An intelligent terrain scanning and three-dimensional modeling method based on a UAV airfield and a laser radar

Abstract

The application discloses an intelligent terrain scanning and three-dimensional modeling method based on an unmanned aerial vehicle airport and a laser radar, and particularly relates to the field of unmanned aerial vehicle terrain mapping, and is used for solving the problem of lacking sustainable tracking and verifiable consistency anchor points in cross-end attitude and time, and is achieved by reading an inertial navigation attitude reference to generate an anchor record as a cross-end tracking starting point when scanning is started, generating a reaching order signature for each collection segment in the airport forwarding stage and binding the signature to the anchor record to form a tracking linked list, performing verification according to the tracking linked list on the ground side, dividing available intervals and breakpoint intervals according to the continuous relationship of the order signature and the attitude reference to form an interval index, calculating an attitude angle difference and an order dislocation rate in the breakpoint interval, generating a splicing decision by using an evidence synthesis method, selecting an order playback path or an attitude inheritance path according to the splicing decision to complete splicing, and implementing joint inspection on the splicing result to determine a final strip and a re-navigation reference to establish a traceable record.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) terrain mapping, and more specifically, to an intelligent terrain scanning and 3D modeling method based on UAV airports and lidar. Background Technology

[0002] The UAV completes self-checks and takeoffs / landings within the airport, and, equipped with a lidar system, performs terrain scanning along the flight path, generating point clouds, attitude references, and time reference markers. After the operation, the original files are output to the onboard unit via a wired network, and then relayed to the ground workstation via the airport's wireless link. The ground station receives and categorizes the files according to the task directory in unattended mode, and then triggers attitude calculation and point cloud fusion to generate 3D results that fit the terrain undulations. It is expected that the strip references will remain consistent, the stitching will be smooth, and the results will be reusable during multiple flights.

[0003] However, the existing technical problems are: there is a lack of sustainable tracking and verifiable consistency anchors for attitude and time at different ends. Data from the aircraft, airport, and ground ends are triggered by the arrival of documents. The processing entry point does not establish a common binding and consistent verification between attitude reference and time benchmark. The alignment relationship inside and outside the strip cannot be confirmed in the early stages of the process, the strip base plane is difficult to continue, the results of the re-flight cannot be used with existing benchmarks, inconsistencies appear in the terrain model at the joints, and the overall stability and comparability of the three-dimensional structure are limited. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an intelligent terrain scanning and 3D modeling method based on UAV airports and lidar. This method generates anchor records at the start of scanning as the starting point for cross-end tracking. During the airport forwarding phase, arrival sequence signatures are established and a tracking list is formed. On the ground side, verification is performed to divide the interval index. At the breakpoint intervals, attitude angle differences and sequence misalignment rates are calculated. An evidence synthesis method is used to generate a splicing decision to complete the splicing. Seam inspection determines the final strip and re-entry benchmark, establishing a traceable record to solve the problems mentioned in the background art.

[0005] According to a first aspect of the embodiments of this application, a method for intelligent terrain scanning and 3D modeling based on unmanned aerial vehicle (UAV) airports and lidar is provided, comprising: S1: Read the inertial navigation attitude reference when the scan starts and generate an anchor record as the starting point for cross-end tracking; S2: During the airport transfer phase, generate an arrival sequence signature for each collected segment and bind the signature to the anchor record to form a tracking list; S3: The ground side performs verification based on the tracking chain, and divides the available interval and breakpoint interval according to the sequential signature and attitude reference continuity to form an interval index; S4: Calculate the attitude angle difference and sequence misalignment rate in the breakpoint interval, generate splicing decision using the evidence synthesis method, and select the sequential replay path or attitude inheritance path to complete the splicing. S5: Conduct joint inspection on the splicing results, determine the final strip and re-entry benchmark based on the inspection results, and establish traceable records.

[0006] Further, step S1 includes: At the start of the scan, read the attitude reference and start time, and embed the timestamp into the attitude data structure; The attitude reference is converted into a three-dimensional vector form represented by Euler angles, the noise is optimized, the attitude orientation is obtained, and a start frame marker containing frame serialization and initial point cloud index is generated for the first point cloud data frame generated by the lidar. Using a hash function, a unique identifier is calculated for the start time, attitude reference, attitude orientation, and start frame marker, forming an anchor record as the starting point for cross-end tracking.

