A method and system for protecting data transmission security of a UAV

Abstract

The present disclosure provides a kind of unmanned aerial vehicle data transmission security protection method and system, it includes the original trajectory data of unmanned aerial vehicle and train is continuously sampled, constructs the data segmented space mapping structure based on relative motion trajectory;The displacement change relationship between adjacent segments is connected in series, executes the sequence construction of sliding trajectory constraint based on space displacement interval;The data segmented sequence of receiving end is matched with sliding trajectory constraint sequence and is aligned;The abnormal rearrangement candidate segment in the rearrangement feature calibration result set is respectively evaluated time continuity and space continuity, executes the abnormal behavior discrimination processing based on time and space coupling consistency;Abnormal data segment is eliminated, and based on the sequence reordering and splicing of the sliding trajectory constraint sequence according to the trusted data segment, execute the data transmission security output processing based on abnormal elimination and sequence reconstruction, this method realizes the reliable and safe protection of unmanned aerial vehicle data transmission in the scene of train accompanying inspection.

Description

Technical Field

[0001] This disclosure relates to the field of unmanned aerial vehicle (UAV) data transmission security technology, and in particular to a method and system for protecting the security of UAV data transmission. Background Technology

[0002] During train-accompanied inspections, the drone maintains relative motion with the high-speed train, and the node connections in the communication link continuously shift over time. Traditional technologies rely on a fixed sequence mechanism for temporal consistency judgment. However, the continuous spatial shifting and rearrangement of the sequence leads to temporal continuity but spatial shifts, making it impossible for anomaly detection mechanisms to distinguish between normal shifts and anomalous rearrangements. Specifically, when the drone accompanies the high-speed train, its relative position constantly changes, and the normal order of data segments itself slides continuously in space. Malicious attackers may exploit this dynamic characteristic to anomalously rearrange data segments while maintaining temporal continuity. Since traditional methods rely solely on temporal consistency for judgment, they struggle to differentiate between normal shifts under physical laws and artificially created anomalous rearrangements, resulting in vulnerabilities in data transmission security.

[0003] Therefore, there is an urgent need for a data transmission security protection method that can construct temporal and spatial coupling constraints to solve the problem that existing technologies cannot distinguish between normal sliding rearrangement and abnormal rearrangement behavior, and to achieve reliable security protection for UAV data transmission in train-accompanying inspection scenarios. Summary of the Invention

[0004] In view of this, in order to solve the problems caused by the existing technology, this application provides a method and system for protecting the security of UAV data transmission.

[0005] In a first aspect, this disclosure provides a method for ensuring the security of data transmission from unmanned aerial vehicles (UAVs), the method comprising:

[0006] S1: Continuously sample the raw trajectory data of the UAV and the train to obtain the relative motion trajectory, establish a one-to-one correspondence between data segments and spatial displacement intervals, and construct a data segment spatial mapping structure based on the relative motion trajectory;

[0007] S2: Connect the displacement change relationship between adjacent segments in the data segmentation spatial mapping structure, apply monotonicity and continuity constraints, and perform the construction of a sliding trajectory constraint sequence based on the spatial displacement interval to form a sliding trajectory constraint sequence that satisfies the laws of physical motion.

[0008] S3: Match and align the data segment sequence from the receiving end with the slip trajectory constraint sequence, calculate the matching score, perform sequence rearrangement feature calibration processing based on slip trajectory consistency, and output the rearrangement feature calibration result set;

[0009] S4: Perform temporal continuity evaluation and spatial continuity evaluation on the abnormal rearrangement candidate segments in the rearrangement feature labeling result set, construct a spatiotemporal coupling anomaly index, perform abnormal behavior discrimination processing based on the consistency of temporal and spatial coupling, and generate an abnormal behavior label set and a reliable data segment set.

[0010] S5: Based on the abnormal behavior marker set and the trusted data segment set, the abnormal data segments are eliminated, and the trusted data segments are reordered and spliced ​​according to the slip trajectory constraint sequence, and data transmission security output processing based on abnormal elimination and sequence reconstruction is performed.

[0011] Optionally, S1 includes:

[0012] Perform unified time-base resampling processing on the raw transmission data stream of the UAV, the UAV trajectory sampling sequence, and the train trajectory sampling sequence to obtain a unified time-base sampling sequence;

[0013] The unified time-base sampling sequence is segmented according to a fixed time window to form data segments with time identifiers and sequence numbers;

[0014] Based on the three-dimensional coordinates of the UAV and the train at the center value of each data segment time, calculate the relative displacement in the track direction, the lateral relative displacement, the height relative displacement, and the relative displacement state value.

[0015] Based on the relative displacement and relative displacement state value of each data segment in the orbital direction, the data segment is mapped to the corresponding spatial displacement interval, forming a one-to-one correspondence between time identifier, sequence number, relative displacement state and spatial displacement interval.

[0016] Optionally, S2 includes:

[0017] Extract the spatial displacement interval number, interval center position, relative displacement state value and time center value between adjacent data segments, and construct the interval association sequence of adjacent data segments;

[0018] The adjacent data segment interval association sequence is subjected to forward concatenation calculation to generate the original slip trajectory expression sequence containing continuous displacement accumulation and state change suppression;

[0019] The original slip trajectory expression sequence is subjected to monotonicity constraint processing to remove pseudo-slip segments with non-physical direction flips, resulting in a monotonicity constraint trajectory sequence.

[0020] Apply continuous constraints to the monotonic constraint trajectory sequence to limit abrupt changes in segmented slip velocity and sharp bends in relative displacement state values, and output the slip trajectory constraint sequence.

[0021] Optionally, S3 includes:

[0022] Using the segment sequence number in the slip trajectory constraint sequence as the retrieval key, the corresponding receiving segment record is extracted from the receiving buffer to establish an aligned observation sequence between the receiving sequence and the constraint trajectory.

[0023] For each observation pair in the aligned observation sequence, the spatial center offset, velocity difference, and difference in relative displacement state value are calculated and combined to obtain a single-segment matching score;

[0024] Based on the single-segment matching score, the spatial displacement change trends of the received sequence and the constraint trajectory within a continuous window are compared to calculate the continuous slip trend matching sequence;

[0025] Based on the calculated continuous slip trend aggregated score, combined with the comprehensive constraint factor and score fluctuation amplitude of the constraint trajectory nodes, the received data is segmented and marked as normal slip sequence or abnormal rearrangement candidate sequence, generating a set of rearrangement feature calibration results.

[0026] Optionally, S4 includes:

[0027] Extract abnormal rearrangement candidate segments from the rearrangement feature labeling result set, construct local spatiotemporal correlation units around the candidate segments and their preceding and following segments, and form an abnormal candidate spatiotemporal correlation sequence;

[0028] The temporal continuity evaluation value and spatial continuity evaluation value are calculated for the abnormal candidate spatiotemporal correlation sequence to obtain the temporal continuity evaluation sequence and the spatial continuity evaluation sequence.

[0029] The temporal continuity evaluation value and the spatial continuity evaluation value are coupled and consistent to determine the spatiotemporal coupling anomaly index. Based on the comparison between the corrected spatiotemporal coupling anomaly index and the anomaly determination threshold, anomalous rearrangement behavior that is temporally continuous but spatially abrupt is identified. The identified anomalous rearrangement behavior is assigned to the anomalous behavior label set, and the remaining segments, together with the normal slip sequence, are assigned to the reliable data segment set.

[0030] Optionally, S4 further includes:

[0031] For segments where the corrected spatiotemporal coupling anomaly index is higher than the anomaly discrimination threshold, they are marked as anomalous rearrangement behavior and written into the anomalous behavior label set;

[0032] For segments that do not exceed the threshold, along with segments that have been marked as normal slip sequences, they are written into the trusted data segment set.

[0033] Multiple consecutive abnormal segments are merged. If the time interval between adjacent abnormal segments does not exceed one segment time window and the spatial displacement interval numbers do not differ by a preset range, they are merged into the same abnormal behavior segment.

[0034] Optionally, S5 includes:

[0035] The abnormal behavior marker set and the trusted data segment set are jointly indexed and organized to generate a trusted retention sequence and an abnormal removal index sequence.

[0036] The combination of the center position of the spatial displacement interval, the constraint trajectory expression value, and the segmented time center value is used as the rearrangement basis to reorder the reliable preserved sequence according to the slip trajectory constraint sequence.

[0037] The adjacent trusted segments after reordering are spliced ​​and reconstructed to determine whether the time gap and spatial gap are within the allowable range, and a reconstructed data sequence is generated based on the determination result.

[0038] Based on the reconstructed data sequence, calculate the local reconstruction consistency coefficient and the global reconstruction consistency coefficient to generate the slip trajectory consistency state result;

[0039] The reconstructed data sequence and the anomaly removal index sequence are jointly encapsulated to generate a transmission security result that includes the anomaly rearrangement position result, the slip trajectory consistency status result, and the reconstructed data sequence body.

[0040] Secondly, this disclosure provides a UAV data transmission security protection system applied to the method of the first aspect, the system comprising:

[0041] The data segmentation spatial mapping construction module is used to continuously sample the raw trajectory data of UAVs and trains to obtain relative motion trajectories, establish a one-to-one correspondence between data segments and spatial displacement intervals, and construct a data segmentation spatial mapping structure based on relative motion trajectories.

