A multi-data fusion method, system and medium in airport apron safety supervision

Abstract

The present application relates to the technical field of data processing, and discloses a multi-data fusion method, system and medium in airport apron safety supervision. The method includes: based on a target feature matrix, judging potential conflicts through set safety distance thresholds and speed thresholds, constructing risk assessment indicators by combining historical risk event data, and obtaining safety risk scores; according to the safety risk scores, establishing spatial association relationships and risk propagation relationships among apron targets, calculating regional risk density through the risk accumulation effect, and generating real-time safety situation data; for the real-time safety situation data, calculating the objective function value of safety control, setting safety constraint conditions for optimization and solution, and forming control strategy parameters. The present application realizes the effective fusion of multi-source heterogeneous data, constructs accurate apron safety situation perception, and generates optimized control strategies based on the situation information, thereby improving the accuracy and real-time performance of apron safety supervision.

Description

Technical Field

[0001] This application relates to the field of data processing, and particularly to a multi-data fusion method, system and medium in airport apron safety supervision. Background Art

[0002] The airport apron is an important area for various aircraft support operations, involving multiple operations such as aircraft taxiing, towing, aircraft maintenance inspection, refueling, and catering. The existing apron safety supervision mainly relies on a single data source for management. For example, video surveillance systems are used for target detection and tracking, radar systems are used for position measurement, and ADS-B systems are used for aircraft identification, etc. These supervision means operate independently, forming multiple information islands. At the same time, there are problems such as time delay, inconsistent spatial reference, and uneven data quality in the data collected by different monitoring devices, which bring many challenges to apron safety management.

[0003] The main problem existing in the prior art is the lack of an effective fusion mechanism for multi-source heterogeneous data. The limitations of a single data source lead to the existence of monitoring blind spots, and there is a lack of collaboration between different data sources, making it impossible to form a comprehensive understanding of the apron situation. In addition, the existing risk assessment methods often only consider static safety interval requirements, ignoring the propagation effect and cumulative effect of risks in space, and it is difficult to accurately reflect the dynamic risk distribution in the complex apron environment. In the formulation of control strategies, the existing methods mostly adopt preset rules, lacking consideration of multi-objective collaborative optimization, and it is difficult to balance operation efficiency while ensuring safety. Summary of the Invention

[0004] This application provides a multi-data fusion method, system and medium in airport apron safety supervision, which is used to achieve the effective fusion of multi-source heterogeneous data, construct an accurate apron safety situation perception, and generate optimized control strategies based on the situation information, so as to improve the accuracy and real-time performance of apron safety supervision.

[0005] In a first aspect, the present application provides a multi - data fusion method for airport apron safety supervision. The multi - data fusion method for airport apron safety supervision includes: for the video data, radar data, and position data collected by multi - source sensors on the apron, generating a data credibility value according to data integrity and time consistency, filtering and denoising the original data through the data credibility value to obtain a pre - processed data set; for the pre - processed data set, establishing a unified spatio - temporal reference through timestamp and coordinate transformation, calculating the matching degree between different data sources based on the similarity of target features to obtain the fused target data; according to the fused target data, extracting feature parameters from three dimensions of position, speed, and motion trajectory, screening and combining the feature parameters according to the correlation coefficient to form a target feature matrix; based on the target feature matrix, judging potential conflicts through set safety distance thresholds and speed thresholds, constructing a risk assessment index in combination with historical risk event data to obtain a safety risk score; according to the safety risk score, establishing a spatial association relationship and a risk propagation relationship between apron targets, calculating the regional risk density through the risk accumulation effect to generate real - time safety situation data; for the real - time safety situation data, calculating the objective function value of safety control, setting safety constraint conditions for optimization to form control strategy parameters.

[0006] In a second aspect, the present application provides a multi - data fusion system for airport apron safety supervision. The multi - data fusion system for airport apron safety supervision includes:

[0007] A generation module, configured to generate a data credibility value according to data integrity and time consistency for the video data, radar data, and position data collected by multi - source sensors on the apron, and filter and denoise the original data through the data credibility value to obtain a pre - processed data set;

[0008] A fusion module, configured to establish a unified spatio - temporal reference through timestamp and coordinate transformation for the pre - processed data set, calculate the matching degree between different data sources based on the similarity of target features to obtain the fused target data;

[0009] An extraction module, configured to extract feature parameters from three dimensions of position, speed, and motion trajectory according to the fused target data, screen and combine the feature parameters according to the correlation coefficient to form a target feature matrix;

[0010] A judgment module, configured to judge potential conflicts through set safety distance thresholds and speed thresholds based on the target feature matrix, construct a risk assessment index in combination with historical risk event data to obtain a safety risk score;

[0011] A establishment module, configured to establish a spatial association relationship and a risk propagation relationship between apron targets according to the safety risk score, calculate the regional risk density through the risk accumulation effect to generate real - time safety situation data;

[0012] A solution module, configured to calculate the objective function value of safety control for real-time safety situation data, set safety constraint conditions for optimization and solution, and form control strategy parameters.

[0013] The third aspect of the present application provides a computer-readable storage medium, in which instructions are stored. When the instructions run on a computer, the computer is enabled to execute the multi-data fusion method in the above airport apron safety supervision.

[0014] In the technical solution provided by the present application, the original data is filtered and denoised through the data credibility value, solving the problem of quality differences in multi-source heterogeneous data and improving data reliability; secondly, a unified spatio-temporal reference is established by using timestamps and coordinate transformation, and the matching degree between data sources is calculated based on the similarity of target features, realizing the precise alignment and association between different data sources; thirdly, feature parameters are extracted from three dimensions of position, speed, and motion trajectory and correlation analysis is performed to construct a multi-dimensional target feature description, enhancing the expression ability of the target state; potential conflicts are judged by setting safety distance thresholds and speed thresholds, and an evaluation index system is constructed in combination with historical risk event data, realizing the precise quantification of risks; at the same time, by establishing the spatial association relationship and risk propagation relationship between apron targets and introducing the concept of risk accumulation effect, the distribution and propagation characteristics of risks in space are accurately characterized; finally, the objective function value is calculated based on real-time safety situation data and constraint conditions are set, and an optimization algorithm is used to solve the control strategy parameters, realizing the collaborative optimization of multiple objectives on the premise of ensuring safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0016] Figure 1 It is a schematic diagram of an embodiment of the multi-data fusion method in airport apron safety supervision in an embodiment of the present application;

[0017] Figure 2 It is a radar multi-dimensional information waveform diagram provided by an embodiment of the present application;

[0018] Figure 3 It is a schematic diagram of an embodiment of the multi-data fusion system in airport apron safety supervision in an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] The embodiments of the present application provide a multi - data fusion method, system and medium in airport apron safety supervision. Terms such as "first", "second", "third", "fourth", etc. (if any) in the specification, claims and the above - mentioned drawings of the present application are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that such data used can be interchanged under appropriate circumstances so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "comprising" or "having" and any variation thereof are intended to cover non - exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these process, method, product or device.

