changes, the algorithms often cannot achieve good results. At the same time, due to the influence of factors such as the environment and shooting time, the detection accuracy will also be affected.

[0009] 2) The problem of objective evaluation of algorithm effectiveness. Currently, there is no unified test sample for runway foreign objects, making it difficult to conduct a unified objective evaluation of all the proposed detection algorithms. For the special application of airport runway foreign object detection, the detection algorithms can be evaluated from indicators such as precision, recall, false alarm rate, and missed detection rate. At the same time, a relatively comprehensive runway foreign object sample library needs to be established.

[0010] 3) The problem of systematicness in algorithm research. Current research only focuses on the detection of runway foreign objects, while less research has been done on the recognition of runway foreign objects and their corresponding risk level classification. Since there are different handling methods for runway foreign objects of different risk levels, that is, for high-risk FOD, it needs to be removed immediately; for medium-risk FOD, it can be removed after the current runway inspection; for low-risk FOD, it can be removed after the runway is closed. If the same handling method of immediate removal is applied to all risk levels of FOD, it will greatly increase the workload of runway inspection personnel, increase the stay time on the runway, and reduce the runway utilization rate.

[0011] 4) The problem of limitations in detection range. Most of the currently studied runway foreign object detection methods use fixed cameras, which requires building relatively high towers beside the runway to install cameras and transform the airport. At the same time, the relatively high towers pose safety hazards to the takeoff and landing of aircraft. Moreover, the fixed-installed cameras can only detect foreign objects in fixed areas, with poor flexibility and unable to detect foreign objects in areas such as taxiways and apron according to actual needs.

[0012] 5) The problem of algorithm efficiency. Since the algorithm needs to process pavement images in various different situations and needs to have good robustness for different scenarios, the algorithm complexity is relatively high, the processing time is long, the detection efficiency is low, and it cannot meet the real-time requirements.

[0013] Although radar can have the strongest reflection on metallic objects and can achieve effective detection, it is still relatively difficult to detect small objects such as small screws and nuts due to their small size. This is because the reflected echo is very small. In addition, for small non-metallic objects such as stones, leaves, rubber, and plastic products, their radar reflections are even smaller. To detect various small foreign objects on the runway, echo accumulation is required. It is required that the radar signals be accumulated to enhance the signal-to-clutter ratio. When the signal exceeds the threshold, it can be preliminarily considered as a target, but further confirmation is still needed. Due to the busy airport runway and tight time, the number of accumulations cannot be too many. Some echo accumulations still cannot reach the threshold, and the resulting missed detection consequences may be very serious safety hazards. There is a need to develop new algorithms to reduce missed detections and false alarms. It is very difficult to suppress various clutters caused by weather conditions. This method mainly resolves its problems, and the new method plays an extremely important role.

[0014] Patent application document CN113009507A discloses a distributed airport runway FOD monitoring system based on lidar, including an integrated detection unit, which includes a lidar module, an optical detection module, and an identification and control module, and is arranged in an array on both sides of the airport runway for real-time detection and identification of FOD on the airport runway; a data processing center for processing the lidar and optical alarm information sent from the integrated detection unit, performing information fusion, and obtaining a judgment on whether it is FOD; a display control subsystem for displaying and controlling the monitoring information and the alarm information of the target, and completing manual FOD confirmation through human-computer interaction; a communication network subsystem and a power management subsystem for realizing functions such as information transmission and interaction, power supply, etc.; having advantages such as high detection accuracy, low false alarm rate, no electromagnetic radiation, and providing uninterrupted monitoring. However, this patent cannot completely solve the existing technical problems and also cannot meet the requirements of the present invention. Summary of the Invention

[0015] Aiming at the defects in the prior art, the purpose of the present invention is to provide a detection method and system for airport FOD targets based on three-dimensional radar.

[0016] According to the detection method for airport FOD targets based on three-dimensional radar provided by the present invention, it includes:

[0017] Step S1: Preprocess the airport radar data to obtain the track parameters of each target;

[0018] Step S2: Use the kernel density estimation method to detect, analyze, and screen FOD targets;

[0019] Step S3: The radar guides the optoelectronic device to confirm the authenticity of the FOD target and the safety of the true FOD target;

[0020] Step S4: Perform system global optimization and adaptive parameter adjustment for adaptive performance verification.