[0042] The sliding trajectory constraint sequence construction module is used to connect the displacement change relationship between adjacent segments in the data segmentation spatial mapping structure, apply monotonicity and continuity constraints, and perform sliding trajectory constraint sequence construction based on spatial displacement interval to form a sliding trajectory constraint sequence that satisfies the laws of physical motion.

[0043] The sequence rearrangement feature calibration module is used to match and align the data segment sequence at the receiving end with the sliding trajectory constraint sequence, calculate the matching score, perform sequence rearrangement feature calibration processing based on sliding trajectory consistency, and output a set of rearrangement feature calibration results.

[0044] The abnormal behavior discrimination module is used to evaluate the temporal continuity and spatial continuity of the abnormal rearrangement candidate segments in the rearrangement feature labeling result set, construct a spatiotemporal coupling anomaly index, perform abnormal behavior discrimination processing based on the consistency of temporal and spatial coupling, and generate an abnormal behavior label set and a reliable data segment set.

[0045] The data transmission security output module is used to remove abnormal data segments based on the abnormal behavior marker set and the trusted data segment set, and to reorder and splice the trusted data segments according to the slip trajectory constraint sequence, and perform data transmission security output processing based on abnormal removal and sequence reconstruction.

[0046] Thirdly, this disclosure provides an electronic device including a memory and at least one processor, the memory storing a computer program, and the processor executing the computer program to implement the method of the first aspect described above.

[0047] Fourthly, this disclosure provides a computer storage medium storing a computer program that, when executed, implements the method described in the first aspect.

[0048] The beneficial effects of the present invention are as follows: Compared with the prior art, the present invention has the following advantages:

[0049] (1) By constructing a spatial mapping expression of data segments based on relative motion trajectories, a one-to-one correspondence is established between data segments and spatial displacement intervals. Furthermore, a sliding trajectory constraint sequence is constructed, and spatial displacement intervals are introduced as independent constraint dimensions. This solves the problem that traditional methods rely solely on time series consistency and cannot introduce spatial constraints. It realizes the quantitative expression of data segments in spatial displacement intervals and provides a spatial dimension for distinguishing between normal sliding and abnormal rearrangement.

[0050] (2) By performing abnormal behavior discrimination processing based on the consistency of time and space coupling, the data segments are evaluated for time continuity and space continuity respectively, and a time-space coupling abnormal index is constructed to identify abnormal rearrangement behavior that is continuous in time but jumps in space. This solves the hidden danger that malicious attackers use dynamic characteristics to perform abnormal rearrangement while maintaining the continuity of time order, which is difficult to detect. It realizes high-precision detection of abnormal behavior that is continuous in time but has a shift in spatial correspondence. Attached Figure Description

[0051] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0052] Figure 1A flowchart of a method for protecting the security of unmanned aerial vehicle data transmission provided in an embodiment of this disclosure is shown;

[0053] Figure 2 A flowchart illustrating the abnormal rearrangement behavior discrimination based on spatiotemporal coupling consistency provided in an embodiment of this disclosure is shown.

[0054] Figure 3 A schematic diagram of the structure of the UAV data transmission security protection system provided in this embodiment is shown.

[0055] The accompanying drawings have illustrated specific embodiments of this disclosure, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this disclosure to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0056] The present disclosure will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present disclosure and should not be used to limit the scope of protection of the present disclosure.

[0057] The components of the embodiments of the invention described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0058] In the following, the terms “comprising,” “having,” and their cognates, which may be used in various embodiments of the invention, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as excluding, firstly, the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more features, numbers, steps, operations, elements, components, or combinations thereof.

[0059] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the invention pertain. Terms (such as those defined in commonly used dictionaries) shall be interpreted as having the same meaning as in their contextual meaning in the relevant technical field and shall not be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of the invention.

[0060] Figure 1A flowchart of the UAV data transmission security protection method provided in this disclosure embodiment is shown below. Figure 1 As shown, the process may include the following steps:

[0061] S1: Continuously sample the raw trajectory data of the UAV and the train to obtain the relative motion trajectory, establish a one-to-one correspondence between data segments and spatial displacement intervals, and construct a data segment spatial mapping structure based on the relative motion trajectory.

[0062] The relative motion trajectories of the UAV and the train are continuously sampled. The data stream is divided into continuous data segments according to fixed time windows, and each segment is assigned a time stamp, sequence number, and relative displacement status. Each data segment is then mapped to a corresponding spatial displacement interval based on changes in relative spatial position, forming a spatial mapping structure where the time series and spatial displacement intervals correspond one-to-one. This is achieved through the following sub-steps.

[0063] S1.1: Perform unified time-base resampling processing on the raw transmission data stream of the UAV, the UAV trajectory sampling sequence and the train trajectory sampling sequence to obtain a unified time-base sampling sequence.

[0064] To eliminate the asynchronous deviation between the arrival time of the original data stream and the time of relative motion, it is first necessary to perform unified time-base resampling processing on the original UAV transmission data stream, the UAV trajectory sampling sequence, and the train trajectory sampling sequence. This processing compresses the data stream carrying time and the time of relative motion occurrence into the same sampling clock, so that subsequent data segmentation no longer depends on the arrival order of the original link, but on a unified motion reference time.

[0065] In practice, a combination of fixed-cycle resampling and exponential smoothing is used. In this embodiment, a uniform sampling period of 0.05 seconds is used; if the train speed is high, a smaller sampling period, such as 0.02 seconds, can be selected. For the UAV position coordinates and the train position coordinates, exponential weighted smoothing is performed respectively to suppress the amplified effect of short-term jitter during flight on the relative displacement calculation.

[0066] Taking the orbital direction coordinates as an example, the smoothing process is performed according to the following formula:

[0067]

[0068]

[0069] in, Let be the coordinates of the UAV's orbital direction at the k-th original sampling time. Let be the train track orientation coordinates at the k-th original sampling time. , These are the smoothed orbital direction coordinates; To smooth out the attenuation coefficient, the value ranges from 0.15 to 0.45, and in this embodiment, it is 0.30; The smoothing window length ranges from 3 to 9; in this embodiment, it is set to 5. This smoothing formula assigns greater weight to samples taken at closer time points and less weight to samples taken at farther time points, thereby obtaining a continuous motion reference trajectory that can be used for subsequent spatial mapping. The lateral and altitude coordinates of the UAV and the train are processed in the exact same way.

[0070] In practice, a reference time sequence is first generated according to a unified sampling period. Then, the original trajectory sampling points are projected onto the corresponding reference times. Finally, smoothing and data encapsulation are completed. After the above processing, a unified time-base sampling sequence containing the unified time, UAV position, train position, and the data stream segment carried at that time can be obtained. This sequence will serve as the input basis for the next step.

[0071] S1.2: Divide the unified time base sampling sequence into segments according to a fixed time window to form data segments with time identifiers and sequence numbers.

[0072] After obtaining the unified time-base sampling sequence, the sequence needs to be segmented according to a fixed time window to form multiple continuous data segments. Each data segment will correspond to a time identifier and a sequence number. The selection of the time window needs to consider two points: first, it must cover multiple sampling points to ensure that the segment contains effective relative motion information; second, it cannot be too long to avoid compressing the actual sliding process. In this embodiment, the segmented time window... Take 0.60 seconds, with a value range of 0.20 seconds to 1.50 seconds.

[0073] For the j-th sampling time in the unified time-base sampling sequence We need to determine which data segment it belongs to. Let the starting reference time be... Then the data segment sequence number corresponding to that sampling time. Calculated by the following formula:

[0074]

[0075] in, This indicates rounding down. The formula divides the continuous time axis into a series of time slots of fixed length. Each sampling moment is assigned to its corresponding time slot by dividing its time difference from the start moment by the window length and then rounding down.

[0076] The start time of this data segment and end time They are respectively:

[0077]

[0078]

[0079] In practice, all data stream segments within the same time window are merged into a single data segment, and the start and end times and number of this segment are recorded to obtain a time-identified data segment sequence. This sequence contains the time boundary information of each data segment, providing a time reference for subsequent calculations of relative displacement states.

[0080] S1.3: Based on the three-dimensional coordinates of the UAV and the train at the center value of each data segment time, calculate the relative displacement in the track direction, the lateral relative displacement, the height relative displacement, and the relative displacement state value.

[0081] For each data segment, its relative displacement state needs to be calculated. This state cannot be represented solely by instantaneous distance, otherwise it would be impossible to distinguish between normal slippage and abnormal rearrangement; it must simultaneously reflect the relative offset in track direction, lateral offset, height offset, and the rate of change between adjacent segments. Therefore, the relative position quantities at the segment's time center value are first obtained, and then the segment-level relative displacement state values ​​are constructed.

[0082] For the j-th data segment, first determine its time center value, and then read the corresponding UAV three-dimensional coordinates from the unified time-base sampling sequence. and train three-dimensional coordinates Then calculate the relative displacement along the orbital direction. Lateral relative displacement and height relative displacement :

[0083]

[0084]

[0085]

[0086] Furthermore, the relative spatial distance of the j-th data segment for:

[0087]

[0088] To couple the three-dimensional relative position and the rate of change of distance between adjacent segments to the same state variable, a relative displacement state value is constructed. :

[0089]

[0090] in, , , These are the normalized scales for the track direction, lateral direction, and height direction, respectively, with values ​​ranging from 10 to 200, 2 to 30, and 1 to 20. In this embodiment, they are taken as 100 meters, 15 meters, and 8 meters, respectively. The distance change normalized velocity scale has a value range of 5 to 120, and in this embodiment it is taken as 30 meters per second; The segmented time window set in step S1.2. This formula embeds both spatial location and spatial slip trend into the data segments, providing a unified comparison benchmark for establishing a one-to-one correspondence between time series and spatial displacement intervals.