[0020] For ease of understanding, the specific process of the embodiments of the present application will be described below. Please refer to Figure 1 , an embodiment of the multi - data fusion method in airport apron safety supervision in the embodiments of the present application includes:

[0021] Step S101: For the video data, radar data and position data collected by the apron multi - source sensors, generate a data credibility value according to data integrity and time consistency, filter and denoise the original data through the data credibility value to obtain a pre - processed data set;

[0022] Step S102: For the pre - processed data set, establish a unified spatio - temporal benchmark through timestamp and coordinate transformation, calculate the matching degree between different data sources based on the similarity of target features, and obtain the fused target data;

[0023] Step S103: According to the fused target data, extract feature parameters from three dimensions of position, speed and motion trajectory, screen and combine the feature parameters according to the correlation coefficient to form a target feature matrix;

[0024] Step S104: Based on the target feature matrix, judge potential conflicts through the set safety distance threshold and speed threshold, and construct a risk assessment index in combination with historical risk event data to obtain a safety risk score;

[0025] Step S105: According to the safety risk score, establish the spatial association relationship and risk propagation relationship between apron targets, calculate the regional risk density through the risk accumulation effect, and generate real - time safety situation data;

[0026] Step S106: For the real - time safety situation data, calculate the objective function value of safety control, set safety constraint conditions for optimization and solution, and form control strategy parameters.

[0027] It is understandable that the execution entity of this application can be a multi - data fusion system in airport apron safety supervision, or it can also be a terminal or a server. Specifically, it is not limited here. In the embodiments of this application, the server is taken as an execution entity for illustration.

[0028] Specifically, the video data, radar data, and position data are processed. The video data is an image sequence collected by video monitoring devices in the apron. The radar data includes the distance, azimuth, and speed information of the target, and the position data records the longitude, latitude, and timestamp of the target. The data integrity is obtained by calculating the data missing rate and sampling uniformity. The image sequence of the video data is analyzed in units of 50 milliseconds, and the frame integrity within each time window is statistically calculated. For the radar data, the data continuity is judged by detecting the time interval between sampling points, and the integrity of the position data is determined by the uniformity of the sampling time. The time consistency is based on the comparison of timestamps from different data sources, calculating the deviation degree at the sampling moment, and generating a time synchronization index. The data credibility value is obtained by weighting the integrity index and the time consistency index, and the weight coefficient is set according to the importance of different data sources. In the data pre - processing stage, data from different sources is unified to the same spatio - temporal reference. The time reference adopts the UTC time standard, and the spatial reference adopts the local coordinate system of the airport apron. The timestamp conversion involves time zone adjustment and alignment at the millisecond level of precision, and the coordinate conversion requires establishing a mapping relationship between different coordinate systems. The calculation of the target feature similarity includes three aspects: geometric features, motion features, and temporal features, and the Euclidean distance between feature vectors is used to measure whether different data sources describe the same target. The calculation of the matching degree uses a similarity matrix, and the matrix elements represent the intensity of the corresponding relationship between different data sources. For the fused target data, features are extracted from three dimensions: position, speed, and motion trajectory. The position feature includes the two - dimensional coordinates and height information of the target, the speed feature includes linear velocity and angular velocity, and the motion trajectory feature describes the historical motion path of the target. The correlation analysis of feature parameters uses the Pearson correlation coefficient to calculate the linear correlation degree between different features and screen out the feature combinations with significant correlations. Each row of the feature matrix represents a target, and each column corresponds to a feature dimension.

[0029] In the risk assessment stage, the judgment of potential conflicts is based on the target feature matrix. The safety distance threshold is determined according to the operation specifications of different types of targets, and the speed threshold is set based on the apron speed limit requirements. The historical risk event data contains records of past safety events such as collisions, incursions, and speeding. The risk assessment index system is constructed through statistical analysis. The safety risk score is normalized to unify risk indicators in different dimensions within the range of 0-1. In the process of generating real-time safety situation data, the spatial correlation relationship between targets is calculated, including distance, relative position, and movement trend. The risk propagation relationship describes the diffusion effect of risk between adjacent targets, and the risk propagation matrix is used to represent the transmission intensity of risk. The calculation of regional risk density takes into account the risk accumulation effect, and the risk values of all targets within the unit area are weighted and superimposed. In the control strategy generation stage, the real-time safety situation data is transformed into an objective function, and the function value reflects the overall risk level in the current situation. The safety constraint conditions include hard constraints such as the minimum safety distance, the maximum allowable speed, and the no-entry area. The optimization solution process uses an iterative calculation method to gradually adjust the control parameters until the constraint conditions are met and the objective function reaches the optimal value.

[0030] For example, when multiple aircraft and multiple ground vehicles are operating simultaneously on the apron, it is found through video monitoring that there is an intersection point in the movement trajectories of target A and target B. The acquired data is processed to extract the position and speed information of the targets, calculate the relative distance and speed between the two targets, and combine with the risk event records in similar scenarios in historical data to generate the current risk level. Then, analyze the spatial propagation law of the risk. Considering that target C is also active in this area, the risk states of the three targets are analyzed in association to generate targeted control suggestions, such as adjusting the travel route of target B or reducing the operating speed of target A.

[0031] In the embodiments of the present application, the original data is filtered and denoised by the data credibility value, solving the problem of differences in the quality of multi-source heterogeneous data and improving the data reliability. Secondly, a unified spatio-temporal reference is established by using timestamps and coordinate transformation, and the matching degree between data sources is calculated based on the similarity of target features, realizing the precise alignment and association between different data sources. Thirdly, feature parameters are extracted from three dimensions of position, speed, and motion trajectory and correlation analysis is performed to construct a multi-dimensional target feature description, enhancing the expression ability of the target state. The potential conflicts are judged by setting safety distance thresholds and speed thresholds, and an evaluation index system is constructed by combining historical risk event data, realizing the precise quantification of risks. At the same time, by establishing the spatial association relationship and risk propagation relationship between apron targets and introducing the concept of risk accumulation effect, the distribution and propagation characteristics of risks in space are accurately described. Finally, the objective function value is calculated based on real-time safety situation data and constraint conditions are set, and an optimization algorithm is used to solve the control strategy parameters, realizing the collaborative optimization of multiple targets on the premise of ensuring safety.

[0032] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0033] (1) Segment the video data according to a fixed time window, divide the segmented video data into an image frame sequence, calculate the data integrity according to the pixel gray values in the image frame sequence, and obtain the video data integrity index;

[0034] (2) Detect the sampling rate of the radar data, extract the signal waveform in the radar data, and count the number of signal interruptions according to the continuity of the waveform to obtain the radar data integrity index;

[0035] (3) Parse the time stamps of the position data, count the variance of the sampling intervals of the position data, and judge the data acquisition stability according to the size of the variance to obtain the position data integrity index;

[0036] (4) Perform weighted summation on the video data integrity index, the radar data integrity index, and the position data integrity index to generate a comprehensive integrity score;

[0037] (5) Compare the time stamps of the video data, the radar data, and the position data, calculate the deviation value of the data arrival time, and generate a time consistency score;

[0038] (6) Calculate the data credibility value according to the weighted combination of the comprehensive integrity score and the time consistency score, filter the data by the set threshold, and eliminate the noise points to obtain the preprocessed data set.