[0021] Preferably, step S1 includes:

[0022] Unify the data formats of multiple radar stations and perform spatio-temporal alignment, including mapping the data of each radar station to a unified runway coordinate system through coordinate transformation;

[0023] Based on the echo intensity threshold and spatial correlation analysis, mask the target echoes in the area outside the runway;

[0024] Convert the radar echo intensity value through the min-max normalization formula to the range of [0, 1], where x is the original data value, and min and max are the minimum and maximum values of the current frame data respectively;

[0025] Accumulate the current frame and the 8 frames generated by expanding in 8 adjacent directions to form a three-dimensional array frames_mtd of 27-frame historical data for subsequent kernel density estimation.

[0026] Preferably, step S2 includes:

[0027] Use the Epanechnikov kernel function for probability density estimation, and the kernel function is defined as: where u is the normalized distance between the sample point and the point to be estimated, which is used to measure the contribution of the sample point to the density estimation;

[0028] Calculate the bandwidth parameter h using the Silverman rule, and the formula is: where n is the number of historical frames, d = 1 is the one-dimensional data dimension, and σ is the sample standard deviation;

[0029] For each pixel position (i, j), extract the sample data at the corresponding position of the historical frame and calculate the probability density estimation value. The expression is: L is the number of historical frames, sample_date[l] is the sample value of the l-th frame; frame[i, j] is a data point in a two-dimensional array of a certain frame in the radar scan data;

[0030] If and the normalized value of the current frame is greater than the sample mean μ, it is determined that there is a FOD target at this position, and the result matrix is updated, where λ kde is the detection threshold;

[0031] For non-uniformly distributed sample data, adopt local adaptive bandwidth adjustment, and the formula is:

[0032]

[0033] where, μ density is the global average density, and α is the smoothing factor.

[0034] Preferably, the step S3 includes:

[0035] Calculating the pointing parameters of the optoelectronic device according to the FOD target coordinates detected by the radar, including the azimuth angle θ and the elevation angle The formula is: where, (x target , y target , z target ) are the three-dimensional coordinates of the FOD target;

[0036] The optoelectronic system collects target images, combines with a predefined threat level library, and judges the target type and danger level through image recognition;

[0037] Generate a comprehensive judgment report, and recommend the clearance priority and countermeasures.

[0038] Preferably, the step S4 includes:

[0039] Statistical environmental clutter characteristic parameters, including the background noise mean μ noise and the standard deviation σ noise , dynamically adjust the detection threshold λ kde , the expression is: λ kde = μ noise + k·σ noise , where k is an adaptive coefficient, which is corrected in real time according to environmental factors;

[0040] After processing each frame, randomly replace one frame in frames_mtd to ensure dynamic data update.

[0041] According to the detection system for FOD targets at airports based on three-dimensional radar provided by the present invention, it includes:

[0042] Module M1: Preprocess the airport radar data to obtain the track parameters of each target;

[0043] Module M2: Use the kernel density estimation method to detect, analyze and screen FOD targets;

[0044] Module M3: The radar guides the optoelectronics to confirm the authenticity of the FOD target and the safety of the true FOD target;

[0045] Module M4: Perform system global optimization and adaptive parameter adjustment for adaptive performance verification.

[0046] Preferably, the module M1 includes:

[0047] Unify the data formats of multiple radar stations, and perform spatio-temporal alignment, including mapping the data of each radar station to a unified runway coordinate system through coordinate transformation;

[0048] Based on the echo intensity threshold and spatial correlation analysis, the target echo outside the runway area is masked;

[0049] The radar echo intensity value is transformed to the range of [0, 1] through the min-max normalization formula where x is the original data value, and min and max are the minimum and maximum values of the current frame data respectively;

[0050] Accumulate the current frame and the 8 frames generated by expanding in 8 adjacent directions to form a three-dimensional array frames_mtd of 27 frames of historical data for subsequent kernel density estimation.