[0007] Further, step S2 includes: Each acquisition segment is processed according to the segment reception order. The acquisition sequence tag in each acquisition segment is read, and the difference between the acquisition sequence tag and the actual arrival position is compared to generate the sequence signature basic data, including the difference sequence and the out-of-order position list. The acquisition segment is a data unit formed by packaging on the machine according to a preset time window or a preset number of frames, and includes at least a continuous point cloud frame sequence, segment identifier, acquisition sequence tag, start and end timestamps, and attitude / position information corresponding to the point cloud frame. Based on the difference sequence, out-of-order features and main out-of-order patterns are extracted, and the main out-of-order patterns are encoded into fragment-level sequential signatures. The fragment-level sequential signature is bound to the anchor record, and the linked list nodes are appended to the tracking linked list in the order of fragment arrival to form a linear link. The tracking linked list includes fragment representations and fragment-level sequential signature-level anchor record references. During the appending process, the reference consistency is verified. The reference consistency includes at least the following: the anchor IDs of all linked list nodes are consistent and match the anchor record hash identifier transmitted with the linked list.

[0008] Further, based on the difference sequence, disorder features and main disorder patterns are extracted, including: The feature extraction method is applied to the difference sequence of the collected segments. Specifically, the disordered features are characterized by calculating the cumulative sum of continuous differences. Principal component analysis (PCA) is used to isolate the main disordered patterns. Specifically, the covariance matrix of the difference sequence is constructed, where each element is the covariance value of each pair of elements in the sequence. Eigenvalue decomposition is performed to solve for the eigenvectors and eigenvalues ​​of the covariance matrix. The eigenvalues ​​are then sorted in descending order to identify the dominant components. The difference sequence is projected onto the first few eigenvectors to separate the main disordered patterns.

[0009] Further, step S3 includes: The ground station receives the tracking list, reads the segment identifier, segment-level sequential signature, and anchor record reference for each node, confirms the segment timing relationship according to the segment-level sequential signature, and generates a timing consistency tag list. Based on the temporal consistency label list, the attitude continuity of adjacent segments is examined by the attitude reference recorded by anchor and the attitude jump position list is recorded. The attitude jump position list is merged with the temporal consistency label list to form a joint consistency evaluation dataset. Using the joint consistency evaluation dataset, the fragment sequences in the tracking chain are divided. A cluster of continuous fragments that simultaneously satisfies the sequential signature temporal consistency and attitude continuity is divided into a usable interval; otherwise, it is divided into a breakpoint interval. For each interval, the start and end positions, the segment to which it belongs, the interval ID, and the consistency type are assigned to form an interval index.

[0010] Further, step S4 includes: The start and end positions and the segments to which a specific breakpoint interval belongs are read from the interval index. The attitude orientation in the anchor record is used as a reference. The attitude orientation of adjacent segments before and after the breakpoint is extracted. The angular distance between vectors is calculated pair by pair to obtain the attitude angular difference of all breakpoint positions in each breakpoint interval. The segment identifier list formed based on the actual arrival sequence within the breakpoint interval is selected and compared with the collection sequence marker list to obtain the sequence misalignment rate; The attitude angle difference is mapped to the initial confidence of the attitude inheritance path, and the sequence misalignment rate is mapped to the initial confidence of the sequence replay path. The two initial confidences are fused and the confidence of each path is calculated. The path with the highest confidence is used as the splicing decision. If the splicing decision is a sequential replay path, the breakpoint interval segment sequence is reconstructed according to the acquisition order. If the decision is an attitude inheritance path, the attitude reference of the anchor record is locked at the breakpoint position as the inheritance benchmark, the attitude orientation of subsequent segments is adjusted to match the previous segment, and the splicing decision record is retained, including attitude angle difference, sequence misalignment rate, initial confidence level, fusion result and path selection. Among them, the sequential replay path refers to reconstructing and replaying the segment sequence of the breakpoint interval according to the original acquisition sequence marking on the machine, and the attitude inheritance path refers to locking the anchor record attitude reference at the breakpoint as the inheritance benchmark, and uniformly correcting the attitude orientation of the segment after the breakpoint to make it continuous with the strip base plane before the breakpoint.

[0011] Furthermore, the attitude angle difference is mapped to the initial confidence level of the attitude inheritance path, and the sequence misalignment rate is mapped to the initial confidence level of the sequence replay path. The two initial confidence levels are fused and the confidence level of each path is calculated. The path with the highest confidence level is used as the splicing decision, specifically: The attitude angle difference is passed through an exponential decay function. The inverse mapping is the initial trust level of the pose inheritance path, and the order misalignment rate is reduced by an exponential decay function. The inverse mapping is the initial trust level of the sequential replay path; The Dempster-Shafer evidence theory is used to fuse two initial confidence levels to assign a basic probability mass function to the pose inheritance path. Uncertainty ,in The framework includes pose inheritance paths and sequential replay paths; a basic probability mass function is assigned to the sequential replay paths. Uncertainty ; The Dempster combinatorial rule is applied to compute the fusion quality function for each non-empty subset A. Where K is the conflict factor To resolve conflicting evidence and normalize it, where A, B, and C are all... A subset of; The confidence level of each path is calculated using pignamistic probability. , ,in This represents the confidence / probabilistic support of selecting the pose inheritance path as the correct splicing path; This represents the confidence / probabilistic support level of selecting the sequential replay path as the correct concatenation path. , , ; The path with higher credibility is selected as the sole concatenation decision.