[0091] In practice, the three-dimensional positions of the drone and the train are first read from the segmented time center value, and then the segmented state values ​​are calculated. If the train inspection speed is higher than 60 meters per second, it is recommended to increase it appropriately. and The values ​​are selected to avoid excessive amplification of normal accompanying slip. After the above calculations, a data segment sequence with relative displacement states is obtained.

[0092] S1.4: Based on the relative displacement and relative displacement state value of each data segment in the orbital direction, the data segment is mapped to the corresponding spatial displacement interval, forming a one-to-one correspondence between time identifier, sequence number, relative displacement state and spatial displacement interval.

[0093] Based on the relative displacement and relative displacement state of each data segment along the track direction, the data segment is mapped to the corresponding spatial displacement interval, forming a one-to-one correspondence between time stamp, sequence number, relative displacement state, and spatial displacement interval. The spatial displacement interval is primarily established along the train track direction because, in this scenario, continuous sliding rearrangement first manifests as a spatial offset along the track direction. The interval length should not be too large, otherwise multiple segments with different physical locations will be mapped to the same interval; the interval length should also not be too small, otherwise it is easily affected by instantaneous jitter. In this embodiment, the spatial displacement interval length... Take 8 meters.

[0094] Determine the minimum relative displacement of the track direction for this inspection mission. This value can be obtained by iterating through all segments. We obtain the spatial displacement interval number of the j-th data segment. for.

[0095]

[0096] The lower boundary of the spatial displacement interval corresponding to this data segment. and upper boundary They are respectively:

[0097]

[0098]

[0099] Finally, each data segment is encapsulated into a mapping record item, which includes at least a time identifier consisting of the segment start time and segment end time, a segment number, relative displacement in the track direction, relative displacement state value, spatial displacement interval number, and interval boundary. This establishes a one-to-one correspondence between the time identifier, segment number, relative displacement state value, and spatial displacement interval. The resulting data segment spatial mapping structure can be directly used as input data for the sliding trajectory constraint sequence construction process in step S2.

[0100] In the technical solution of this disclosure, by constructing a spatial mapping representation of data segments based on relative motion trajectories, a one-to-one correspondence is established between the temporally continuous data stream and the spatial displacement interval. This ensures that data segments no longer rely solely on temporal order but also simultaneously carry spatial location information. This step eliminates the dependence of the original link arrival order on the data segment organization, providing a unified data foundation in both time and space for subsequently distinguishing between normal slippage and abnormal rearrangement. It fundamentally solves the problem in the prior art where temporally continuous data but spatial correspondence offsets cannot be effectively represented.

[0101] S2: Connect the displacement change relationships between adjacent segments in the data segmentation spatial mapping structure, apply monotonicity and continuity constraints, and execute the construction of a sliding trajectory constraint sequence based on the spatial displacement interval to form a sliding trajectory constraint sequence that satisfies the laws of physical motion.

[0102] The spatial displacement interval correlation analysis is performed on continuous data segments in the data segmentation spatial mapping structure. The displacement change relationship between adjacent segments is concatenated into a continuous sliding trajectory expression sequence, and monotonicity and continuity constraints are applied to form a sliding trajectory constraint sequence that satisfies the laws of physical motion. The output is an information expression structure with trajectory constraints. This is achieved through the following sub-steps.

[0103] S2.1: Extract the spatial displacement interval number, interval center position, relative displacement state value and time center value between adjacent data segments, and construct the interval association sequence of adjacent data segments.

[0104] The data segment spatial mapping structure formed in step S1 is sorted in ascending order of segment sequence number. After sorting, the spatial displacement interval number, interval center position, relative displacement state value, and time center value between two adjacent data segments are extracted sequentially to construct the interval correlation relationship. The core here is not to directly compare whether two segments are adjacent, but to uniformly express which spatial displacement interval the previous segment is in, how much the subsequent segment has moved relative to the previous segment along the track direction, and whether this movement still satisfies the physical slip characteristics accompanying the inspection.

[0105] Therefore, we first calculate the center position of the spatial displacement interval corresponding to each data segment. For the j-th data segment, the lower boundary of its spatial displacement interval is... The upper boundary is Then the center position of the interval for:

[0106]

[0107] Calculate the change in center displacement between adjacent segments That is, the difference between the center positions of the j-th segment and the (j-1)-th segment:

[0108]

[0109] Simultaneously, the time interval between the time center values ​​of two adjacent data segments is calculated. Let the time center value of the j-th data segment be... ,but:

[0110]

[0111] Calculate the spatial displacement interval jump of adjacent segments That is, the difference between the interval numbers:

[0112]

[0113] in, Number the spatial displacement interval of the j-th data segment. Since there is no preceding segment, the change in center displacement is defined. Time interval Interval jump variables .

[0114] The above formula transforms the discrete interval mapping structure into a relationship of change in the spatial center position, providing an accumulative displacement basis for the subsequent serial construction of the slip trajectory.

[0115] In practical implementation, if the time interval It exceeded the segmented time window set in step S1. If the distance is twice the length of the adjacent data segment, the adjacent relationship is temporarily marked as a candidate for breakage. This prevents pseudo-continuous concatenation caused by link caching or sudden rearrangement, ensuring the reliability of subsequent trajectory construction. After the above processing, an adjacent data segment interval association sequence is obtained, which is directly used to generate the original slip trajectory expression sequence in the next step.

[0116] S2.2: Perform forward concatenation calculation on the adjacent data segment interval correlation sequence to generate the original slip trajectory expression sequence containing continuous displacement accumulation and state change suppression.

[0117] After obtaining the correlation sequence of adjacent data segments, a forward concatenation calculation is performed to generate the original slip trajectory expression sequence. This concatenation incorporates the current displacement change, the relative displacement state difference between adjacent segments, and the time interval to construct a trajectory expression value that reflects the continuous slip process. In train-accompanied inspection scenarios, normal slip is often characterized by slow advancement or slow retraction along the track direction, while abnormal rearrangement typically manifests as excessively large displacement jumps or abrupt state changes within a short period. Therefore, the following recursive calculation formula is constructed. This formula achieves continuous displacement accumulation by summing the center displacement change, and simultaneously suppresses abrupt state changes by introducing a penalty term for the relative displacement state difference between adjacent segments.

[0118] Let the original slip trajectory representation of the j-th data segment be value. Its recursive calculation formula is as follows:

[0119]

[0120] in, The original slip trajectory representation value (in meters) for the (j-1)th data segment; This represents the change in center displacement between the j-th data segment and the (j-1)-th data segment; The relative displacement state value for the j-th data segment; The state difference suppression coefficient ranges from 0.20 to 2.00; in this embodiment, it is set to 1.00. In this recursive formula, the summation... It achieves continuous accumulation of displacement, and the denominator contains... This achieves suppression of state abrupt changes, and the combined effect of both factors makes the calculated original slip trajectory expression value... This formula reflects normal continuous slip without misinterpreting abrupt changes in state as normal displacement. While accumulating displacement continuously, it suppresses displacement changes with large differences in state between adjacent segments, ensuring that abrupt changes in state are not directly interpreted as normal slip.

[0121] Simultaneously, the series displacement relationship is converted into a velocity expression, facilitating the subsequent application of continuity constraints. The segmented sliding velocity corresponding to the j-th data segment. for:

[0122]

[0123] in, The time interval between the center values ​​of adjacent data segments.

[0124] For the initial segment ,because and ,definition Recursive calculation When required Pick .

[0125] In practice, the initial value Desirable This refers to the center position of the spatial displacement interval of the first data segment; if the system emphasizes relative change, it can also be set to 0. In forward concatenation, each time a new... This, along with the segment number, time center value, and segment-level slip velocity, is encapsulated into a single original slip trajectory node. All trajectory nodes are arranged in chronological order to form the original slip trajectory representation sequence, and this output will be directly used for the next step of monotonicity constraint processing.

[0126] S2.3: Apply monotonicity constraints to the original slip trajectory expression sequence to remove pseudo-slip segments with non-physical direction flips, and obtain a monotonicity constrained trajectory sequence.

[0127] A monotonicity constraint is applied to the original slip trajectory representation sequence. This monotonicity does not require the trajectory to only increase or decrease in a simple manner, but rather that the slip direction should conform to the physical laws governing train-accompanied inspection within a short, continuous time window. Typically, the slip of the UAV relative to the train track direction either evolves slowly in the same direction or fluctuates slightly within a small range; high-frequency directional reversals should not occur in consecutive adjacent segments. Therefore, a monotonicity consistency factor needs to be constructed to assess whether there is a non-physical reversal between the current segment's displacement direction and the previous segment's displacement direction.