[0039] Specifically, when processing video data, a fixed-time window refers to dividing a continuous video stream into segments of a fixed time length. For example, video data is divided into a time window every 50 milliseconds. Within each time window, the video data is decomposed into a series of consecutive image frames, and each image frame contains the complete information of the apron scene at a specific moment. For calculating the data integrity of the pixel gray values in the image frame sequence, the following formula is used:

[0040] ;

[0041] where represents the video data integrity index, W represents the total number of frames within the time window, M and N respectively represent the number of rows and columns of the image, represents the pixel gray value at the coordinate (x, y) in the i-th frame image, is the pixel quality evaluation function.

[0042] The processing of radar data focuses on the detection of the sampling rate. The signal waveform is extracted from the original radar data, and the waveform data contains the distance, azimuth, and speed information of the target. As Figure 2 shown, it is the radar multi-dimensional information waveform diagram provided by the embodiment of the present invention. Among them, Figure 2 includes three subgraphs, which respectively represent the distance signal waveform, the azimuth signal waveform, and the speed signal waveform. Among them, the distance signal waveform is represented by a red curve, which is used to display the relative distance change between the target and the radar; the azimuth signal waveform is represented by a blue curve, which is used to display the azimuth angle change of the target relative to the radar; the speed signal waveform is represented by a green curve, which is used to display the movement speed change of the target. The black dots on each waveform represent the actual sampling points, which are used to reflect the sampling rate of the radar data. The vertical coordinates in the figure respectively represent the amplitude of the distance signal, the angle value of the azimuth signal, and the speed value of the speed signal, and the horizontal coordinate represents time. By detecting the sampling rate and analyzing the waveform continuity of these three-dimensional signals, the integrity index of the radar data is calculated, and then the data quality is evaluated. When the waveform is discontinuous or sampling points are missing, it indicates that there is a signal interruption in the radar data, and other data sources need to be used for supplementation. The continuity of the waveform is judged by statistically analyzing the time interval between adjacent sampling points. When the time interval exceeds the preset threshold, it is recorded as a signal interruption. The ratio of the number of signal interruptions to the total number of sampling points is used as the integrity index of the radar data. The time stamp parsing of the position data includes extracting the time stamp information from the data packet and converting it into a unified time format. The calculation of the sampling interval variance is carried out by statistically analyzing the time difference between adjacent sampling points. The smaller the variance value, the more stable the data acquisition. The position data integrity index is obtained by normalizing according to the size of the sampling interval variance.

[0043] The generation of the comprehensive integrity score adopts the method of weighted summation, and different weights are assigned to the integrity indicators of the three data sources. The setting of the weights is based on the importance of the data sources in safety supervision, and the sum of the weights of video data, radar data, and position data is 1. The calculation of the time consistency score aligns the time stamps of the three data sources to the same time reference, and then calculates the deviation between the actual arrival time and the theoretical arrival time of each data source. The smaller the deviation value, the better the time consistency of the data. The time consistency score of the data is obtained by normalizing the deviation value.

[0044] The calculation of the data credibility value performs a weighted combination of the comprehensive integrity score and the time consistency score. The setting of the weight coefficient reflects the relative importance of integrity and consistency in data quality assessment. During the data filtering process, the data credibility value is compared with a preset threshold, and data points below the threshold are marked as noise points and excluded.

[0045] For example, taking the process of an aircraft taxiing out from a parking position to a taxiway as an example, through the gray value analysis of the image frame sequence collected by the video monitoring device, it is found that some frames are blurred. There is a brief signal loss in the radar data when the aircraft turns, and the position data has uneven sampling intervals due to occlusion. By calculating the comprehensive integrity score and the time consistency score, the reliability of the information from these three data sources is evaluated. It is found that the credibility of the radar data is relatively low during the turning stage. Therefore, during the generation of the preprocessed data set, the fusion results of video data and position data are mainly relied on in this stage, thus ensuring the accuracy of the data. The entire data preprocessing process effectively screens out reliable data sources and excludes unreliable data points through the quantitative assessment of the quality of multi-source data, providing high-quality input data for subsequent data fusion.

[0046] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0047] (1) Parse the time stamps in the preprocessed data set, align the sampling moments of different data sources to a unified time axis, and generate a time synchronization sequence;

[0048] (2) Perform a reference transformation on the coordinate data in the time synchronization sequence, map the position information in different coordinate systems to a unified apron coordinate system, and obtain spatially aligned data;

[0049] (3) Extract the geometric dimensions, motion directions, and speed information of the target from the spatially aligned data, and construct a target feature description set;

[0050] (4) Calculate the Euclidean distance between feature vectors according to the target feature description set, and generate a similarity matrix between data sources;

[0051] (5) Normalize the similarity matrix, and screen out the optimal matching pairs by setting a matching threshold to form a target association table;

[0052] (6) Merge the information of multi-source data according to the target association table, and fuse the target parameters by using the weighted average method to obtain the fused target data.

[0053] Specifically, parse the timestamps in the preprocessed dataset. The timestamp is an accurate record of the data collection moment, including year, month, day, hour, minute, second and millisecond information. The sampling moments of different data sources need to be aligned to a unified time axis. Uniformly use UTC time as the benchmark, perform time zone conversion and millisecond-level accuracy alignment on the timestamps of each data source to generate a time synchronization sequence. For the coordinate data in the time synchronization sequence, a datum transformation is required to unify the spatial reference system. The image coordinates in video data use the pixel coordinate system, radar data uses the polar coordinate system, and position data uses the longitude and latitude coordinate system. Unify the position information in these different coordinate systems to the local coordinate system of the airport apron. The apron coordinate system takes the runway entrance as the origin and establishes a northeast celestial coordinate system. During the coordinate transformation process, the installation positions and attitude parameters of different devices need to be considered, and various data are mapped to a unified spatial benchmark through a coordinate transformation matrix.

[0054] Extract target features from the spatially aligned data. The geometric dimensions include the length, width and height of the target, which are obtained through image segmentation and bounding box extraction. The motion direction is obtained by calculating the change in the target position at adjacent moments, and the speed information is calculated by dividing the displacement distance by the time interval. The target feature description set organizes these features into a unified data structure, and each target corresponds to a feature vector.

[0055] The Euclidean distance between feature vectors is calculated using the following formula:

[0056] ;

[0057] where, represents the distance between the i-th feature vector and the j-th feature vector, and respectively represent the k-th components of the two feature vectors, represents the weight coefficient of the k-th feature, represents the time decay factor, and n represents the feature dimension. By calculating the distances between all pairs of feature vectors, a similarity matrix between data sources is generated.