[0012] Further, step S5 includes: Read the spliced ​​strip data from the splicing decision record, identify the overlapping areas between adjacent strips, perform geometric consistency inspection on the point cloud within the overlapping areas, and generate a consistency report; Based on the consistency report, local deformation is analyzed in the overlapping area, abnormal deformation locations are marked, and the list of geometric consistency deviation locations and abnormal deformation locations are merged to form a comprehensive inspection index list. Using the comprehensive inspection index list, the spliced ​​strips are inspected as a whole. If the inspection passes, the current strip is determined to be the final strip. The strip base plane and attitude reference are extracted from the final strip as the re-navigation benchmark and archived. If the inspection fails, the problem section is marked and the entry into the database is stopped. The specific location and anomaly type of the problem section are marked in the task directory. Save the judgment trajectory, that is, the entire process log from the start of the joint inspection to the conclusion, including the consistency report, the list of inspection indicators, and the archived operation record.

[0013] According to a second aspect of the embodiments of this application, an electronic device is provided, comprising: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors perform the method as described in the first aspect.

[0014] According to a third aspect of the embodiments of this application, a computer-readable storage medium is provided that stores computer instructions thereon, which, when executed by a processor, implement the steps of the method as described in the first aspect.

[0015] The technical solutions provided by the embodiments of this application may include the following beneficial effects: As can be seen from the above embodiments, this application uses scanning to initiate the anchoring attitude reference, the airport side accumulates the arrival sequence signature, the ground side completes the double consistency verification at the process entry, and uses the attitude angle difference and sequence misalignment rate as the sole criteria at the breakpoint. A single splicing decision is formed through evidence synthesis, and the end uses the seam inspection feedback as the re-flight benchmark, forming a continuous tracking chain from collection, transmission, preprocessing to result solidification. This ensures that the strip geometry remains stable during cross-end conversion, the seam misalignment no longer accumulates, the re-flight results remain comparable using the same source benchmark, the audit trajectory is complete and can be reviewed, the parameter dimensions are simplified and the judgment is clear, and the consistent attitude and time reference is maintained even under unattended conditions. The accuracy, coherence and reusability of 3D modeling are improved simultaneously.

[0016] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] Figure 1 This is a flowchart illustrating an intelligent terrain scanning and 3D modeling method based on an unmanned aerial vehicle (UAV) airport and LiDAR, according to an exemplary embodiment.

[0019] Figure 2 This is a schematic diagram of an electronic device according to an exemplary embodiment. Detailed Implementation

[0020] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application.

[0021] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0022] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0023] Figure 1 This is a flowchart illustrating an intelligent terrain scanning and 3D modeling method based on an unmanned aerial vehicle (UAV) airport and LiDAR, according to an exemplary embodiment. Figure 1 As shown, this method, when applied to a terminal, may include the following steps: S1: Read the inertial navigation attitude reference when the scan starts and generate an anchor record as the starting point for cross-end tracking.

[0024] S2: During the airport transfer phase, generate an arrival sequence signature for each collected segment and bind the signature to the anchor record to form a tracking list.

[0025] S3: The ground side performs verification based on the tracking chain, and divides the available intervals and breakpoint intervals according to the sequential signature and attitude reference continuity to form an interval index.

[0026] S4: Calculate the attitude angle difference and sequence misalignment rate in the breakpoint interval, generate a splicing decision using the evidence synthesis method, and select the sequential replay path or attitude inheritance path to complete the splicing.

[0027] S5: Conduct joint inspection on the splicing results, determine the final strip and re-entry benchmark based on the inspection results, and establish traceable records.

[0028] In a drone airport environment, establishing a reliable cross-end tracking starting point is crucial as the drone, equipped with lidar, is about to initiate a terrain scanning mission. This ensures that attitude references and time bases possess sustainable continuity and verifiability from the outset, preventing potential alignment deviations and stripe inconsistencies during subsequent transmission and processing. However, existing technologies often neglect precise anchoring at the start, leading to a loss of unified reference for onboard data during airport forwarding and ground fusion, causing problems such as inconsistent stitching and unusable results. Therefore, this step systematically reads, records, and merges key information to generate anchor records, laying the foundation for consistency at the starting point of the entire intelligent terrain scanning and 3D modeling process.

[0029] Specifically, step S1 may include the following sub-steps: S1.1 Read and record the attitude reference and start time.