[0128] For the j-th data segment, the monotonic consistency factor Calculated by the following formula:

[0129]

[0130] in, and These represent the changes in displacement between two adjacent pairs; The displacement change normalization scale ranges from 1 to 30; in this embodiment, 10 meters is used. The numerator of this formula is essentially equal to the deviation that occurs when two adjacent displacement changes have different directions: if they have the same sign, the deviation approaches zero, and the monotonic consistency factor approaches 1; if they have opposite signs and large amplitudes, the deviation increases, and the monotonic consistency factor decreases significantly. For Directly set .for ,calculate When required As defined in step S2.1, , Set to 0. This formula transforms the consistency of adjacent displacement directions into a continuously calculable penalty term.

[0131] In practice, a sliding window with a length of 3 to 6 data segments is used for monotonicity verification. In this embodiment, a length of 4 is used. Calculations are performed on each segment within the window. If multiple consecutive If all values ​​are below a set threshold (0.60 in this embodiment), it indicates that the trajectory has undergone non-physical direction reversal, and the corresponding original trajectory node should be marked as a monotonic abnormal node. The higher the train speed and the more stable the accompanying flight, the higher the threshold should be.

[0132] Finally, trajectory nodes not marked as monotonic anomalies are retained, resulting in a monotonicity-constrained trajectory sequence. This sequence eliminates pseudo-slip segments that are temporally continuous but exhibit high-frequency spatial orientation flips, preventing anomalous rearrangements from being incorrectly concatenated as normal slip trajectories. This output will be directly used in the next step of continuity constraint processing.

[0133] S2.4: Apply continuous constraints to the monotonic constraint trajectory sequence to limit abrupt changes in segmented slip velocity and sharp bends in relative displacement state values, and output an information expression structure with trajectory constraints.

[0134] Building upon monotonicity constraints, continuity constraints are further applied to the monotonicity-constrained trajectory sequence to obtain a sliding trajectory constraint sequence that satisfies the laws of physical motion. This sequence is then further encapsulated into an information representation structure with trajectory constraints. The focus of continuity constraints is to limit the abrupt changes in segmented sliding velocity and to limit the drastic bending of relative displacement state values ​​within a short period of time.

[0135] Constructing continuity factors The calculation formula is as follows:

[0136]

[0137] in, and These represent the segment-level sliding velocities of two adjacent data segments; To normalize the velocity variation, this embodiment uses 15 meters per second, with a value range of 2 to 40. , , These represent the relative displacement state values ​​of three consecutive data segments; The normalized scale for state bending is set to 1.00 in this embodiment, with a value ranging from 0.10 to 2.50. For Directly set , .for ,calculate When required Pick , Pick This formula simultaneously constrains both sudden velocity jumps and state bends, ensuring that the sliding trajectory is not only directionally reasonable but also changes smoothly.

[0138] Couple the monotonic consistency factor and the continuity factor into a unified comprehensive constraint factor. :

[0139]

[0140] in, The monotonicity and continuity weighting coefficient is set to 0.50 in this embodiment, with a range of 0.30 to 0.70. This formula couples monotonicity and continuity into a unified comprehensive discriminant.

[0141] In practical implementation, if If the value is not lower than the set threshold (in this embodiment, the threshold is 0.70, and the value range is 0.50 to 0.85), then the trajectory node is retained and marked as a constrained valid node; if it is lower than the threshold, then it is marked as a constrained mismatch node.

[0142] Finally, all constrained valid nodes are repackaged in chronological order to form a sliding trajectory constraint sequence, i.e., an information representation structure with trajectory constraints. This structure includes at least the following: segment sequence number, time center value, spatial displacement interval number, interval center position, original sliding trajectory expression value, segment-level sliding velocity, monotonic consistency factor, continuity factor, and comprehensive constraint factor. This output will be directly used as input data for the sequence rearrangement feature calibration processing based on sliding trajectory consistency in step S3.

[0143] In the technical solution of this disclosure, by performing a sliding trajectory constraint sequence construction process based on spatial displacement intervals, the displacement change relationship between adjacent data segments is concatenated, and monotonicity and continuity constraints are applied, thereby forming a sliding trajectory constraint sequence that reflects the laws of physical motion. This step quantifies the normal sliding law during the flight of the UAV accompanying the train into a computable constraint expression, so that any abnormal rearrangement behavior that deviates from the physical sliding law has a comparable benchmark in subsequent discrimination, effectively suppressing the fuzzy boundary between normal sliding and abnormal rearrangement.

[0144] S3: Match and align the segmented data sequence from the receiving end with the slip trajectory constraint sequence, calculate the matching score, perform sequence rearrangement feature calibration processing based on slip trajectory consistency, and output the rearrangement feature calibration result set.

[0145] The data segment sequence at the receiving end is subjected to rearrangement feature analysis. The changing trend of the received sequence in the spatial displacement interval is compared with the matching relationship of the sliding trajectory constraint sequence. Segments that meet the constraints are marked as normal sliding sequences, and those that do not meet the constraints are marked as abnormal rearrangement candidate sequences. The rearrangement feature labeling result set is output. Figure 2 The flowchart illustrating the abnormal rearrangement behavior discrimination based on spatiotemporal coupling consistency provided in this disclosure embodiment is shown, as follows: Figure 2 As shown, this is achieved through the following sub-steps.

[0146] S3.1: Using the segment sequence number in the slip trajectory constraint sequence as the retrieval key, extract the corresponding receiving segment record from the receiving buffer and establish an aligned observation sequence between the receiving sequence and the constraint trajectory.

[0147] The sliding trajectory constraint sequence output in step S2 is processed by the receiving end. Using the segment sequence number, time center value, spatial displacement interval number, interval center position, and comprehensive constraint factor already encapsulated in step S2 as search keys, the corresponding numbered data segment reception records are extracted from the receiving buffer, forming observation pairs where each constraint trajectory node corresponds one-to-one with the receiving segment node. For each observation pair, the interval center position, segment-level sliding velocity, and relative displacement state value from the transmitting constraint side are uniformly extracted, along with the interval center position, receiving time center value, and receiving state value reconstructed from the receiving side. Thus, subsequent comparisons are no longer based on the original disordered segments, but rather on the received sequence observation results bound with the same number.

[0148] In practice, the sliding trajectory constraint sequence is first arranged in ascending order according to the segment sequence number, and then the corresponding receiving segment is retrieved from the receiving buffer according to the same number. If there are multiple receiving candidates for a certain number, all candidates are retained to form a multi-candidate observation pair, which will be resolved in the subsequent matching and scoring stage.

[0149] The receiving end first determines the relative displacement of the track direction in the received data segments. Combined with the known spatial displacement interval length Relative displacement with respect to the minimum orbital direction Calculate the interval number of this segment. Thus, the lower boundary is obtained. and upper boundary Then the center position of the receiving side interval. for:

[0150]

[0151] The receiving end calculates the relative displacement state value from the received data segments using the same method as the sending end. Segmented sliding speed Spatial center increment and speed increment .

[0152] Simultaneously, the time offset between the center value of the received data segment time and the center time of the constraint trajectory node is calculated. Let the center value of the j-th received data segment be... The time center value of the j-th constrained trajectory node in the output of step S2 is ,but:

[0153]

[0154] for ,definition This indicates that the start time of the first received segment is aligned with the start time of the constrained trajectory node.

[0155] This set of formulas transforms the received sequence into a space-time observation form isomorphic to the constrained trajectory, providing a unified benchmark for subsequent same-scale comparisons. It should be noted that if there is a short-term delay in the receive buffer, then... There might be an overall offset within a local continuous window. This should not be immediately considered an anomaly, but rather a unified determination should be made in the subsequent trend matching stage. After the above processing, an observation sequence aligned with the constraint trajectory is obtained. Each element in this sequence is called an observation pair, and each observation pair contains one constraint trajectory node and one receiving segment node. This sequence will be directly used for the next step of single-segment matching score calculation.

[0156] S3.2: For each observation pair, calculate the difference in spatial center offset, velocity difference, and relative displacement state value, and combine them to obtain a single-segment matching score.

[0157] For each observation pair, a single-segment matching score is calculated, focusing on the degree of fit between the receiving segment and the constraint trajectory nodes in terms of spatial displacement interval position, slip velocity change, and relative displacement state. It's not sufficient to simply compare interval numbers, as normal slip allows for gradual interval migration in continuous slip scenarios; what truly needs to be identified is whether the migration is smooth and conforms to the evolution of the constraint trajectory. Therefore, three sub-items—spatial offset consistency, velocity consistency, and state consistency—are constructed simultaneously and combined to form the single-segment matching score.

[0158] Calculate the spatial center offset This is the difference between the center position of the receiving side interval and the center position of the constraint side interval. Let the center position of the spatial displacement interval of the constraint trajectory node be... ,but:

[0159]

[0160] Calculation speed difference This is the difference between the segment-level sliding velocity on the receiving side and the segment-level sliding velocity on the constraint side. Let the segment-level sliding velocity on the receiving side be... The segmented sliding velocity on the constrained side is ,but:

[0161]

[0162] Calculate the difference in relative displacement state values Let the relative displacement state value of the receiving side be... The relative displacement state value on the constraint side is ,but:

[0163]

[0164] Combining the three biases mentioned above yields the single-segment matching score for the j-th observation pair. :

[0165]

[0166] in, The spatial offset normalization scale ranges from 1 to 20, and is set to 8 meters in this embodiment. The velocity offset normalization scale ranges from 1 to 30, and is set to 15 meters per second in this embodiment. The state offset normalization scale ranges from 0.10 to 2.00, and is set to 1.00 in this embodiment. This formula uses an exponential term to strengthen the consistency constraint of the spatial center position and a fractional penalty term to suppress the deviation of velocity and state, so that only receiving segments that simultaneously satisfy the requirements of close position, similar velocity, and compatible state can obtain a high matching score.