[0058] The normalization of the similarity matrix adopts the maximum-minimum normalization method, mapping the similarity values to the range of 0 to 1. By setting a matching threshold to screen for the optimal matching pairs, when the similarity exceeds the threshold, it is considered that the two data sources describe the same target, ensuring that the target of one data source matches at most one target of another data source, and forming a target association table.

[0059] According to the target association table, information merging is performed on multi-source data. For the matching targets, the weighted average method is used to fuse their parameters. The setting of the weights is based on the data credibility, and the data source with high credibility has a larger weight. For the position information, the weights are determined by combining the GPS positioning accuracy and the radar ranging accuracy; for the speed information, the weights are set according to the Doppler velocity measurement accuracy and the image target tracking accuracy.

[0060] For example: when an aircraft is taxiing on the apron, the video surveillance equipment collects the image sequence of the target, the radar equipment records the distance and azimuth data of the target, and the ADS-B receiver obtains the position broadcast information of the target. The timestamps of these data are unified to UTC time, and the frame timestamps of the video data, the radar scanning period, and the ADS-B broadcast time form an aligned time series. Then, the target pixel coordinates in the image, the polar coordinate measurements of the radar, and the longitude and latitude information of the ADS-B are all converted into the apron coordinate system. The characteristic information of the aircraft is extracted from the converted data, including the fuselage size, taxiing direction, and ground speed. The distances between the characteristic vectors obtained from different data sources are calculated to generate a similarity matrix, and it is confirmed through threshold screening that these data describe the same aircraft. According to the accuracy characteristics of each data source, the position and speed of the target are weighted and fused to obtain a more accurate target state estimate.

[0061] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0062] (1) Extract the position coordinates from the fused target data, calculate the instantaneous longitude, latitude, and altitude values of the target, and obtain the target position characteristic parameters;

[0063] (2) Perform time difference operation on the position characteristic parameters, calculate the instantaneous speed and acceleration values of the target, and obtain the target speed characteristic parameters;

[0064] (3) Perform time series connection on the position characteristic parameters, construct the continuous motion trajectory point set of the target, and obtain the target trajectory characteristic parameters;

[0065] (4) Combine the position characteristic parameters, speed characteristic parameters, and trajectory characteristic parameters to generate a feature vector, and construct the original feature data set;

[0066] (5) Standardize the parameters in the original feature dataset, calculate the Pearson correlation coefficient between the parameters, and obtain the feature correlation matrix;

[0067] (6) Combine the features with correlation coefficients greater than the threshold in the feature correlation matrix, and perform dimensionality reduction and reconstruction on the features through principal component analysis to form the target feature matrix.

[0068] Specifically, to process the fused target data, the position coordinates of the target need to be extracted. The position coordinates include three dimensions: longitude, latitude, and altitude. The longitude and latitude values are represented in decimal degrees, accurate to 6 decimal places after the decimal point, and the altitude value is in meters, based on the WGS84 ellipsoid datum. Extract these coordinate values from the fused target data to form a spatial position description of the target at a specific moment.

[0069] For the instantaneous velocity of the target and acceleration calculations, the time difference method is used, and its calculation formula is:

[0070] ;

[0071] where represents the position vector at time t, is the sampling time interval, and are high-order correction coefficients respectively, and are the time window weights, and H and J are the sizes of the historical data windows.

[0072] The continuous motion trajectory of the target is obtained by connecting the position feature parameters in time series. The trajectory point set contains the position information of the target at different times. Each trajectory point contains a timestamp, coordinate values, and a sampling serial number. Connect these points in time order to form a complete motion trajectory.

[0073] The construction of the feature vector combines three types of feature parameters: position, velocity, and trajectory. The position feature includes the three-dimensional coordinates at the current moment, the velocity feature includes the linear velocity and angular velocity, and the trajectory feature includes the relative position relationship of the historical trajectory points. These features form a multi-dimensional vector, and each dimension represents an attribute feature of the target. When standardizing the original feature dataset, the Z-score standardization method is used to convert features with different dimensions to the same scale. Calculate the Pearson correlation coefficient between the standardized features. The Pearson correlation coefficient reflects the linear correlation degree between the features, and its value range is between -1 and 1. The calculation results of the correlation coefficients form the feature correlation matrix.

[0074] Principal component analysis is used for feature dimensionality reduction. Feature combinations with a correlation coefficient greater than the threshold are selected. Then, the eigenvalues and eigenvectors of the feature covariance matrix are calculated, and several principal component directions with the largest contribution rate are selected to perform a projection transformation on the original features to obtain the feature representation after dimensionality reduction.

[0075] For example, the position coordinates of an aircraft at a certain moment are extracted from the fusion data, including longitude, latitude, and altitude. Based on the continuous position sampling data, the instantaneous speed of the aircraft, including ground speed and vertical speed components, is calculated through time difference, and then the acceleration value is calculated through the change in speed. These position points are connected in chronological order to form the taxiing trajectory of the aircraft. The feature vector contains the current position, movement speed, acceleration, and trajectory characteristics of the aircraft. After standardizing these features, the correlation between features is calculated, and it is found that there is a strong correlation between speed and trajectory curvature, indicating that the aircraft will reduce speed when turning. Through principal component analysis, the original high-dimensional features are reduced to several main feature components, and these feature components comprehensively reflect the motion state of the aircraft.

[0076] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0077] (1) Calculate the relative distance between targets for the position data in the target feature matrix, compare the relative distance with the safety distance threshold, and obtain the distance risk index;

[0078] (2) Calculate the relative speed of the targets for the speed data in the target feature matrix, compare the relative speed with the speed threshold, and obtain the speed risk index;

[0079] (3) Combine the distance risk index and the speed risk index to generate the risk state vector at the current moment, and obtain the target potential conflict probability;

[0080] (4) Conduct statistical analysis on the historical risk event data, extract the occurrence frequency and harm degree of different types of risk events, and generate the historical risk weight;

[0081] (5) Perform a weighted calculation on the target potential conflict probability and the historical risk weight, construct a comprehensive risk assessment function, and obtain the risk quantification value;

[0082] (6) Normalize the risk quantification value, and convert it into a risk level through hierarchical mapping to obtain the safety risk score.

[0083] Specifically, the position data in the target feature matrix is processed. The target feature matrix contains the position information of all targets on the apron. The relative distance values are obtained by calculating the Euclidean distance between pairwise targets. The safety distance threshold is the minimum separation distance determined according to the safety operation specifications of different types of targets. Different thresholds are set for between aircraft and aircraft, between aircraft and ground vehicles, and between ground vehicles and ground vehicles. The calculated relative distance is compared with the corresponding safety distance threshold to generate a distance risk indicator, which reflects the spatial proximity between targets. For the speed data in the target feature matrix, the relative speed between any two targets is calculated. The calculation of relative speed needs to consider the directionality of target movement. The speed vectors of the two targets are subjected to vector subtraction operation to obtain the relative speed vector. The speed threshold is set based on the safety management regulations of the airport apron, including speed limit requirements in different areas and speed limits under different weather conditions. By comparing the relative speed value with the speed threshold, a speed risk indicator is generated.