[0051] Preferably, the module M2 includes:

[0052] Use the Epanechnikov kernel function for probability density estimation, and the kernel function is defined as: where u is the normalized distance between the sample point and the point to be estimated, which is used to measure the contribution of the sample point to the density estimation;

[0053] Adopt the Silverman rule to calculate the bandwidth parameter h, and the formula is: where n is the number of historical frames, d = 1 is the one-dimensional data dimension, and σ is the sample standard deviation;

[0054] For each pixel position (i, j), extract the sample data at the corresponding position of the historical frame and calculate the probability density estimation value. The expression is: L is the number of historical frames, sample_date[l] is the sample value of the l-th frame; frame[i, j] is a data point in the two-dimensional array of a certain frame in the radar scan data;

[0055] If and the normalized value of the current frame is greater than the sample mean μ, it is determined that there is a FOD target at this position, and the result matrix is updated, where λ kde is the detection threshold;

[0056] For non-uniformly distributed sample data, adopt local adaptive bandwidth adjustment, and the formula is:

[0057]

[0058] where, μ density is the global average density, and α is the smoothing factor.

[0059] Preferably, the module M3 includes:

[0060] According to the FOD target coordinates detected by the radar, calculate the pointing parameters of the optoelectronic device, including the azimuth angle θ and the elevation angle The formula is: where, (xtarget , y target , z target ) are the three-dimensional coordinates of the FOD target;

[0061] The optoelectronic system acquires the target image, combines the predefined threat level library, and determines the target type and danger level through image recognition;

[0062] Generate a comprehensive judgment report, and recommend the clearance priority and countermeasures.

[0063] Preferably, the module M4 includes:

[0064] Statistical environmental clutter characteristic parameters, including the background noise mean μ noise and the standard deviation σ noise , dynamically adjust the detection threshold λ kde , and the expression is: λ kde = μ noise + k · σ noise , where k is an adaptive coefficient and is corrected in real time according to environmental factors;

[0065] After processing each frame, randomly replace one frame in frames_mtd to ensure dynamic data update.

[0066] Compared with the prior art, the present invention has the following beneficial effects:

[0067] 1. The present invention combines the radar data to improve the ES algorithm model and the new method of kernel density estimation, realizes the fast, stable and reliable detection of the airport FOD target, and can solve the target distance of optoelectronics. Its processing efficiency is one order of magnitude smaller than other methods, with fast speed and high efficiency, and the system has good economy and practicability;

[0068] 2. The present invention adopts the new radar data processing algorithm and guided image observation according to the historical and on-site data of the three-dimensional radar FOD target, realizes the high-precision radar and optoelectronic system performance, functions such as the identification and safety analysis of the FOD target, and greatly reduces the system's leakage and false alarm rates; the system can be further expanded, such as the radar guiding multiple optoelectronics, and the optoelectronic equipment can search for the runway FOD target by itself, etc.;

[0069] 3. After preliminary verification, the present invention has been improved many times, and the implemented module is applied in the multi-source joint perception system, achieving good results in the fast, stable and reliable detection of the airport FOD target. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] By reading the detailed description of the non-limiting embodiments with reference to the following drawings, other features, objects and advantages of the present invention will become more apparent:

[0071] Figure 1 System structure diagram;

[0072] Figure 2 Block diagram of the implementation of the FOD algorithm based on kernel density estimation

[0073] Figure 3 Block diagram of the algorithm steps

[0074] Figure 4 Polar coordinate diagram of the first frame of the long-term test data of the left edge

[0075] Figure 5 Polar coordinate diagram of the second frame of the long-term test data of the left edge

[0076] Figure 6 Polar coordinate diagram of the third frame of the long-term test data of the left edge

[0077] Figure 7 Polar coordinate diagram of the fourth frame of the long-term test data of the left edge

[0078] Figure 8 Polar coordinate diagram of the fifth frame of the long-term test data of the left edge Specific implementation manners

[0079] The present invention will be described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any form. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several changes and improvements can still be made. These all belong to the protection scope of the present invention

[0080] Embodiment 1

[0081] The present invention provides a detection method for FOD targets based on 3D radar at airports, as Figure 1 and Figure 2 shown, steps of the batch Epanechnikov kernel density estimation (Batch KDE) algorithm

[0082] 1) Initial frame accumulation

[0083] Before processing each frame of data, first accumulate 3 frames of data

[0084] For each frame, generate 8 neighbor points in front, behind, left, and right of each point in this frame to form 8 new frames

[0085] Therefore, 3 + 8 * 3 = 27 frames of data will be generated in the initial stage. This is used as frames_mtd for the subsequent kde calculation

[0086] 2) Input data definition

[0087] frame: The data of the current frame, represented as a two-dimensional array with a shape of N*M, where N = 1000 is the number of azimuths and M = 4000 is the number of range cells.