[0167] In practice, if a certain receiving number corresponds to multiple candidate receiving segments, then each segment is calculated separately. The candidate segment corresponding to the maximum value is selected as the preferred observation segment for the current number, and the remaining candidates are retained as alternative anomaly candidates. After the above processing, a single-segment matching score sequence is obtained, which will be directly used for the next step of continuous sliding trend matching calculation.

[0168] S3.3: Based on the single-segment matching score, compare the spatial displacement change trends of the received sequence and the constraint trajectory within a continuous window, and calculate the continuous slip trend matching sequence.

[0169] Building upon the single-segment matching score, we further compare the degree of agreement between the spatial displacement trend of the received sequence within a continuous window and the trend of the constraint trajectory. This is because anomalous rearrangements do not necessarily manifest as a severe mismatch in a single segment; more commonly, a single point may appear reasonable, but when three or more consecutive segments are strung together, the direction, magnitude, or rate of change of spatial displacement does not align with the physical slip pattern. Therefore, it is necessary to construct a trend difference measure and a window aggregation score.

[0170] Calculate the adjacent displacement increments and adjacent velocity increments on both the receiving and constrained sides. Also calculate the spatial center increment on the receiving side. for:

[0171]

[0172] Spatial center increment on the constrained side for:

[0173]

[0174] Speed ​​increment on the receiving side for:

[0175]

[0176] velocity increment on the constraint side for:

[0177]

[0178] for ,definition , Accordingly, , .

[0179] The aforementioned incremental differences and time offset changes are combined to construct a trend matching function. The trend matching score for the j-th segment is then calculated. for:

[0180]

[0181] in, The spatial increment normalization scale ranges from 0.5 to 1.5, and is 1.5 meters in this embodiment; The speed increment normalization scale ranges from 0.5 to 20, and is set to 8 meters per second in this embodiment; The time offset normalization scale ranges from 0.01 to 0.50; in this embodiment, it is set to 0.10 seconds. This formula requires not only that the received segment itself is close to the constraint node, but also that the manner of change is consistent. If a received sequence exhibits local packet switching, skipping, or insertion rearrangement, then... , Changes in time offset often produce anomalous abrupt changes within a continuous window, leading to Significant decrease.

[0182] To avoid misjudgments caused by single-point fluctuations, it is also necessary to perform decay aggregation on the single-segment matching scores and trend matching scores within the continuous window. The continuous sliding trend aggregation score is calculated with the j-th segment as the end of the window. for:

[0183]

[0184] Where Q represents the length of the sliding window, which ranges from 3 to 8, and is set to 5 in this embodiment; This is the time decay coefficient, with a value ranging from 0.60 to 0.95; in this embodiment, it is set to 0.85. This represents the single-segment matching score for the i-th segment; This represents the trend matching score of the i-th segment. The formula calculates a weighted average of the matching results of the most recent segments and increases the weight of segments closer to the current time, thus making the determination of continuous slipping trends both locally robust and sufficiently sensitive to the latest anomalies.

[0185] In practice, it is recommended to use a larger Q and a higher Q during the stable train speed phase. In areas with significant relative motion fluctuations, such as curves and station entry / exit points, Q is appropriately reduced to prevent excessive accumulation of normal motion disturbances. After the above processing, a continuous slip trend matching sequence is obtained, which will be directly used for the next step of rearrangement feature calibration.

[0186] S3.4: Based on the calculated continuous slip trend aggregated score value, combined with the comprehensive constraint factor and score fluctuation amplitude of the constraint trajectory node, the received data is segmented and marked as normal slip sequence or abnormal rearrangement candidate sequence, generating a set of rearrangement feature calibration results.

[0187] Based on the calculated continuous slip trend aggregate score, the segmented data sequences at the receiving end are labeled and classified to form normal slip sequences and abnormal rearrangement candidate sequences. This classification should simultaneously consider the current window aggregate score, the decrease in scores of adjacent windows, and the comprehensive constraint factor of the constraint trajectory nodes. If the constraint trajectory itself is in a weak constraint range, such as during train deceleration or normal fluctuations caused by drone active obstacle avoidance, then even... A slight decrease is not necessarily an anomaly; however, during the high-constraint stable phase, a smaller decrease... A decrease may correspond to a significant anomaly.

[0188] Therefore, the final calibration criterion value is constructed. (Dimensionless quantity):

[0189]

[0190] in, This represents the aggregated score of the continuous sliding trend corresponding to the j-th segment; This represents the comprehensive constraint factor of the j-th constrained trajectory node in the output of step S2; The value is used as the normalization scale for score fluctuation, ranging from 0.05 to 0.50, and is set to 0.20 in this embodiment. Based on the continuous trend matching score, this formula further introduces constraint strength weighting and penalizes drastic fluctuations in scores of adjacent windows to prevent mislabeling caused by accidental high or low scores.

[0191] In practice, two judgment thresholds can be set: a higher threshold is used to mark normal sliding sequences, and a lower threshold is used to mark abnormal rearrangement candidate sequences. The intermediate interval can be temporarily stored as a region to be reviewed. In this embodiment, the normal sliding threshold is set to 0.70, and the abnormal rearrangement candidate threshold is set to 0.40. A normal sliding sequence indicates that the receiving segment satisfies the sliding trajectory constraint within a continuous window and can be interpreted as a physical sliding rearrangement under accompanying inspection; an abnormal rearrangement candidate sequence indicates that although the receiving segment maintains temporal continuity or local numbering continuity, its spatial displacement interval change trend does not match the constraint trajectory output in step S2, and there is a possibility of abnormal rearrangement.

[0192] Finally, the segment sequence number, time center value, spatial displacement interval number, single segment matching score, trend matching score, continuous slip trend aggregation score, final calibration criterion value, and category label of each data segment are uniformly encapsulated to form a rearranged feature calibration result set. This result set will be directly used as the input data for the abnormal behavior discrimination processing based on the consistency of temporal and spatial coupling in step S4.

[0193] In the technical solution of this disclosure, by performing sequence rearrangement feature calibration processing based on slip trajectory consistency, the data segment sequence at the receiving end is matched and aligned with the slip trajectory constraint sequence, and a single segment matching score and a continuous slip trend matching score are calculated, thereby identifying normal sequences that meet the slip constraints and abnormal candidate sequences that deviate from the constraints. This step achieves refined feature calibration of the received sequence, which can distinguish between normal rearrangement caused by physical slip and artificially created abnormal rearrangement, avoiding misjudgment caused by sequence space sliding in traditional methods.

[0194] S4: Perform temporal continuity evaluation and spatial continuity evaluation on the abnormal rearrangement candidate segments in the rearrangement feature labeling result set, construct a spatiotemporal coupling anomaly index, perform abnormal behavior discrimination processing based on the consistency of temporal and spatial coupling, and generate an abnormal behavior label set and a reliable data segment set.

[0195] Further discrimination is performed on candidate segments of abnormal rearrangement. The coupling relationship between time series continuity and spatial displacement continuity is jointly analyzed to determine whether there are abnormal behaviors that are continuous in time but abrupt in space. An abnormal behavior label set and a reliable data segment set are generated. Figure 2 The flowchart illustrating the abnormal rearrangement behavior discrimination based on spatiotemporal coupling consistency provided in this disclosure embodiment is shown, as follows: Figure 2 As shown, this is achieved through the following sub-steps.

[0196] S4.1: Extract abnormal rearrangement candidate segments from the rearrangement feature labeling result set, construct local spatiotemporal correlation units around the candidate segments and their preceding and following segments, and form an abnormal candidate spatiotemporal correlation sequence.

[0197] Extract data segments marked as anomalous rearrangement candidate sequences from the rearrangement feature calibration result set output in step S3, and reorder them according to the segment sequence number. The focus here is not on performing a normal reordering, but on constructing local spatiotemporal correlation units around the preceding, current, and following segments of the candidate segment. This is to determine whether the candidate segment belongs to a short-term disturbance in a normal slip process or to an anomalous rearrangement where time remains continuous but spatial position has deviated from the physical slip chain. For each candidate segment, its time center value, spatial displacement interval number, spatial displacement interval center position, final calibration criterion value, and corresponding parameters of adjacent preceding and following segments need to be extracted simultaneously to form a local three-segment spatiotemporal correlation record. If the candidate segment is at the beginning or end, an asymmetric correlation unit is constructed using the two most recent valid segments.

[0198] To ensure consistent judgment later, we first define the time interval, spatial central difference, and interval jump variable between candidate segments and adjacent valid segments. For the j-th candidate segment, let its reception time central value be... The center position of the spatial displacement interval is The spatial displacement interval is numbered as The time interval between the candidate segment and the previous segment. for:

[0199]

[0200] Time interval between candidate segments and the next segment for:

[0201]

[0202] Spatial central difference between candidate segment and previous segment for:

[0203]

[0204] Spatial central difference between candidate segment and subsequent segment for:

[0205]

[0206] Interval jump variables between candidate segments and the previous segment for:

[0207]

[0208] Interval jump variables between candidate segments and the next segment for:

[0209]

[0210] The above formula embeds the abnormal candidate segments into the local continuous chain, no longer viewing individual segments in isolation, but establishing basic observations for subsequent analysis of whether time is continuous and whether space is broken.