[0084] The combination of the distance risk indicator and the speed risk indicator constitutes the risk state vector at the current moment. The risk state vector describes the possibility of a conflict occurring between two targets. Through the prediction of the target's motion state and path extrapolation analysis, the potential conflict probability of the target is obtained. The calculation of the potential conflict probability comprehensively considers factors such as the current distance, relative speed, motion direction, and acceleration of the target. For the analysis of historical risk event data, the risk events are classified according to types, including collision accidents, intrusion into restricted areas, speeding, etc. The occurrence frequencies of each type of risk event in different time periods and different regions are counted, and their proportions in the overall risk events are calculated. The assessment of the harm degree is based on the consequences caused by the event, including multiple dimensions such as casualties, property losses, and flight delays. Through the mining and analysis of historical data, corresponding weight values are assigned to each type of risk event.

[0085] The weighted calculation of the target potential conflict probability and the historical risk weight adopts a multi-level analysis method. A risk assessment index system is established, with the current conflict probability and historical statistical data as the first-level indicators, each containing several second-level indicators. The weight coefficients of each indicator are determined through hierarchical analysis, and the weighted results of all indicators are comprehensively obtained to get the risk quantification value. The risk quantification value reflects the comprehensive risk level in the current scenario. The normalization process of the risk quantification value adopts a piecewise linear mapping method, mapping the risk value to the interval of 0 to 1. The hierarchical mapping divides the risk level into multiple levels according to the safety management requirements of the airport, such as low risk, medium risk, high risk, etc. Each risk level corresponds to a score interval, and the normalized risk quantification value is converted into a specific risk level through interval mapping to obtain the safety risk score.

[0086] For example, taking the intersection scenario of an aircraft and a shuttle bus on the airport apron as an example, the data processing process of risk assessment is described in detail. The position data of the aircraft and the shuttle bus are extracted from the target feature matrix, the relative distance between the two is calculated, and this distance is compared with the specified minimum safety interval between the aircraft and the ground vehicle. At the same time, the relative speed of the two targets is calculated. Considering that the aircraft is taxiing and the shuttle bus is crossing the taxiway laterally, the speed vectors of the two need to be analyzed. Combining the distance and speed indicators, the movement trajectories of the two targets in the next few seconds are predicted, and the potential intersection risks are evaluated. Querying the historical database, it is found that similar crossing events have occurred in this area, and some of these events have caused flight delays. Therefore, this type of risk event has a relatively high weight. Combining the current conflict probability and the historical risk weight, the risk quantification value is calculated and converted into the corresponding risk level, so as to give an accurate risk assessment result for this intersection scenario.

[0087] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0088] (1) Assign weights to the safety risk scores, calculate the distance relationship between the targets, and obtain the spatial influence coefficient matrix;

[0089] (2) Perform distance attenuation calculation on the spatial influence coefficient matrix, analyze the risk transmission intensity between adjacent targets, and obtain the risk propagation matrix;

[0090] (3) Perform iterative operations on the risk propagation matrix, count the cumulative risk values on the risk propagation chain, and obtain the risk cumulative effect value;

[0091] (4) Perform spatial distribution mapping on the risk cumulative effect value, calculate the total risk accumulation per unit area, and obtain the regional risk density;

[0092] (5) Organize the regional risk density according to the time series, construct a dynamic risk distribution map, and generate a regional situation change sequence;

[0093] (6) Perform spatio-temporal correlation analysis on the regional situation change sequence, combine with the risk level distribution for data integration, and generate real-time safety situation data.

[0094] Specifically, weights are assigned to the processing of the safety risk scores, and the weight values are determined based on the target type, activity area, and time period. Different types of targets (such as aircraft, ground vehicles, and personnel) have different risk impact weights. The distance relationship between the targets is obtained by calculating the Euclidean distance, forming the spatial influence coefficient matrix, which describes the spatial correlation intensity between all targets on the apron.

[0095] The following formula is used for the distance attenuation calculation of the spatial influence coefficient matrix:

[0096] ;

[0097] wherein, represents the risk transfer intensity of target i to target j, is the benchmark transfer coefficient of the k-th type of risk, is the distance attenuation rate, is the distance between targets, is the direction influence factor, is the environmental factor, is the barrier coefficient of the m-th type of obstacle, is the obstacle distribution function.

[0098] The iterative operation process of the risk propagation matrix considers the chain transfer of risks among multiple targets. Determine the initial risk source, and then calculate the cumulative effect of the risk transferred to other targets through different paths. The risk state of the target is updated in each iteration until the risk transfer intensity is lower than the threshold or the maximum number of iterations is reached. The risk cumulative effect value reflects the comprehensive influence of all surrounding risk sources on a certain target. The spatial distribution mapping of the risk cumulative effect value divides the apron area into grid cells. For each grid cell, count the risk cumulative effect values of all targets within its range, calculate the total risk within the unit area, and obtain the regional risk density. The calculation of the risk density considers the spatial distribution characteristics of the targets and the risk influence range.

[0099] The time series organization of the regional risk density sorts the risk density data within continuous time periods according to the time stamp. The dynamic risk distribution map shows the spatial distribution and time evolution characteristics of the risk level through the visual expression of the risk density. The regional situation change sequence records the dynamic change process of the risk distribution. The spatio-temporal correlation analysis focuses on discovering the correlation patterns of the risk distribution in the time and space dimensions. By analyzing the change trends of the risk distribution at adjacent moments and combining the risk level distributions in different regions, the risk situation data are integrated to form a complete real-time safety situation description.

[0100] For example: During a busy period at the airport apron, multiple aircraft are operating simultaneously, including an aircraft A being pushed back, an aircraft B taxiing, and several ground support vehicles. Based on their safety risk scores, aircraft A is assigned a higher weight because it is in the process of being pushed back, followed by aircraft B, and the ground vehicles have a lower weight. Calculate the distance relationships between these targets to generate a spatial influence coefficient matrix. Perform distance attenuation calculation on this matrix, taking into account the blocking effect of fixed facilities on the apron on risk propagation, to obtain the actual risk propagation intensity. Through multiple rounds of iterative calculation, analyze the process of risk spreading from aircraft A to surrounding targets. The risk impact on aircraft B attenuates with increasing distance, but since it is also a risk source itself, a secondary risk accumulation area is formed around it. Map these risk accumulation effects to the apron grid and calculate the risk density distribution of each area. As time goes by, record the change process of the risk density and find that the risk distribution shows an obvious migration characteristic as the aircraft moves. Through spatio-temporal correlation analysis, integrate this dynamically changing information into complete safety situation data.