[0088] frames_mtd: A three-dimensional array containing multi-frame data, with a shape of L*N*M, where L is the number of historical frames.

[0089] λ_kde: A threshold parameter used to compare the results of probability density estimation.

[0090] result: The returned frame used to save the results of rate density estimation.

[0091] 3) Define the kernel function

[0092] The Epanechnikov kernel function K(u) is defined as:

[0093]

[0094] This kernel function is smooth in the interval |u| ≤ 1 and drops to zero at the boundaries; u represents the independent variable of the kernel function, which is the length between the sample point and the point to be estimated.

[0095] 4) Data normalization

[0096] Normalize the data in frame and frames_mtd to the range [0, 1]. This can be achieved through Min-Max Normalization:

[0097]

[0098] where x is the original data value, and min and max are the minimum and maximum values in the data respectively.

[0099] Actual method:

[0100] Normalize the data in frame and frames_mtd to the range [0, 1]. This can be achieved by dividing each pixel value by 255.

[0101] 5) Calculate the bandwidth

[0102] Use the Silverman rule to estimate the bandwidth h. For one-dimensional data, the bandwidth h is calculated as follows:

[0103]

[0104] where n is the number of samples, d = 1 indicates that the data is one-dimensional, and σ is the standard deviation of the samples.

[0105] 6) Kernel density estimation

[0106] For each pixel position (i, j) in the frame, extract the sample data sample_data at the corresponding positions in all historical frames.

[0107] Calculate the mean μ and standard deviation σ of the sample data.

[0108] Use the Epanechnikov kernel function to calculate the probability density estimate at this position

[0109]

[0110] 7) Result update

[0111] If the probability density estimate is less than the threshold λ_kde, and the normalized value of the current frame is greater than the sample mean, then update the value at the corresponding position in the result matrix result to the original value of the current frame:

[0112]

[0113] 8) Output result

[0114] Finally, result is an array of shape n×m, which contains the processed image data.

[0115] 9) Random frame update

[0116] After processing each frame, select an index r (ranging from 0 to 26), and replace the r-th frame in frames_mtd with the newly processed frame.

[0117] For each frame in frames_mtd, traverse each pixel position, generate a random number p, ranging from 0 to 1. If p is less than a predefined probability threshold, then replace the value at this position in the current frame with the value at the corresponding position in the newly processed frame. This can keep the data in frames_mtd dynamically updated and ensure that the latest data is incorporated into subsequent KDE calculations

[0118] Test condition: After the runway is closed to ensure that there are no interfering targets in the radar detection area of the runway, the "target placement personnel" go onto the runway and horizontally place targets (1 standard target, 10 physical samples, 1 standard sample) at the radar detection edge area (far side of the left and right edge runways). Two standard parts are used to locate the target placement area. After placement, notify the test record personnel to start the test. See Figure 3 as shown.

[0119] The order of placing the samples is as follows: standard parts, asphalt blocks, long metal bars, high metal bars, rubber sheets, fuel tank caps, wrenches, plastic pipes, hydraulic pipes, nuts, sockets, and standard parts.

[0120] Each file directory contains the radar scan record data for a single test period (from the start of the test to the withdrawal from the site). Different numbers represent different batches of tests, and there are differences in the detection environment; left and right represent the positions of the samples relative to the radar. The data is the raw radar data without processing, and the signal values are relatively small.

[0121] A display of the detection effect of the samples in each data is presented. In the picture, yellow is the background and black is the detected target. Each data has multiple frames. A complete display of the latest left-edge data is made, and for other data, the frame with the largest number of detected samples among multiple frames is selected.

[0122] It should be noted that the detection method adopted is to try to detect the foreign object within the first few frames when it first appears, but it will not continuously detect the targets that have already appeared. For practical applications, considering storing the positions of all detected targets in the area of concern (the runway belongs to the area of concern, while the grassland does not. According to the current detection algorithm, there will be detected targets in the grassland, but since it does not belong to the runway, it does not cause interference), and removing the record of the target after optical and electronic confirmation and removal of the target. With the accumulation of data ( Figures 5 to 8 ) as the number of times increases, it can be seen that the detection effect is gradually improved, and the application level can be reached after 3 times.