[0030] At the moment the UAV scanning mission begins, attitude references, including spatial orientation data such as roll, pitch, and yaw angles, are directly acquired from the inertial navigation system. This ensures that a synchronous sampling mechanism is used to minimize errors caused by sensor latency. Subsequently, the start time is immediately captured and timestamped using a high-precision clock source synchronized with GPS. This timestamp is stored in UTC format and accurate to the millisecond level to prevent time drift from affecting cross-end timing. The recording process of attitude references and start timestamps emphasizes their binding; the timestamps are embedded in the attitude data structure to form an initial temporal-attitude correlation. After this process, the attitude references and start timestamps serve as the spatiotemporal basis for anchor recording, providing a traceable initial reference for subsequent attitude orientation refinement and tracking list construction, ensuring reliable input for continuity judgments during breakpoint interval verification.

[0031] S1.2 Refine the recording of attitude orientation and start frame marker.

[0032] Based on the acquired attitude reference, the roll, pitch, and yaw angles are further converted into three-dimensional vectors represented by Euler angles and stored as attitude orientation. The attitude orientation data structure carries a start-time field (or a reference to the start-time in the anchor record) to ensure the homogeneity of the attitude reference, attitude orientation, and time base. Least squares optimization is used to process sensor noise, improving the accuracy and stability of the attitude orientation. Simultaneously, when the lidar generates its first point cloud data frame, it is marked as a start-frame marker, including a frame sequence number and initial point cloud index. A frame synchronization protocol ensures no offset in the association with the start-time and attitude orientation. The recording of attitude orientation and start-frame markers focuses on the complete capture of static moments, preventing frame-level start-point deviations from affecting the continuity of the strip reference. After recording, these elements enhance the attitude and frame dimensions of the anchor record, supporting the calculation of attitude angle differences in the evidence synthesis method and the preprocessing entry partitioning of the interval index.

[0033] S1.3 Merge the three pieces of information to generate an anchor record.

[0034] The startup time, attitude orientation, and starting frame marker are integrated with the attitude reference, and a hash function is used to calculate a unique identifier for this information, forming a single anchor record structure. This structure is stored in binary format for easy transmission via the wireless link during airport transit and for retrieval on the ground side. The merging process includes an integrity verification step; if any inconsistency is detected in any component, a scan restart mechanism is triggered to ensure consistency of the starting point. After the anchor record is generated, it serves as the sole starting point for the scan task, defining the reference system for cross-end tracking. It supports sequential signature binding and appending to the tracking list, thereby maintaining alignment within and outside the strip in unattended mode.

[0035] Step S1 establishes a unified anchor point for scanning initiation, ensuring that attitude reference and time reference are bound together from the source of acquisition. This provides a consistent verification basis for the airport and ground phases of verification and stitching, avoids the accumulation of early alignment deviations, and thus significantly improves the accuracy, consistency and reusability of 3D modeling results.

[0036] In the stage of collecting raw files on the UAV airport receiver, after the operation is completed, the raw files are output to the airport unit via a wired network. At this time, it is urgent to generate a reliable arrival sequence signature for each collected segment and bind it to the anchor record to fix the cross-end timing reference and ensure that the order during transmission is verifiable, thereby supporting the interval division and splicing decision on the ground side. However, in existing technologies, airport forwarding is only triggered by file arrival, ignoring the extraction and signature binding of segment-level out-of-order features. This results in a lack of consistency verification between attitude reference and time base at the cross-end, causing problems such as difficulty in continuing the strip base plane and the inability to reuse the results of re-flight. Therefore, this step establishes a tracking chain through comparison, extraction and binding operations to accumulate the sequence information on the airport side, providing a stable timing chain for 3D modeling in unattended mode, ensuring smooth strip splicing and significantly enhancing the reusability of results.

[0037] S2.1 Receives the collected segments and compares them to generate a sequential signature basis.

[0038] Before relaying via the wireless link, the airport unit processes each acquired segment according to the segment reception order. Each acquired segment is a data unit packaged on-board according to a preset time window or preset frame number, and includes at least: a continuous point cloud frame sequence, segment identifier, acquisition sequence tag, start and end timestamps, and attitude / position information corresponding to the point cloud frame. First, the acquisition sequence tag within the segment is read. This tag originates from the built-in sequence number in the point cloud data generated on-board. Then, this tag is compared with the actual arrival position (the enqueue index position in the airport receiving buffer). The potential out-of-order degree is quantified by calculating the numerical difference between the acquisition sequence tag and the arrival position for each segment, ensuring that the comparison process covers all segments to capture global transmission deviations, including cases of reversed order, insertion, or missing segments. The comparison uses the Levenshtein distance algorithm. Specifically, a matrix is ​​initialized, where rows represent the acquisition sequence tag sequence and columns represent the arrival position sequence. The matrix elements calculate the cost of editing operations such as insertion, deletion, or replacement. The minimum edit distance quantifies the out-of-order degree, avoiding the omission of complex out-of-order patterns by simple differences. After the comparison is completed, the basic data of the sequential signature is generated, including the difference sequence and the list of out-of-order positions, which serve as input for extracting out-of-order features and support the temporal fixation of subsequent tracking of the linked list.