[0211] In specific implementation, if and All fall within the segmented time window in step S1 If the difference is between 0.5 and 1.5 times, the candidate segment is given priority in subsequent coupling discrimination; if the two deviate significantly, it is first recorded as a time discontinuity candidate, rather than directly judged as an anomalous rearrangement. After the above processing, an anomalous candidate spatiotemporal correlation sequence is obtained, which will be directly used for the next step of temporal and spatial continuity evaluation.

[0212] S4.2: Calculate the temporal continuity evaluation value and spatial continuity evaluation value for the abnormal candidate spatiotemporal correlation sequence, respectively, to obtain the temporal continuity evaluation sequence and spatial continuity evaluation sequence.

[0213] Temporal continuity evaluation and spatial continuity evaluation are performed on the candidate spatiotemporal correlation sequences of the anomalous structures. These two evaluations must be calculated separately because this step aims to identify special anomalies that are temporally continuous but spatially abrupt. Therefore, only after obtaining two independent evaluation metrics can subsequent coupled comparisons be performed.

[0214] The goal of time continuity evaluation is to determine whether a candidate segment is still within a normal transmission or reception cycle. If the time interval before and after a candidate segment is close to the segment's time window, it indicates that the candidate segment has temporal continuity. Time continuity evaluation value. Calculated by the following formula:

[0215]

[0216] in, The segmented time window determined in step S1; The time offset normalization scale ranges from 0.02 to 0.50, and is set to 0.10 seconds in this embodiment; The normalized scale for the difference in time intervals between segments ranges from 0.01 to 0.30; in this embodiment, it is set to 0.05 seconds. The core essence of this formula is: if the time intervals before and after a candidate segment are both close to the segment time window, and the differences between them are not significant, then... A value close to 1 indicates that the candidate segment is continuous in time.

[0217] The goal of spatial continuity evaluation is to determine whether the changes in a candidate segment over its spatial displacement interval still conform to the physical laws of slip. If the spatial center difference before and after a candidate segment is too large, or if the interval numbering changes too abruptly, it indicates that spatial continuity has been disrupted. Spatial continuity evaluation value. Calculated by the following formula:

[0218]

[0219] in, The normalized scale for the spatial central difference ranges from 1 to 20, and is 10 meters in this embodiment. The normalization scale for the interval jump variable ranges from 1 to 6, and is set to 3 in this embodiment. The scale used for spatial bilateral equilibrium normalization ranges from 8 to 20; in this embodiment, 10 meters is chosen. In this formula, if the spatial displacement difference before and after the candidate segment is small, the interval jumps are limited, and the bilateral spatial changes are approximately balanced, then... Higher; if a large step size spatial jump occurs, then A significant decrease. The balancing term added to the formula is used to suppress the misinterpretation of normal bilateral smoothing slip as spatial discontinuity.

[0220] In practice, if the train is accelerating, decelerating, or traveling on a curve, the speed can be appropriately increased. and To tolerate normal amplification of spatial changes; if the train is in a steady-speed straight-line inspection phase, the speed should be reduced. and This improves the sensitivity of spatial transition identification. After the above processing, a temporal continuity evaluation sequence and a spatial continuity evaluation sequence are obtained. These two sequences will be directly used for the next step of spatiotemporal coupling anomaly detection.

[0221] S4.3: Couple the temporal continuity evaluation value and the spatial continuity evaluation value to determine their consistency, construct a spatiotemporal coupling anomaly index, and identify anomalous rearrangement behavior that is temporally continuous but spatially abrupt.

[0222] After obtaining independent temporal and spatial continuity evaluation values, a coupling consistency judgment is performed on both to identify genuine anomalous rearrangement behavior. In this scenario, genuine anomalous rearrangement is usually not a situation where both time and space are disordered, but rather where time appears continuous, but space has jumped off the original physical slip trajectory. Therefore, the discriminant function must strengthen the combined characteristics of temporal continuity and spatial abrupt change, while weakening ordinary perturbations of slight fluctuations in both time and space.

[0223] Therefore, a spatiotemporal coupling anomaly index is constructed and superimposed with the final calibration criterion value output in step S3 to form a more robust discrimination result. Initial spatiotemporal coupling anomaly index for:

[0224]

[0225] in, This represents the final calibration criterion value of the j-th candidate segment in step S3; This represents the calibration compensation coefficient, ranging from 0.20 to 1.50; in this embodiment, it is set to 0.80. This formula is only applicable when there is high temporal continuity, low spatial continuity, and a strong tendency for anomalies has already been observed in step S3. Only then will it increase significantly.

[0226] However, if the spatial differences before and after a candidate segment are large, but a certain bilateral symmetry balance is still maintained, it may be normal obstacle avoidance or normal accompanying adjustment, and should not be directly judged as anomaly. Therefore, the initial index needs to be modified and suppressed. Modified spatiotemporal coupling anomaly index for:

[0227]

[0228] in, The normalization scale for bilateral spatial differences ranges from 0.5 to 1.5; in this embodiment, 1.5 meters is used. The formula uses the denominator to correct and suppress this bilateral balance disturbance: if the spatial differences are large but exhibit symmetrical balance, i.e. and If the squares are close, the squared term in the denominator is smaller, and the correction effect is weaker; if the spatial differences are asymmetrical, the denominator increases. It was effectively suppressed.

[0229] In practice, an anomaly detection threshold should be set. If If the value is higher than the threshold, the candidate segment is determined to be an abnormal rearrangement behavior; if it is lower than the threshold, it is retained as a non-abnormal candidate and transferred to the trusted set. In this embodiment, the anomaly discrimination threshold can be 0.55, with a value range of 0.35 to 0.75. A lower value is preferable during the train steady-speed inspection stage to improve detection sensitivity; a higher value is preferable during stages with more trackside obstructions or when the drone needs to frequently correct its position to reduce false positives. After the above processing, a spatiotemporal coupling anomaly discrimination sequence is obtained. This sequence records the corrected spatiotemporal coupling anomaly index of each candidate segment and the comparison result with the anomaly discrimination threshold, which will be directly used for the next step of anomaly behavior labeling and the generation of the trusted data segment set.

[0230] S4.4: Based on the comparison results between the corrected spatiotemporal coupling anomaly index and the anomaly discrimination threshold, generate an anomaly behavior label set and a reliable data segment set.

[0231] The final classification and encapsulation are performed on the spatiotemporal coupling anomaly discrimination sequence. For the corrected spatiotemporal coupling anomaly index... Segments exceeding the anomaly detection threshold are marked as abnormal rearrangement behavior and written into the abnormal behavior marker set; segments that do not exceed the threshold, along with segments marked as normal slip sequences in step S3, are written into the trusted data segment set.

[0232] To ensure that the output can be directly called in step S5, the abnormal behavior marker set should at least include the following: segment sequence number, time center value, spatial displacement interval number, and time continuity evaluation value. Spatial continuity evaluation value Corrected spatiotemporal coupling anomaly index And anomaly type markers. A reliable data segment set should at least include the following: segment sequence number, time center value, spatial displacement interval number, and final calibration criterion value. And trusted category tags.

[0233] Before encapsulation, adjacent segments of multiple consecutive abnormal segments can be merged. This process is performed if the time interval between adjacent abnormal segments does not exceed one segment time window. If the spatial displacement interval numbers differ by no more than 1 to 2 intervals, they can be merged into the same abnormal behavior segment so that subsequent step S5 can perform elimination and reconstruction according to the abnormal segment instead of a single segment.

[0234] Ultimately, a structured set of abnormal behavior markers and a set of trusted data segments are formed, which are directly used as input data for the secure output processing of data transmission based on anomaly removal and sequence reconstruction in step S5.

[0235] In the technical solution of this disclosure, anomaly behavior discrimination processing based on temporal and spatial coupling consistency is performed to jointly evaluate the temporal and spatial continuity of anomaly candidate segments and construct a spatiotemporal coupling anomaly index, thereby identifying anomalous rearrangement behavior that is temporally continuous but spatially abrupt. This step utilizes the inconsistency between temporal and spatial continuity as a discrimination criterion, solving the technical deficiency of existing technologies that cannot distinguish between normal sliding rearrangement and anomalous rearrangement behavior, and significantly improving the accuracy and robustness of anomaly detection.

[0236] S5: Based on the abnormal behavior marker set and the trusted data segment set, the abnormal data segments are eliminated, and the trusted data segments are reordered and spliced ​​according to the slip trajectory constraint sequence, and data transmission security output processing based on abnormal elimination and sequence reconstruction is performed.

[0237] Abnormal data segments are removed, and based on the trusted data segments, they are reordered and concatenated according to the sliding trajectory constraint sequence to restore the data stream that conforms to the law of relative motion. The output includes the abnormal rearrangement position, the consistency status of the sliding trajectory, and the transmission security result of the reconstructed data sequence. This is achieved through the following sub-steps.

[0238] S5.1: Perform joint indexing and organization on the abnormal behavior marker set and the trusted data segment set to generate a trusted retention sequence and an anomaly removal index sequence.