[0123] Implementation process and improvement methods:

[0124] The present invention first preprocesses the three-dimensional radar echo data, proposes to modify the ES algorithm and the batch-based kernel density estimation method, uses the radar to guide the optoelectronic or optoelectronic autonomous search for runway FOD targets, comprehensively determines the safety of FOD targets using radar data and images, gives a comprehensive determination basis, and recommends taking necessary measures. During the R & D process, the improvement methods are as follows: 1) Focus on improving the calculation method of kernel density estimation; 2) The method of comprehensively determining safety; 3) The radar data-guided optoelectronic image authentication FOD module; 4) Continuously improve the statistical and correction processes to achieve fast, stable, and accurate guidance and the detection effect of FODMB. Many small problems were also found in the actual test, and corresponding modifications and improvements were made to the software modules. From the actual test results, this method is effective.

[0125] The utility of the detection method for three-dimensional radar airport FOD targets of the present invention realizes the effective detection of fast, stable and reliable sea targets. Applying this system will play an important role in the takeoff and landing safety of airport aircraft. This patent gives the implementation principle and improvement method, provides a detailed description of the system implementation process and discussion of problems. After preliminary verification, the method is improved many times. The implemented module is applied in the multi-source joint perception system, achieving fast, stable and reliable tracking effects, and greatly improving the detection and evidence collection capabilities of various illegal targets.

[0126] Embodiment 2

[0127] The present invention also provides a detection system based on three-dimensional radar airport FOD targets, including:

[0128] Module M1: Radar data detection module: The three-dimensional radar target echo data is applied to the ES detection module, and the module needs to be improved. Since the original ES module is for two-dimensional radar, an elevation channel needs to be added, and the depth is combined with the kernel density estimation method to achieve rapid detection of airport FOD targets.

[0129] Module M2: Kernel density estimation module: Analyze and screen important targets according to the characteristics of the concerned airport FOD targets, input radar data, refer to the echo database of historical targets, the time and environmental and target characteristics such as past aircraft, airport vehicles, personnel, and weather, and adjust the kernel density estimation parameters through the parameter settings of the foreground and background, and output feedback to optimize the system.

[0130] Module M3: Radar-guided optoelectronic and comprehensive determination module: Calculate the radar target coordinates and the optoelectronic center pointing parameters, adjust the optoelectronic aiming at the specified target, determine the nature and area of the target according to the image, determine the safety level of the FOD target, and formulate measures to be taken.

[0131] Module M4: System overall optimization module: According to the system output or user information feedback, adjust the system parameters and automatically fine-tune the threshold, adaptively process the clutter change caused by the rapid change of the environment, and stabilize the system performance index.

[0132] The said Module M1 includes: unifying the data formats of multiple radar stations, performing spatio-temporal alignment, including mapping the data of each radar station to a unified runway coordinate system through coordinate transformation; based on the echo intensity threshold and spatial correlation analysis, masking the target echoes in the area outside the runway; converting the radar echo intensity value to the range of [0,1] through the minimum-maximum normalization formula where x is the original data value, and min and max are respectively the minimum and maximum values of the current frame data; accumulating the current frame and the 8 frames generated by the 8-direction extension of its adjacent frames to form a three-dimensional array frames_mtd of 27-frame historical data for subsequent kernel density estimation.

[0133] The module M2 includes: performing probability density estimation using the Epanechnikov kernel function, and the kernel function is defined as: where u is the normalized distance between the sample point and the point to be estimated, which is used to measure the contribution of the sample point to the density estimation; the bandwidth parameter h is calculated using the Silverman rule, and the formula is: where n is the number of historical frames, d = 1 is the one-dimensional data dimension, and σ is the sample standard deviation; for each pixel position (i, j), the sample data at the corresponding position in the historical frames is extracted, and the probability density estimation value is calculated. The expression is: L is the number of historical frames, sample_date[l] is the sample value of the l-th frame; frame[i, j] is a data point in a two-dimensional array of a certain frame in the radar scan data; if and the normalized value of the current frame is greater than the sample mean μ, it is determined that there is a FOD target at this position, and the result matrix is updated, where λ kde is the detection threshold; for non-uniformly distributed sample data, local adaptive bandwidth adjustment is adopted, and the formula is: where, μ density is the global average density, and α is the smoothing factor.