[0039] S2.2 Extract disordered features and form fragment-level sequential signatures.

[0040] Based on the generated sequential signature data, a feature extraction method is applied to the difference sequence of the collected segments. Specifically, the disordered features are characterized by calculating the cumulative sum of continuous differences. The cumulative sum is added item by item to reflect the magnitude of the order reversal during transmission. At the same time, the peak position of the cumulative sum exceeding the preset threshold is identified to highlight significant deviations. This ensures that the extraction process only targets discontinuous segments rather than continuous normal parts, thereby simplifying the computational load and focusing on the problem area.

[0041] Principal component analysis (PCA) is employed to isolate the primary out-of-order patterns. First, a covariance matrix of the difference sequence is constructed, where each element represents the covariance of each pair of elements in the sequence, used to capture linear relationships between variables. Second, eigenvalue decomposition is performed to solve for the eigenvectors and eigenvalues ​​of the covariance matrix, which are then sorted in descending order of eigenvalues ​​to identify the dominant components. Next, the difference sequence is projected onto the first few eigenvectors, retaining the main components that explain more than 80% of the variance, thus separating the primary out-of-order patterns, such as continuous inversions or periodic shifts, while eliminating secondary noise represented by the remaining components to avoid noise interference with signature accuracy. After out-of-order feature extraction, the separated primary out-of-order patterns are encoded into fixed-length hash values, forming fragment-level sequential signatures. These signatures uniquely identify the arrival time sequence characteristics of the corresponding collected fragments, facilitating binding to anchor records and confirming the fragment time sequence relationship during ground verification.

[0042] S2.3 Bind sequential signatures and append them to form a tracking linked list.

[0043] The generated fragment-level sequential signatures are directly bound to anchor records. Anchor records are generated only once at the task level, producing a unique anchor record identifier (anchor ID). At the airport side, a tracking list node is created for each acquired fragment. This node includes at least a fragment identifier (unique fragment ID), a fragment-level sequential signature, an anchor ID (an anchor record reference), and an optional arrival sequence number / arrival time. Subsequently, these linked list nodes are appended to the tracking list in the fragment arrival order to form a linear link. The anchor record itself does not need to embed all fragment-level sequential signatures; these signatures are stored in their respective nodes and associated with the anchor record via the anchor ID. During the appending process, reference consistency is verified. Reference consistency includes at least: the anchor IDs of all linked list nodes are consistent and match the anchor record hash identifier transmitted with the linked list, preventing different task fragments from being mixed in or anchor records from being tampered with. If inconsistencies are found, an anomaly is flagged, triggering retransmission or manual review. After the tracking list is formed, it serves as the transmission carrier relayed from the wireless link to the ground workstation, supporting fragment-by-fragment verification and interval division on the ground side.

[0044] Step S2 accumulates arrival sequence signatures and constructs a tracking list during the airport transfer stage to ensure that the temporal information of the collected segments is bound to the anchor record from the same source. This provides a verifiable and consistent foundation for ground attitude calculation and point cloud fusion, avoids strip alignment deviations caused by out-of-order transmission, and thus maintains benchmark consistency and overall stability of 3D results during multiple re-flights.

[0045] During the stage where the ground workstation receives the original files and tracking lists from the airport relay, after classifying the files according to the task directory in unattended mode, rigorous cross-end verification needs to be performed based on the tracking lists to isolate potential breakpoints and form interval indexes. This ensures that the preprocessing entry point before attitude calculation and point cloud fusion has reliable temporal and attitude continuity, thereby preventing the accumulation of strip alignment deviations and inconsistencies in the re-flight baseline. However, in existing technologies, the ground processing entry point is only triggered by file arrival, ignoring the joint inspection of sequence signatures and attitude references. This leads to the inability to confirm the relationship between inside and outside strips in the early stages, causing inconsistencies in terrain model seams and limited overall stability. Therefore, this step establishes interval indexes through reading, verification, division, and calibration operations, pre-isolates breakpoints, and provides a consistent preprocessing foundation for subsequent splicing decisions and result solidification. This ensures that the 3D model achieves baseline continuity and significantly improves comparability in multiple re-flights.

[0046] Step S3 may include the following sub-steps: S3.1 Read the traced list and verify the timing relationship of the segments.

[0047] After receiving the tracking list from the wireless link, the ground station first reads the contents of each node, including the segment identifier, segment-level sequential signature, and anchor record reference, ensuring that the reading process traverses all nodes in the list to cover all acquired segments. Then, it confirms the segment timing relationship based on the segment-level sequential signature: comparing the arrival order reflected by the segment identifier and sequential signature of adjacent nodes. If they match, the adjacent segment pair is marked "timing consistent"; otherwise, it is marked "timing inconsistent." After verification, the labeling results of the above adjacent segment pairs are summarized to form a timing consistency label list, which serves as input for subsequent attitude continuity inspection and interval division.