[0101] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0102] (1) Analyze the real-time safety situation data, extract the risk level and spatial distribution information, and obtain the current situation characteristic quantity;

[0103] (2) Substitute the current situation characteristic quantity into the safety control function, calculate the ratio of risk control cost to benefit, and obtain the objective function value;

[0104] (3) Set constraint conditions for the objective function value, convert the safety distance requirement and speed limit into mathematical constraints, and obtain a constraint equation system;

[0105] (4) Combine the constraint equation system and the objective function value to construct an optimization problem, and calculate the optimal solution through gradient descent to obtain the initial control parameters;

[0106] (5) Conduct a conflict check on the initial control parameters, analyze the degree of mutual influence between the parameters, and obtain the feasible control range;

[0107] (6) Refine and adjust the parameters within the feasible control range, select the optimal control scheme, and form the control strategy parameters.

[0108] Specifically, when processing real-time security situation data, risk level information is extracted. The risk level includes three levels: high, medium, and low, and each level corresponds to a different numerical range. The spatial distribution information records the distribution state of risks within the apron area, including the coordinate positions and influence ranges of risk points. The current situation characteristic quantities are composed of multiple indicators such as risk level, distribution density, and spatial aggregation degree. The construction of the security control function needs to consider multiple factors, including the implementation cost of control measures, the expected risk value reduction, and the execution difficulty. The risk control cost includes human resource input, equipment scheduling cost, and impact on operating efficiency. The benefits are reflected in aspects such as the degree of risk reduction and the improvement of safety margin. The objective function value is obtained through the calculation of the cost-benefit ratio, and this function value reflects the overall quality of the control plan.

[0109] The setting of constraint conditions is based on airport operation specifications and safety management requirements. The safety distance requirement stipulates the minimum interval that must be maintained between different types of targets, and the speed limit sets the maximum allowable speed of various vehicles and aircraft in different areas. These physical constraints are transformed into mathematical expressions to form a system of constraint equations. The system of constraint equations includes equality constraints and inequality constraints to ensure that the control plan meets safety requirements. The construction of the optimization problem combines the objective function and the system of constraint equations, and the gradient descent method is used to solve the optimal solution. During the gradient descent process, the control parameters are randomly initialized, and then the gradient of the objective function with respect to each parameter is calculated, and the parameter values are updated along the gradient direction until convergence to a local optimal solution. The initial control parameters include specific control quantities such as speed adjustment amount and path offset amount.

[0110] During the conflict check of the initial control parameters, it is necessary to analyze the correlation between different control parameters. The speed adjustment of a certain target will affect its relative position relationship with other targets, and the path adjustment will change the intersection time sequence of multiple targets. By establishing a parameter influence relationship diagram, the positive promotion and negative inhibition relationships between parameters are identified, and the feasible value ranges of each parameter are determined. When making a refined adjustment of parameters within the feasible control range, the grid search method is used. The value range of each parameter is divided into several discrete points, and the optimal parameter combination is found through exhaustive combination or heuristic search. The optimization objectives include multiple dimensions such as risk control effect, operating efficiency, and implementation difficulty, and the plan with the highest comprehensive score is selected as the control strategy parameter.

[0111] For example, real-time safety situation data shows that there is a high-risk area near the No. 1 position, where there is an aircraft about to push back, a refueling vehicle, and two baggage carts. The risk level and spatial distribution information indicate that the risk mainly stems from the conflict between the refueling operation and the timing of aircraft push-back. According to the current situation characteristic quantities, calculate the cost-benefit ratios of different control schemes, including options such as adjusting the refueling vehicle operation time and changing the aircraft push-back direction. The constraint conditions require that the refueling vehicle and the aircraft maintain a safety distance of not less than 5 meters, and the aircraft push-back speed does not exceed 5 kilometers per hour. After converting these constraints into mathematical expressions, through optimization calculation, the initial control parameters are obtained: delaying the aircraft push-back time. Conduct a conflict check on this initial scheme, analyze the impact on the subsequent flight support work, and obtain a feasible solution set including the time adjustment range. Finally, through refined adjustment, comprehensively considering the coordinated cooperation of multiple objectives, a specific control strategy is formed: delaying the aircraft push-back by 2 minutes, and at the same time adjusting the working order of other support vehicles, which not only ensures the safety interval but also reduces the impact on the overall operation efficiency.

[0112] In a specific embodiment, the process of performing the step of conflict checking on the initial control parameters may specifically include the following steps:

[0113] (1) Conduct a numerical range analysis on the initial control parameters, count the upper and lower limits of the values of each parameter, and obtain the parameter change interval;

[0114] (2) Cross-combine the parameter change intervals, calculate the control effects under different parameter combinations, and obtain the parameter correlation matrix;

[0115] (3) Conduct a sensitivity analysis on the parameter correlation matrix, calculate the influence intensity of parameter changes on the control effect, and obtain the parameter sensitivity value;

[0116] (4) Arrange the parameter sensitivity values in descending order, identify the key influencing parameters and secondary influencing parameters, and obtain the parameter importance sequence;

[0117] (5) Use the parameter importance sequence as the weight coefficient to perform weighted constraints on the parameter value range, and obtain the parameter adjustment boundary;

[0118] (6) Conduct a feasibility verification on the parameter adjustment boundary, eliminate the value intervals that do not meet the constraint conditions, and obtain the feasible control range.

[0119] Specifically, when performing a numerical range analysis on the initial control parameters, it is necessary to consider the physical meaning and actual constraints of the parameters. The control parameters include the speed adjustment amount, distance threshold, time interval, etc., and each parameter has a reasonable value range. The speed adjustment amount is determined by the airport operation specifications, such as the allowable change range of the taxiing speed; the distance threshold is determined based on the safety interval requirements; and the time interval needs to consider the flight schedule and support operation requirements. Statistical analysis is performed on the historical data of each parameter to extract the maximum and minimum values, and the reasonable change interval of the parameter is determined in combination with expert experience. The cross-combination of the parameter change intervals uses the grid search method, which equally divides the value range of each parameter to form a discrete set of value points. The value points of different parameters are fully combined, and each combination corresponds to a specific set of control parameter values. Simulation calculations are performed on each set of parameter combinations to evaluate their impact on the control objectives, such as the degree of reduction in the risk level and the change in operating efficiency. The evaluation results are organized into a parameter correlation matrix, and the matrix elements represent the contribution degree of the parameter combination to the control effect.

[0120] The sensitivity analysis of the parameter correlation matrix adopts the single-factor change method. By fixing other parameters unchanged and changing the value of a single parameter alone, the change amplitude of the control effect is observed. By calculating the partial derivative of the control effect with respect to the parameter change, the sensitivity value of the parameter is obtained. The larger the sensitivity value, the more significant the impact of the parameter on the control effect. This process is repeated for all parameters to obtain a complete set of parameter sensitivity values. The descending order of the parameter sensitivity values sorts all parameters according to their influence intensity from large to small. The parameters ranked in the front play a decisive role in the control effect and are defined as key influencing parameters; the parameters ranked in the back have relatively less influence and are called secondary influencing parameters. This division helps to determine the priority order of parameter adjustment in actual control and form a parameter importance sequence.