[0134] The module M3 includes: calculating the pointing parameters of the optoelectronic device according to the FOD target coordinates detected by the radar, including the azimuth angle θ and the elevation angle The formula is: where, (x target , y target , z target ) are the three-dimensional coordinates of the FOD target; the optoelectronic system collects the target image, combines the predefined threat level library, and judges the target type and danger level through image recognition; generates a comprehensive judgment report, and recommends the clearance priority and countermeasures.

[0135] The module M4 includes: statistically analyzing the environmental clutter characteristic parameters, including the background noise mean μ noise and the standard deviation σ noise , dynamically adjusting the detection threshold λ kde , and the expression is: λ kde = μ noise + k·σ noise , where k is the adaptive coefficient and is corrected in real time according to environmental factors; after processing each frame, randomly replace one frame in frames_mtd to ensure dynamic data update.

[0136] Under rainfall conditions, k is adjusted according to the linear relationship k = 1.2 + 0.05R, where R is the rainfall intensity; under strong wind conditions, k is adjusted according to the linear relationship k = 1.5e 0.1VAdjustment, where V is the wind speed.

[0137] Those skilled in the art know that in addition to implementing the systems, devices, and their respective modules provided by the present invention in the form of pure computer-readable program code, the method steps can be logically programmed to enable the systems, devices, and their respective modules provided by the present invention to be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers, etc., to achieve the same program. Therefore, the systems, devices, and their respective modules provided by the present invention can be considered as a kind of hardware component, and the modules included therein for implementing various programs can also be regarded as the structures within the hardware component; the modules for implementing various functions can also be regarded as either software programs for implementing the method or the structures within the hardware component.

[0138] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. Without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A method for detecting airport FOD targets based on three-dimensional radar, characterized in that: include: Step S1: pre-processing the airport radar data to obtain the track parameters of each target; Step S2: using the kernel density estimation method to detect, analyze and screen FOD targets; Step S3: The radar guides the photoelectric confirmation of the authenticity of the FOD target and the safety of the real FOD target; Step S4: Perform system global optimization and adaptive parameter adjustment for adaptive performance testing.

2. The method for detecting airport FOD targets based on three-dimensional radar according to claim 1, characterized in that: The step S1 comprises: Unify the data formats of multiple radar stations and perform time and space alignment, including mapping the data of each radar station to a unified runway coordinate system through coordinate transformation; Based on echo intensity threshold and spatial correlation analysis, target echoes outside the runway area are shielded; The radar echo intensity value is normalized by the minimum-maximum formula Convert to the range of [0,1], where x is the original data value, min and max are the minimum and maximum values ​​of the current frame data respectively; Accumulate the current frame and the 8 frames of data generated by the expansion of the adjacent 8 directions to form a three-dimensional array frames_mtd of 27 frames of historical data for subsequent kernel density estimation.

3. The method for detecting airport FOD targets based on three-dimensional radar according to claim 1, characterized in that: The step S2 comprises: The Epanechnikov kernel function is used for probability density estimation. The kernel function is defined as: Where u is the normalized distance between the sample point and the point to be estimated, which is used to measure the contribution of the sample point to the density estimation; The bandwidth parameter h is calculated using the Silverman rule, and the formula is: Where n is the number of historical frames, d = 1 is the one-dimensional data dimension, and σ is the sample standard deviation; For each pixel position (i, j), extract the sample data of the corresponding position in the historical frame and calculate the probability density estimate, which is expressed as: L is the number of historical frames, sample_date[l] is the sample value of the lth frame; frame[i,j] is a data point in the two-dimensional array of a frame in the radar scan data; like If the normalized value of the current frame is greater than the sample mean μ, it is determined that there is a FOD target at this location and the result matrix is ​​updated, where λ kde is the detection threshold; For non-uniformly distributed sample data, local adaptive bandwidth adjustment is adopted, and the formula is: Among them, μ density is the global average density, and α is the smoothing factor.