[0239] The abnormal behavior marker set and the reliable data segment set output in step S4 are processed by joint indexing. First, a unified segment index table is established based on the segment sequence number, time center value, and spatial displacement interval number. Then, the segments are binarized according to the abnormal markers to obtain the reliable retention sequence and the abnormal removal index sequence. Since there may be multiple consecutive abnormal rearrangement segments in the train accompanying inspection scenario, adjacent abnormal segments should be merged before removal to form an abnormal segment index, so as to avoid a large number of fragmented discontinuities during subsequent sequence reconstruction.

[0240] For the j-th data segment, define a trusted retention weight. and anomaly removal weight The anomaly detection result from step S4 is further transformed from a hard determination into adjustable weights, facilitating flexible retention or removal at anomaly boundary segments. (Confidential retention weights) The calculation formula is:

[0241]

[0242] in, This represents the corrected spatiotemporal coupling anomaly index of the j-th data segment output in step S4. The threshold value for anomaly detection is 0.35 to 0.55, and in this embodiment it is 0.55. This is the weight compression coefficient, ranging from 4 to 6; in this embodiment, it is set to 10. The formula uses the Sigmoid function form, where... far below hour, Approaching 1; when Much higher hour, It approaches 0; it transitions smoothly near the threshold.

[0243] Anomaly Removal Weight Complementary to trusted retention weights:

[0244]

[0245] In specific implementation, if If the value is greater than or equal to 0.50, then the segment is written into the trusted reservation sequence; if... If the value is less than 0.5, the segment is added to the anomaly removal index sequence. For consecutive anomaly segments, the first and last numbers, the first and last times, and the first and last spatial intervals are recorded together. After the above processing, a reliable retention sequence and anomaly removal index sequence are obtained. The former is used for subsequent reordering and concatenation, and the latter is used to generate anomaly location reports.

[0246] S5.2: Using the combination of the center position of the spatial displacement interval, the constraint trajectory expression value, and the segmented time center value as the rearrangement basis, the reliable retained sequence is reordered according to the slip trajectory constraint sequence.

[0247] The trusted reserved sequence is reordered according to the sliding trajectory constraint sequence. It cannot be sorted simply by reception time, because the technical defect that this scheme aims to solve is the problem of temporal continuity but spatial correspondence offset; therefore, the sorting basis must return to the sliding trajectory constraint sequence output in step S2, and the combination of the center position of the spatial displacement interval, the constraint trajectory expression value and the segmented time center value is used as the basis for reordering.

[0248] In practice, the process involves first sorting by spatial displacement interval numbers, and then fine-tuning by the time center value within the same interval. If two segments are in adjacent intervals but their time positions overlap, the comprehensive constraint factor from step S2 is further used as a priority parameter. Therefore, a rearrangement priority value is constructed. :

[0249]

[0250] in, This represents the spatial displacement interval number of the j-th reliable data segment; This represents the time center value of the j-th trusted data segment; This indicates the segmented time window used in step S1; This represents the comprehensive constraint factor of the j-th constrained trajectory node output in step S2; This is the time fine-tuning coefficient, with a value ranging from 0.001 to 0.005; in this embodiment, it is set to 0.01. The constraint compensation coefficient ranges from 0.01 to 0.1, and is set to 0.01 in this embodiment. The formula uses the spatial displacement interval order as the primary sorting criterion, the time center value as the secondary sorting criterion, and a comprehensive constraint factor to suppress the interference of boundary fuzzy segmentation on the sorting results.

[0251] In practice, the rearrangement priority value of all reliable segments is first calculated, and then they are sorted in ascending order to form a constrained rearrangement sequence. This sequence determines the order of each segment during subsequent splicing and reconstruction, and is a key step in restoring the data flow that conforms to the law of relative motion.

[0252] S5.3: Perform splicing and reconstruction processing on adjacent trusted segments, determine whether the time gap and spatial gap are within the allowable range, generate a reconstructed data sequence based on the judgment result, and calculate the local reconstruction consistency coefficient and the global reconstruction consistency coefficient.

[0253] After the trusted segment rearrangement is completed, adjacent trusted segments are spliced ​​and reconstructed to restore the data flow that conforms to the law of relative motion. The splicing process first determines whether there are acceptable time and spatial gaps between adjacent trusted segments. If both time and spatial gaps are within the allowable range, they are spliced ​​directly. If a short gap exists, a transition segment index is generated using the boundary information of adjacent trusted segments, but no forged business content is generated; it is only marked as a transitional empty segment in the result. The allowable range is set based on the length of the spatial displacement interval. That is, if the center difference between the spatial displacement intervals of the preceding and following segments does not exceed one interval length, it is determined that direct splicing is possible; if it exceeds one interval length but is still within a limited expansion range, it is determined as a short gap, and the generation of a transitional empty segment index is allowed.

[0254] To evaluate the overall continuity of the reconstructed sequence, a local reconstruction consistency coefficient is defined. Consistency coefficient of global reconstruction For the j-th trusted data segment ( The local reconstruction consistency coefficient between it and the previous trusted data segment is:

[0255]

[0256] in, and These represent the time center values ​​of the current trusted segment and the previous trusted segment, respectively. For segmented time windows; The time interval normalization scale ranges from 0.2 to 0.5, and is set to 0.10 seconds in this embodiment; and These represent the center positions of the spatial displacement intervals of the current trusted segment and the previous trusted segment, respectively. The normalized scale for spatial clearance ranges from 1 to 2; in this embodiment, 8 meters is used. This formula penalizes deviations of both time and spatial clearances from ideal values; the smaller the clearance, the more... The closer it is to 1.

[0257] Global Reconstruction Consistency Coefficient The average of the consistency coefficients for all local reconstructions:

[0258]

[0259] in, This represents the total number of reliable segments after reconstruction. This value reflects the overall smoothness of the reconstructed data sequence. The closer to 1, the higher the reconstruction quality.

[0260] In specific implementation, if If the value is below a set threshold (0.60 in this embodiment), this location can be marked as the reconstruction stitching boundary; if multiple consecutive values ​​are below this threshold, this location can be marked as the reconstruction stitching boundary. If the value is too low, weak sections will be formed in the reconstruction. Finally, a reconstructed data sequence is obtained, sorted according to the sliding trajectory constraints and then spliced ​​together.

[0261] S5.4: Jointly encapsulate the reconstructed data sequence and the anomaly removal index sequence to generate a transmission security result that includes the anomaly rearrangement position result, the slip trajectory consistency status result, and the reconstructed data sequence body.

[0262] The reconstructed data sequence and the anomaly removal index sequence are jointly encapsulated to generate the final transmission security result. This result contains at least three parts: the first part is the anomaly rearrangement position result, i.e., the sequence number range, time range, and spatial displacement interval range of the anomaly segment; the second part is the slip trajectory consistency status result, i.e., the global reconstruction consistency coefficient obtained in step S5.3. The third part is the reconstructed data sequence ontology, including the boundary states of each local splicing.

[0263] To further quantify the final secure output quality, a transmission security output score is defined. :

[0264]

[0265] in, This represents the global reconstruction consistency coefficient. The total number of segments to be removed due to anomalies; This represents the total number of original received segments; This represents the cumulative time length of the abnormal segment. This represents the total duration of the original received data. Based on global reconstruction consistency, this formula considers both the proportion of abnormal segments and the proportion of abnormal duration to comprehensively evaluate whether the transmission is in a trustworthy and secure state.

[0266] In practice, multi-level output rules are set. When When the value is greater than or equal to 0.80, output a high-confidence reconstruction result; when... When the value is between 0.50 and 0.80, the output is usable but there are exceptions that are discarded; when... When the value is below 0.50, output reconstruction is limited, requiring retransmission or manual review. The final result includes the abnormal rearrangement positions, the consistency status of the slip trajectory, and the reconstructed data sequence, achieving secure data transmission protection in train-accompanied inspection scenarios.

[0267] In the technical solution of this disclosure, by performing data transmission security output processing based on anomaly removal and sequence reconstruction, abnormal data segments are removed, and the segments are reordered and spliced ​​according to the slip trajectory constraint sequence based on the trusted data segments to restore the data stream that conforms to the law of relative motion. This step removes the polluted data caused by abnormal rearrangement while ensuring data integrity, and outputs abnormal location information and reconstruction consistency status, providing an engineering-implementable security protection mechanism for UAV data transmission in train-accompanying inspection scenarios.

[0268] This disclosure also provides a security protection system for unmanned aerial vehicle (UAV) data transmission. Figure 3 This is a schematic diagram of a drone data transmission security protection system provided according to an embodiment of this disclosure. This system is used to run the drone data transmission security protection methods described in the above embodiments. (Refer to...) Figure 3 The system may include:

[0269] The data segmentation spatial mapping construction module is used to continuously sample the relative motion trajectory of the UAV and the train, establish a one-to-one correspondence between data segments and spatial displacement intervals, and construct a data segmentation spatial mapping expression based on the relative motion trajectory.

[0270] The sliding trajectory constraint sequence construction module is used to connect the displacement change relationship between adjacent segments in the data segmentation spatial mapping structure, apply monotonicity and continuity constraints, perform sliding trajectory constraint sequence construction processing based on spatial displacement interval, and form a sliding trajectory constraint sequence that satisfies the laws of physical motion.

[0271] The sequence rearrangement feature calibration module is used to match and align the data segment sequence at the receiving end with the sliding trajectory constraint sequence, calculate the matching score, perform sequence rearrangement feature calibration processing based on sliding trajectory consistency, and output a set of rearrangement feature calibration results.