[0121] When converting the parameter importance sequence into weight coefficients, normalization processing is adopted to make the sum of all weights equal to 1. The parameters with larger weight coefficients have higher priorities during the adjustment process, and their value ranges are subject to stricter constraints. Multiply the original value range of each parameter by the corresponding weight coefficient to obtain the weighted parameter adjustment boundary. The feasibility verification of the parameter adjustment boundary includes static constraint testing and dynamic conflict testing. Static constraint testing ensures that the parameter values meet the physical limitations and operation specifications, and dynamic conflict testing verifies the feasibility of the parameter combination in actual operation. By eliminating the value intervals that do not meet the constraint conditions, the feasible control range is obtained.

[0122] Taking the optimization of control parameters coordinated between aircraft pushback and ground support vehicles as an example, it involves parameters such as aircraft pushback time, pushback speed, and ground vehicle avoidance distance. According to airport operation regulations, the adjustment range of aircraft pushback time is plus or minus 5 minutes, the pushback speed range is 2 - 5 kilometers per hour, and the ground vehicle avoidance distance is not less than 5 meters. These restrictions constitute the initial change interval of the parameters. Combine these parameters for analysis, such as a 2 - minute delay in pushback time combined with a pushback speed of 3 kilometers per hour, and evaluate the impact of each combination on the overall operation. Through sensitivity analysis, it is found that the adjustment of pushback time has the greatest impact on conflict avoidance, followed by the avoidance distance, and the impact of pushback speed is relatively small. Based on this, the parameter importance ranking is obtained: pushback time adjustment, avoidance distance, pushback speed. Convert this importance sequence into weight coefficients to impose weighted constraints on the adjustment range of the parameters. Finally, through feasibility verification, comprehensively considering constraint conditions such as aircraft stand occupancy time and taxiway congestion, determine the parameter adjustment plan.

[0123] The above described the multi - data fusion method in airport apron safety supervision in the embodiments of the present application. Next, the multi - data fusion system in airport apron safety supervision in the embodiments of the present application will be described. Please refer to Figure 2 , an embodiment of the multi - data fusion system in airport apron safety supervision in the embodiments of the present application includes:

[0124] A generation module 201, configured to generate a data credibility value for the video data, radar data, and position data collected by the apron multi - source sensors according to data integrity and time consistency, filter and denoise the original data through the data credibility value, and obtain a pre - processed data set;

[0125] A fusion module 202, configured to, for the pre - processed data set, establish a unified spatio - temporal reference through timestamp and coordinate conversion, calculate the matching degree between different data sources based on the target feature similarity, and obtain the fused target data;

[0126] An extraction module 203, configured to extract feature parameters from the fused target data in three dimensions of position, speed, and motion trajectory, screen and combine the feature parameters according to the correlation coefficient, and form a target feature matrix;

[0127] A judgment module 204, configured to judge potential conflicts based on the target feature matrix through set safety distance thresholds and speed thresholds, construct a risk assessment index in combination with historical risk event data, and obtain a safety risk score;

[0128] A establishment module 205, configured to establish a spatial association relationship and a risk propagation relationship between apron targets according to the safety risk score, calculate the regional risk density through the risk accumulation effect, and generate real - time safety situation data;

[0129] A solution module 206 is configured to calculate the objective function value of security control for real-time security situation data, set security constraint conditions for optimal solution, and form control strategy parameters.

[0130] Through the collaborative cooperation of the above-mentioned various components, the original data is filtered and denoised by the data credibility value, solving the problem of quality differences in multi-source heterogeneous data and improving data reliability. Secondly, a unified spatio-temporal reference is established by using timestamps and coordinate transformation, and the matching degree between data sources is calculated based on the similarity of target features, realizing the precise alignment and association between different data sources. Thirdly, feature parameters are extracted from three dimensions of position, speed, and motion trajectory and correlation analysis is carried out to construct a multi-dimensional target feature description, enhancing the expression ability of the target state. Potential conflicts are judged by setting safety distance thresholds and speed thresholds, and an evaluation index system is constructed in combination with historical risk event data, realizing the precise quantification of risks. At the same time, by establishing the spatial association relationship and risk propagation relationship between apron targets and introducing the concept of risk accumulation effect, the distribution and propagation characteristics of risks in space are accurately characterized. Finally, the objective function value is calculated based on real-time security situation data and constraint conditions are set, and an optimization algorithm is used to solve the control strategy parameters, realizing the collaborative optimization of multiple objectives while ensuring safety.

[0131] The present application also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions run on a computer, the computer is made to execute the steps of the multi-data fusion method in the airport apron security supervision.

[0132] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described system, system, and unit can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0133] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0134] As described above, the above embodiments are only used to illustrate the technical solutions of this application and are not intended to limit them; although this application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of various embodiments of this application.

Claims

1. A multi-data fusion method in airport apron safety supervision, characterized in that, Including: For the video data, radar data, and position data collected by apron multi-source sensors, generate data credibility values according to data integrity and time consistency, filter and denoise the original data through the data credibility values to obtain a preprocessed data set; For the preprocessed data set, establish a unified spatio-temporal reference through timestamp and coordinate transformation, calculate the matching degree between different data sources based on the similarity of target features, and obtain the fused target data; According to the fused target data, extract feature parameters from three dimensions of position, speed, and motion trajectory, and screen and combine the feature parameters according to the correlation coefficient to form a target feature matrix; Based on the target feature matrix, judge potential conflicts through the set safety distance threshold and speed threshold, and construct a risk assessment index in combination with historical risk event data to obtain a safety risk score; According to the safety risk score, establish the spatial association relationship and risk propagation relationship between apron targets, calculate the regional risk density through the risk accumulation effect, and generate real-time safety situation data, including assigning weights to the safety risk score, calculating the distance relationship between targets, and obtaining the spatial influence coefficient matrix; Perform distance attenuation calculation on the spatial influence coefficient matrix, analyze the risk transmission intensity between adjacent targets, and obtain the risk propagation matrix; Perform iterative operations on the risk propagation matrix, count the cumulative risk values on the risk propagation chain, and obtain the risk accumulation effect value; Perform spatial distribution mapping on the risk accumulation effect value, calculate the total risk accumulation per unit area, and obtain the regional risk density; Organize the regional risk density according to the time series, construct a dynamic risk distribution map, and generate a regional situation change sequence; perform spatio-temporal correlation analysis on the regional situation change sequence, and integrate data in combination with the risk level distribution to generate real-time safety situation data; For the real-time safety situation data, calculate the objective function value of safety control, set safety constraint conditions for optimization and solution, and form control strategy parameters.

2. The multi-data fusion method in airport apron safety supervision according to claim 1, characterized in that, For the video data, radar data, and position data collected by apron multi-source sensors, generate data credibility values according to data integrity and time consistency, filter and denoise the original data through the data credibility values to obtain a preprocessed data set, including: Segment the video data according to a fixed time window, divide the segmented video data into an image frame sequence, calculate the data integrity according to the pixel gray values in the image frame sequence, and obtain the video data integrity index; Perform sampling rate detection on the radar data, extract the signal waveform in the radar data, and count the number of signal interruptions according to the waveform continuity to obtain the radar data integrity index; Perform time mark parsing on the position data, count the variance of the sampling interval of the position data, and judge the data acquisition stability according to the variance size to obtain the position data integrity index; Perform weighted summation on the video data integrity index, radar data integrity index, and position data integrity index to generate a comprehensive integrity score; Compare the time marks of the video data, radar data, and position data, calculate the deviation value of the data arrival time, and generate a time consistency score; Calculate the data credibility value based on the weighted combination of the comprehensive integrity score and the time consistency score, filter the data through a set threshold, and eliminate the noise points to obtain a preprocessed data set.