4. The method for detecting airport FOD targets based on three-dimensional radar according to claim 1, characterized in that: The step S3 comprises: According to the FOD target coordinates detected by the radar, the pointing parameters of the optoelectronic device are calculated, including the azimuth angle θ and the pitch angle The formula is: Among them, (x target ,y target ,z target ) is the three-dimensional coordinate of the FOD target; The optoelectronic system collects target images and determines the target type and danger level through image recognition in combination with the predefined threat level library; Generate a comprehensive judgment report and recommend removal priorities and countermeasures.

5. The method for detecting airport FOD targets based on three-dimensional radar according to claim 1, characterized in that: The step S4 comprises: Statistical environmental clutter characteristic parameters, including background noise mean μ noise and standard deviation σ noise , dynamically adjust the detection threshold λ kde , the expression is: kde =μ noise +k·σ noise , where k is the adaptive coefficient, which is modified in real time according to environmental factors; After processing each frame, a frame in frames_mtd is randomly replaced to ensure that the data is dynamically updated.

6. A detection system for airport FOD targets based on three-dimensional radar, characterized in that: include: Module M1: pre-process the airport radar data to obtain the track parameters of each target; Module M2: Detect, analyze and screen FOD targets using kernel density estimation method; Module M3: Radar-guided optoelectronic confirmation of the authenticity of FOD targets and the safety of real FOD targets; Module M4: System global optimization and adaptive parameter adjustment for adaptive performance testing.

7. The airport FOD target detection system based on three-dimensional radar according to claim 6 is characterized in that: The module M1 comprises: Unify the data formats of multiple radar stations and perform time and space alignment, including mapping the data of each radar station to a unified runway coordinate system through coordinate transformation; Based on echo intensity threshold and spatial correlation analysis, target echoes outside the runway area are shielded; The radar echo intensity value is normalized by the minimum-maximum formula Convert to the range of [0,1], where x is the original data value, min and max are the minimum and maximum values ​​of the current frame data respectively; Accumulate the current frame and the 8 frames of data generated by the expansion of the adjacent 8 directions to form a three-dimensional array frames_mtd of 27 frames of historical data for subsequent kernel density estimation.

8. The airport FOD target detection system based on three-dimensional radar according to claim 6 is characterized in that: The module M2 comprises: The Epanechnikov kernel function is used for probability density estimation. The kernel function is defined as: Where u is the normalized distance between the sample point and the point to be estimated, which is used to measure the contribution of the sample point to the density estimation; The bandwidth parameter h is calculated using the Silverman rule, and the formula is: Where n is the number of historical frames, d = 1 is the one-dimensional data dimension, and σ is the sample standard deviation; For each pixel position (i, j), extract the sample data of the corresponding position in the historical frame and calculate the probability density estimate, which is expressed as: L is the number of historical frames, sample_date[l] is the sample value of the lth frame; frame[i,j] is a data point in the two-dimensional array of a frame in the radar scan data; like If the normalized value of the current frame is greater than the sample mean μ, it is determined that there is a FOD target at this location and the result matrix is ​​updated, where λ kde is the detection threshold; For non-uniformly distributed sample data, local adaptive bandwidth adjustment is adopted, and the formula is: Among them, μ density is the global average density, and α is the smoothing factor.

9. The airport FOD target detection system based on three-dimensional radar according to claim 6, characterized in that: The module M3 comprises: According to the FOD target coordinates detected by the radar, the pointing parameters of the optoelectronic device are calculated, including the azimuth angle θ and the pitch angle The formula is: Among them, (x target ,y target ,z target ) is the three-dimensional coordinate of the FOD target; The optoelectronic system collects target images and determines the target type and danger level through image recognition in combination with the predefined threat level library; Generate a comprehensive judgment report and recommend removal priorities and countermeasures.

10. The airport FOD target detection system based on three-dimensional radar according to claim 6, characterized in that: The module M4 comprises: Statistical environmental clutter characteristic parameters, including background noise mean μ noise and standard deviation σ noise , dynamically adjust the detection threshold λ kde , the expression is: kde =μ noise +k·σ noise , where k is the adaptive coefficient, which is modified in real time according to environmental factors; After processing each frame, a frame in frames_mtd is randomly replaced to ensure that the data is dynamically updated.

Images