[0272] The abnormal behavior discrimination module is used to evaluate the temporal continuity and spatial continuity of the abnormal rearrangement candidate segments in the rearrangement feature labeling result set, construct a spatiotemporal coupling anomaly index, perform abnormal behavior discrimination processing based on the consistency of temporal and spatial coupling, and generate an abnormal behavior label set and a reliable data segment set.

[0273] The data transmission security output module is used to remove abnormal data segments based on the abnormal behavior marker set and the trusted data segment set, and to reorder and splice the trusted data segments according to the slip trajectory constraint sequence, and perform data transmission security output processing based on abnormal removal and sequence reconstruction.

[0274] According to embodiments of this disclosure, an electronic device is also provided, which may include a processor, a communications interface, a memory, and a communication bus, wherein the processor, the communications interface, and the memory communicate with each other via the communication bus. The processor can invoke logical instructions stored in the memory to execute the methods provided in the above embodiments.

[0275] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a part of the 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 to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0276] On the other hand, this disclosure also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the methods provided in the above embodiments.

[0277] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0278] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0279] It should be understood that the above embodiments are only used to illustrate the technical solutions of this disclosure, and not to limit them; although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure.

Claims

1. A method for ensuring secure data transmission from unmanned aerial vehicles (UAVs), characterized in that, The method includes: S1: Continuously sample the raw trajectory data of the UAV and the train to obtain the relative motion trajectory, establish a one-to-one correspondence between data segments and spatial displacement intervals, and construct a data segment spatial mapping structure based on the relative motion trajectory; S2: Connect the displacement change relationship between adjacent segments in the data segmentation spatial mapping structure, apply monotonicity and continuity constraints, and perform the construction of a sliding trajectory constraint sequence based on the spatial displacement interval to form a sliding trajectory constraint sequence that satisfies the laws of physical motion. S3: Match and align the data segment sequence from the receiving end with the slip trajectory constraint sequence, calculate the matching score, perform sequence rearrangement feature calibration processing based on slip trajectory consistency, and output the rearrangement feature calibration result set; S4: Perform temporal continuity evaluation and spatial continuity evaluation on the abnormal rearrangement candidate segments in the rearrangement feature labeling result set, construct a spatiotemporal coupling anomaly index, perform abnormal behavior discrimination processing based on the consistency of temporal and spatial coupling, and generate an abnormal behavior label set and a reliable data segment set. S5: Based on the abnormal behavior marker set and the trusted data segment set, the abnormal data segments are eliminated, and the trusted data segments are reordered and spliced ​​according to the slip trajectory constraint sequence, and data transmission security output processing based on abnormal elimination and sequence reconstruction is performed.

2. The method according to claim 1, characterized in that, S1 includes: Perform unified time-base resampling processing on the raw transmission data stream of the UAV, the UAV trajectory sampling sequence, and the train trajectory sampling sequence to obtain a unified time-base sampling sequence; The unified time-base sampling sequence is segmented according to a fixed time window to form data segments with time identifiers and sequence numbers; Based on the three-dimensional coordinates of the UAV and the train at the center value of each data segment time, calculate the relative displacement in the track direction, the lateral relative displacement, the height relative displacement, and the relative displacement state value. Based on the relative displacement and relative displacement state value of each data segment in the orbital direction, the data segment is mapped to the corresponding spatial displacement interval, forming a one-to-one correspondence between time identifier, sequence number, relative displacement state and spatial displacement interval.

3. The method according to claim 1, characterized in that, S2 includes: Extract the spatial displacement interval number, interval center position, relative displacement state value and time center value between adjacent data segments, and construct the interval association sequence of adjacent data segments; The adjacent data segment interval association sequence is subjected to forward concatenation calculation to generate the original slip trajectory expression sequence containing continuous displacement accumulation and state change suppression; The original slip trajectory expression sequence is subjected to monotonicity constraint processing to remove pseudo-slip segments with non-physical direction flips, resulting in a monotonicity constraint trajectory sequence. Apply continuous constraints to the monotonic constraint trajectory sequence to limit abrupt changes in segmented slip velocity and sharp bends in relative displacement state values, and output the slip trajectory constraint sequence.

4. The method according to claim 1, characterized in that, S3 includes: Using the segment sequence number in the slip trajectory constraint sequence as the retrieval key, the corresponding receiving segment record is extracted from the receiving buffer to establish an aligned observation sequence between the receiving sequence and the constraint trajectory. For each observation pair in the aligned observation sequence, the spatial center offset, velocity difference, and difference in relative displacement state value are calculated and combined to obtain a single-segment matching score; Based on the single-segment matching score, the spatial displacement change trends of the received sequence and the constraint trajectory within a continuous window are compared to calculate the continuous slip trend matching sequence; Based on the calculated continuous slip trend aggregated score, combined with the comprehensive constraint factor and score fluctuation amplitude of the constraint trajectory nodes, the received data is segmented and marked as normal slip sequence or abnormal rearrangement candidate sequence, generating a set of rearrangement feature calibration results.

5. The method according to claim 1, characterized in that, S4 includes: Extract abnormal rearrangement candidate segments from the rearrangement feature labeling result set, construct local spatiotemporal correlation units around the candidate segments and their preceding and following segments, and form an abnormal candidate spatiotemporal correlation sequence; The temporal continuity evaluation value and spatial continuity evaluation value are calculated for the abnormal candidate spatiotemporal correlation sequence to obtain the temporal continuity evaluation sequence and the spatial continuity evaluation sequence. The temporal continuity evaluation value and the spatial continuity evaluation value are coupled and consistent to determine the spatiotemporal coupling anomaly index. Based on the comparison between the corrected spatiotemporal coupling anomaly index and the anomaly determination threshold, anomalous rearrangement behavior that is temporally continuous but spatially abrupt is identified. The identified anomalous rearrangement behavior is assigned to the anomalous behavior label set, and the remaining segments, together with the normal slip sequence, are assigned to the reliable data segment set.

6. The method according to claim 5, characterized in that, S4 further includes: For segments where the corrected spatiotemporal coupling anomaly index is higher than the anomaly discrimination threshold, they are marked as anomalous rearrangement behavior and written into the anomalous behavior label set; For segments that do not exceed the threshold, along with segments that have been marked as normal slip sequences, they are written into the trusted data segment set. Multiple consecutive abnormal segments are merged. If the time interval between adjacent abnormal segments does not exceed one segment time window and the spatial displacement interval numbers do not differ by a preset range, they are merged into the same abnormal behavior segment.

7. The method according to claim 1, characterized in that, S5 includes: The abnormal behavior marker set and the trusted data segment set are jointly indexed and organized to generate a trusted retention sequence and an abnormal removal index sequence. The combination of the center position of the spatial displacement interval, the constraint trajectory expression value, and the segmented time center value is used as the rearrangement basis to reorder the reliable preserved sequence according to the slip trajectory constraint sequence. The adjacent trusted segments after reordering are spliced ​​and reconstructed to determine whether the time gap and spatial gap are within the allowable range, and a reconstructed data sequence is generated based on the determination result. Based on the reconstructed data sequence, calculate the local reconstruction consistency coefficient and the global reconstruction consistency coefficient to generate the slip trajectory consistency state result; The reconstructed data sequence and the anomaly removal index sequence are jointly encapsulated to generate a transmission security result that includes the anomaly rearrangement position result, the slip trajectory consistency status result, and the reconstructed data sequence body.

8. A UAV data transmission security protection system, applied to the method of any one of claims 1-7, characterized in that, The system includes: The data segmentation spatial mapping construction module is used to continuously sample the raw trajectory data of UAVs and trains to obtain relative motion trajectories, establish a one-to-one correspondence between data segments and spatial displacement intervals, and construct a data segmentation spatial mapping structure based on relative motion trajectories. The sliding trajectory constraint sequence construction module is used to connect the displacement change relationship between adjacent segments in the data segmentation spatial mapping structure, apply monotonicity and continuity constraints, and perform sliding trajectory constraint sequence construction based on spatial displacement interval to form a sliding trajectory constraint sequence that satisfies the laws of physical motion. The sequence rearrangement feature calibration module is used to match and align the data segment sequence at the receiving end with the sliding trajectory constraint sequence, calculate the matching score, perform sequence rearrangement feature calibration processing based on sliding trajectory consistency, and output a set of rearrangement feature calibration results. The abnormal behavior discrimination module is used to evaluate the temporal continuity and spatial continuity of the abnormal rearrangement candidate segments in the rearrangement feature labeling result set, construct a spatiotemporal coupling anomaly index, perform abnormal behavior discrimination processing based on the consistency of temporal and spatial coupling, and generate an abnormal behavior label set and a reliable data segment set. The data transmission security output module is used to remove abnormal data segments based on the abnormal behavior marker set and the trusted data segment set, and to reorder and splice the trusted data segments according to the slip trajectory constraint sequence, and perform data transmission security output processing based on abnormal removal and sequence reconstruction.

9. An electronic device, characterized in that, The electronic device includes a memory and at least one processor, the memory storing a computer program, and the processor executing the computer program to implement the UAV data transmission security protection method according to any one of claims 1-7.

10. A computer storage medium, characterized in that, It stores a computer program, which, when executed by a processor, implements the UAV data transmission security protection method according to any one of claims 1-7.

Images