3. The multi-data fusion method in airport apron safety supervision according to claim 1, characterized in that For the preprocessed data set, establish a unified spatio-temporal benchmark through timestamp and coordinate transformation, calculate the matching degree between different data sources based on the target feature similarity, and obtain the fused target data, including: parsing the timestamps in the preprocessed data set, aligning the sampling moments of different data sources to a unified time axis, and generating a time synchronization sequence; Perform a reference transformation on the coordinate data in the time synchronization sequence, map the position information in different coordinate systems to a unified apron coordinate system, and obtain spatially aligned data; Extract the geometric dimensions, motion directions, and speed information of the target from the spatially aligned data to construct a target feature description set; Calculate the Euclidean distance between feature vectors based on the target feature description set to generate a similarity matrix between data sources; Normalize the similarity matrix, screen the optimal matching pairs through a set matching threshold, and form a target association table; Merge the information of multi-source data according to the target association table, and fuse the target parameters using the weighted average method to obtain the fused target data.

4. The multi-data fusion method in airport apron safety supervision according to claim 1, characterized in that According to the fused target data, extract feature parameters from three dimensions: position, speed, and motion trajectory, and screen and combine the feature parameters according to the correlation coefficient to form a target feature matrix, including: Extract the position coordinates of the fused target data, calculate the instantaneous longitude, latitude, and altitude values of the target to obtain the target position feature parameters; Perform a time difference operation on the position feature parameters to calculate the instantaneous speed and acceleration values of the target to obtain the target speed feature parameters; Connect the position feature parameters in time series to construct a set of continuous motion trajectory points of the target to obtain the target trajectory feature parameters; Combine the position feature parameters, speed feature parameters, and trajectory feature parameters to generate feature vectors and construct an original feature data set; Standardize the parameters in the original feature data set, calculate the Pearson correlation coefficient between the parameters to obtain a feature correlation matrix; Combine the features with correlation coefficients greater than the threshold in the feature correlation matrix, and perform dimensionality reduction and reconstruction on the features through principal component analysis to form a target feature matrix.

5. The multi-data fusion method in airport apron safety supervision according to claim 1, characterized in that, Based on the target feature matrix, judge potential conflicts through set safety distance thresholds and speed thresholds, and construct a risk assessment index by combining historical risk event data to obtain a safety risk score, including: Calculate the relative distance between targets for the position data in the target feature matrix, compare the relative distance with the safety distance threshold to obtain a distance risk index; Calculate the relative speed of the target for the speed data in the target feature matrix, compare the relative speed with the speed threshold to obtain a speed risk index; Combine the distance risk index and the speed risk index to generate a risk status vector at the current moment to obtain the target potential conflict probability; Conduct statistical analysis on the historical risk event data, extract the occurrence frequency and harm degree of different types of risk events to generate historical risk weights; Perform a weighted calculation of the target potential conflict probability and the historical risk weights, construct a comprehensive risk assessment function to obtain a risk quantification value; Normalize the risk quantification value, convert it into a risk level through hierarchical mapping, and obtain the safety risk score.

6. The multi-data fusion method in airport apron safety supervision according to claim 1, wherein, For real-time safety situation data, calculate the objective function value of safety control, set safety constraints for optimization, and form control strategy parameters, including: Parse the real-time safety situation data, extract the risk level and spatial distribution information, and obtain the current situation feature quantity; Substitute the current situation feature quantity into the safety control function, calculate the ratio of risk control cost to benefit, and obtain the objective function value; Set constraint conditions for the objective function value, convert the safety distance requirement and speed limit into mathematical constraints, and obtain the constraint equation set; Combine the constraint equation set with the objective function value to construct an optimization problem, calculate the optimal solution through gradient descent, and obtain the initial control parameters; Conduct a conflict check on the initial control parameters, analyze the degree of mutual influence between the parameters, and obtain the feasible control range; Refine and adjust the parameters within the feasible control range, select the optimal control scheme, and form the control strategy parameters.

7. The multi-data fusion method in airport apron safety supervision according to claim 6, characterized in that, Conduct a conflict check on the initial control parameters, analyze the degree of mutual influence between the parameters, and obtain the feasible control range, including: conduct a numerical range analysis on the initial control parameters, count the upper and lower limits of the values of each parameter, and obtain the parameter change interval; Cross-combine the parameter change intervals, calculate the control effects under different parameter combinations, and obtain the parameter correlation matrix; Conduct a sensitivity analysis on the parameter correlation matrix, calculate the influence intensity of parameter changes on the control effect, and obtain the parameter sensitivity value; Arrange the parameter sensitivity values in descending order, identify the key influencing parameters and secondary influencing parameters, and obtain the parameter importance sequence; Use the parameter importance sequence as the weight coefficient to perform weighted constraints on the parameter value range, and obtain the parameter adjustment boundary; Conduct a feasibility verification on the parameter adjustment boundary, eliminate the value intervals that do not meet the constraint conditions, and obtain the feasible control range.

8. A multi-data fusion system in airport apron safety supervision, which is used to implement the multi-data fusion method in airport apron safety supervision as described in any one of claims 1-7, characterized in that, The multi-data fusion system in airport apron safety supervision includes: A generation module, which is used to generate a data credibility value for the video data, radar data, and position data collected by the apron multi-source sensors according to data integrity and time consistency, filter and denoise the original data through the data credibility value, and obtain a preprocessed data set; A fusion module, which is used to establish a unified spatio-temporal benchmark for the preprocessed data set through timestamp and coordinate conversion, calculate the matching degree between different data sources based on the target feature similarity, and obtain the fused target data; An extraction module, which is used to extract feature parameters from three dimensions of position, speed, and motion trajectory according to the fused target data, screen and combine the feature parameters based on the correlation coefficient, and form a target feature matrix; A judgment module, which is used to judge potential conflicts based on the target feature matrix through the set safety distance threshold and speed threshold, combine historical risk event data to construct a risk assessment index, and obtain the safety risk score; A construction module, which is used to establish the spatial association relationship and risk propagation relationship between apron targets according to the safety risk score, calculate the regional risk density through the risk accumulation effect, and generate real-time safety situation data; A solution module, configured to calculate an objective function value of security control for real-time security situation data, set security constraint conditions for optimal solution, and form control strategy parameters.

9. A computer-readable storage medium having instructions stored thereon, characterized in that, When executed by a processor, the instruction implements the multi-data fusion method in airport apron security supervision according to any one of claims 1-7.

Images