[0048] S3.2 Examine the pose continuity of adjacent segments.

[0049] Based on the temporal consistency label list, the attitude continuity of adjacent segments is further examined according to the attitude reference of the anchor record. First, the Euler angle vector of the attitude orientation is extracted from the anchor record reference. Then, the vector difference of the attitude orientation between adjacent acquired segments is calculated, and a threshold such as 0.1 radians is set to determine whether it is continuous. If the difference is lower than the threshold, it is considered as attitude continuity; otherwise, it is marked as a potential breakpoint. The examination process is performed on each segment pair to ensure that all adjacent relationships in the chain are covered, and a list of attitude jump positions is recorded to quantify the degree of continuity. The attitude jump position list refers to the boundary index / segment pair position at the boundary of adjacent segments in the tracking chain where the attitude orientation difference exceeds the threshold and is marked as a potential breakpoint. After the attitude continuity examination is completed, these jump position lists are merged with the temporal consistency label list to form a joint consistency evaluation dataset, which facilitates the accurate merging of breakpoint intervals.

[0050] S3.3 Divide the available intervals and the breakpoint intervals.

[0051] Using the joint consistency evaluation dataset, the fragment sequences in the tracking linked list are partitioned. This involves scanning the dataset to find clusters of consecutive fragments that simultaneously satisfy temporal consistency of sequential signatures and pose continuity (the minimum size of a consecutive fragment cluster is 1). These clusters are assigned to usable intervals, indicating that the data within these intervals can be directly used for point cloud fusion. Specifically, when a consecutive fragment cluster contains only a single fragment, if that fragment and its adjacent fragments (if any) simultaneously satisfy both temporal consistency and pose continuity, then that single fragment is designated as a usable interval of length 1; otherwise, it is merged into an adjacent breakpoint interval. For fragments that do not satisfy either consistency requirement, adjacent inconsistent fragments are merged to form breakpoint intervals, ensuring that the merging logic follows the linked list order to avoid fragmentation. The partitioning uses a dynamic programming algorithm to optimize the interval boundaries, which determines the optimal partitioning point by minimizing the merging cost of inconsistent fragments. After partitioning, the lists of usable intervals and breakpoint intervals are established, providing a structured basis for determining the start and end positions.

[0052] S3.4 Define the interval and generate the interval index.

[0053] Based on the divided list of available intervals and breakpoint intervals, the start and end positions and the segment to which each interval belongs are calibrated. First, a unique interval ID is assigned, and then the start and end positions are recorded, i.e., the segment identifiers of the start and end nodes in the tracking list, as well as the complete list of segments covered by the interval (the set of all segment identifiers arranged in the order of the linked list). Subsequently, this calibration information is organized into an interval index structure for easy querying and retrieval, including the interval ID, start and end positions, segment to which it belongs, and consistency type. Among them, the consistency type includes at least: temporally consistent and pose continuous (available interval); temporally inconsistent but pose continuous; temporally consistent but pose discontinuous; and temporally inconsistent and pose discontinuous (breakpoint interval). After the calibration and index are formed, they serve as the transmission carrier for preprocessing entry, supporting subsequent breakpoint parameter calculation and splicing decision.

[0054] Through the above-mentioned advancements, step S3 completes cross-end verification and isolates breakpoints on the ground side, ensuring that the timing and attitude information of the tracking list is transformed into operable interval indexes, providing an isolation basis for attitude angle difference calculation and splicing path selection, avoiding early inconsistencies from affecting the continuity of the strip base surface, thereby enhancing the coherence and reuse potential of the 3D results under unattended conditions.

[0055] During the parameter calculation phase for breakpoint intervals based on interval indexes at the ground workstation, inconsistent areas have been pre-isolated at the preprocessing entry point. At this stage, it is crucial to quantify attitude and temporal deviations and generate a unique splicing decision to select the optimal path for strip splicing. This ensures the homogeneous binding of cross-end attitude references and time bases, preventing issues such as strip base plane offset and the unusability of retracing results. However, existing breakpoint processing entry points lack established continuity criteria, leading to early uncertainty in attitude and time alignment. This results in inconsistencies in terrain model seams and limited comparability of 3D structures. Therefore, this step utilizes metric parameters, mapping confidence, fusion decisions, and path execution operations to achieve evidence-driven splicing. This provides a smooth strip foundation for seam inspection and result consolidation, ensuring a substantial improvement in the accuracy and reusability of 3D modeling in unattended operation.

[0056] Specifically, step S4 may include the following sub-steps: S4.1 Calculate the attitude angle difference by measuring the attitude orientation jump before and after the breakpoint.

[0057] The start and end positions and their corresponding segments of a specific breakpoint interval (specified by the interval ID, which can be traversed sequentially by interval index) are read from the interval index. The attitude orientation Euler angle vector in the anchor record is used as a reference. Then, the attitude orientation Euler angle vectors of adjacent segments before and after the breakpoint are extracted, and the angular distance between each pair of vectors is calculated. This angular distance is the attitude angle difference. To avoid gimbal lock issues, Euler angles are first converted to quaternion representation, and then the cosine value is obtained through the quaternion inner product and then the angle is obtained through the inverse cosine function. This ensures that the attitude angle difference accurately reflects the jump amplitude and is stored in radians. The attitude angle difference calculation process covers all breakpoint positions in each breakpoint interval to avoid missing potential discontinuities. After calculation, the attitude angle difference is directly used as the quantization input for the initial confidence of the subsequent mapped attitude inheritance path.

[0058] S4.2 Compare the degree of sequence misalignment and calculate the order misalignment rate.

[0059] Based on the breakpoint intervals located in the previous step of calculating the attitude angle difference, a list of segment identifiers formed by the actual arrival sequence (i.e., the airport reception sequence) within the breakpoint window is selected and compared with the list of acquisition sequence markers formed by the native acquisition sequence (i.e., the onboard point cloud). Specifically, the two sequences are matched element by element, and the number of misaligned elements (i.e., the number of segments with misaligned positions) is counted. This number is then divided by the total number of segments in the window to obtain a ratio, forming the sequence misalignment rate. This rate value is ensured to represent the temporal reliability from low to high between 0 and 1. The calculation uses a dynamic time warping algorithm to optimize the sequence alignment path. This algorithm handles possible insertion or deletion deviations by constructing a cost matrix and backtracking to the minimum cumulative distance. The sequence misalignment rate is subsequently mapped to the initial confidence level of the sequential replay path, providing another source of evidence for the fusion decision.

[0060] S4.3 Inverse mapping parameters generate initial trust levels and merge them to obtain splicing decisions.

[0061] Using the attitude angle difference and sequence misalignment rate obtained in the first two steps, initial confidence levels for two alternative stitching paths are generated and evidence fusion is performed. The sequential replay path refers to reconstructing and replaying the segment sequence within the breakpoint interval according to the original acquisition sequence markings on the aircraft to restore time / frame sequence consistency. The attitude inheritance path refers to locking the anchor record attitude reference at the breakpoint as the inheritance benchmark, uniformly correcting the attitude orientation of the segments after the breakpoint to ensure continuity with the strip base plane before the breakpoint. The attitude angle difference is inversely mapped to the initial confidence level of the attitude inheritance path using an exponential decay function, for example, the function form is... ,in The initial trust level for the attitude inheritance path, The attenuation constant is set to 1 to control sensitivity. The attitude angle difference indicates that the larger the jump, the lower the confidence level, approaching 0. Similarly, the sequence misalignment rate is inversely mapped to the initial confidence level of the sequential replay path through an exponential decay function, for example, the function has the form: ,in Initial trust level for sequential replay path. The attenuation constant is set to 1. The sequence misalignment rate indicates that the greater the misalignment, the lower the confidence level. Subsequently, the Dempster-Shafer evidence theory is used to fuse the two initial confidence levels, first assigning a basic probability mass function to the pose inheritance path. Uncertainty ,in The framework includes pose inheritance paths and sequential replay paths; a basic probability mass function is assigned to the sequential replay paths. Uncertainty Then, the Dempster combination rule is applied to calculate the fusion quality function for each non-empty subset A. Where K is the conflict factor To resolve conflicting evidence and normalize it, where A, B, and C are all... A subset of; which can be directly obtained within a binary framework: , , To calculate the credibility of each path, the Pignistic probability can be used: , ,in This represents the confidence / probabilistic support of selecting the pose inheritance path as the correct splicing path; This represents the confidence / probabilistic support level for selecting the sequential replay path as the correct concatenation path; the path with higher confidence level is selected as the sole concatenation decision, and paths containing [certain elements] are retained. , , , , , and The judgment record is available for retrospective purposes.

[0062] S4.4 Select a path based on the judgment, perform the splicing, and retain the record.

[0063] Based on the generated splicing decision: If the decision is a sequential replay path, the breakpoint interval segment sequence is reconstructed according to the original acquisition order, and the segment is reordered and fused with point cloud data to restore temporal continuity. If the decision indicates an attitude inheritance path, then the attitude reference recorded at the anchor point is locked as the inheritance benchmark, and the attitude quaternion of the last segment before the breakpoint is calculated. The quaternion of the first segment after the breakpoint Difference quaternions And the orientation of each segment after the breakpoint. Perform correction This ensures the pose before and after the breakpoint is continuous in the same reference frame; simultaneously, overlapping point cloud windows can be selected near the breakpoint, and residual translation can be estimated through iterative nearest point (ICP). And compensate for the point cloud coordinates after the breakpoint, and if necessary... and Smoothing is performed within the window to avoid abrupt changes. The execution process verifies the attitude continuity and temporal consistency of the spliced ​​segments. If any anomalies are found, the process is rolled back and the parameters are recalculated or an alternative path is selected. After splicing is completed, splicing decision records are retained, including attitude angle difference, sequence misalignment rate, initial confidence level, fusion result, path selection, and attitude and translation corrections, forming an audit log to support the traceability and review of the re-entry benchmark.

[0064] Through the step-by-step decision-making of the above sub-steps, step S4 completes the evidence synthesis-driven splicing execution in the breakpoint interval, ensuring the continuity and verification of attitude reference and time benchmark, avoiding the accumulation of misalignment at the seams, and providing a stable and comparable basis for the final strip solidification and reuse of 3D results, thereby significantly enhancing the overall coherence of the model in the intelligent terrain scanning scenario.

[0065] After the ground workstation completes the splicing of the breakpoint sections, the spliced ​​strip has been initially formed. At this point, it is crucial to conduct rigorous joint inspections on the overlapping areas to examine geometric consistency and local deformation, ensuring that the final strip has a stable base surface and can serve as a re-entry benchmark, thereby achieving result solidification and audit trajectory integrity. However, existing technologies do not integrate joint feedback at the result solidification entry point, leading to difficulties in maintaining the strip benchmark and inconsistencies in the 3D structure at the joints. Therefore, this step establishes the final strip and re-entry benchmark through inspection, judgment, and archiving operations, providing reusable results for multiple re-entries in unattended mode, and significantly enhancing the overall stability and comparability of the terrain model.

[0066] Specifically, step S5 may include the following sub-steps: S5.1 Select the overlapping area and check the geometric consistency.

[0067] The ground workstation reads the stitched strip data from the stitching decision record. First, it identifies overlapping areas between adjacent strips by comparing the start and end positions of the segments in the interval index to find spatially overlapping subsets of point clouds, ensuring coverage of all seam locations to avoid omissions. Then, it performs a geometric consistency check on the point clouds within the overlapping areas, specifically comparing the coordinate correspondences of the point clouds on both sides. An iterative nearest-point algorithm is used for point-by-point matching and the average distance deviation is calculated. If the deviation is below a preset threshold (e.g., 0.05 meters), it is considered geometrically consistent; otherwise, a list of deviation locations is recorded. The check process emphasizes global alignment to prevent local deviations from affecting the overall strip base plane. After the geometric consistency check is completed, a consistency report is generated as input for local deformation checks, supporting the comprehensive judgment of subsequent inspection conclusions.

[0068] S5.2 Analyze local deformation and generate test indicators.

[0069] Based on the geometric consistency report generated in the previous step, the analysis of local deformation in the overlapping area continues. First, the curvature features of a subset of the point cloud are extracted. The degree of deformation is quantified by calculating the normal vector of each point and comparing the curvature changes in the neighborhood. Specifically, a principal curvature estimation algorithm is used to process the curvature tensor and solve for eigenvalues ​​to characterize convex and concave deformation. If the rate of change exceeds a threshold, such as 10%, the location of deformation anomalies is marked. Subsequently, the list of geometric consistency deviation locations and the list of deformation anomalies are merged to form a comprehensive list of inspection indicators, ensuring that the indicators cover all dimensions of the overlapping area to reflect potential inconsistencies. After the local deformation analysis is completed, these indicator lists directly drive the generation of inspection conclusions, facilitating the final determination of strips or the labeling of problem sections.

[0070] S5.3 Determine the final stripe and the resumption benchmark based on the inspection results.

[0071] Using a comprehensive list of inspection indicators, the overall evaluation and inspection conclusions of the stitched strips are determined: First, the list is scanned to check whether all indicators meet the consistency threshold. If they pass, the current strip is determined as the final strip. Simultaneously, the strip base plane (e.g., the average altitude plane) and attitude reference are extracted from the final strip as the re-flight baseline, and immediately archived to the mission directory or 3D results library / point cloud strip results library to support the use of the baseline for multiple re-flights. If the inspection fails, the problematic section, i.e., the specific segment range of abnormal indicators, is marked, and the writing of the problematic section into the results library is stopped (i.e., "entry into the library") to prevent unstable results from entering the subsequent production chain. The process verifies the integrity of the archive, ensuring that the re-flight baseline includes references to attitude orientation and the starting frame marker. After the conclusion is determined, the core of the judgment trajectory is formed, supporting the establishment of traceable records.

[0072] S5.4 Mark the problematic sections and establish traceable records.

[0073] Following the inspection conclusions, if any sections fail, the specific location and anomaly type of the problematic section are marked in the task directory, including geometric deviation values ​​and deformation change rates. These markings are stored in XML format for easy auditing and querying. Regardless of pass or fail, a complete log of the entire process from the start of the seam inspection to the conclusion is maintained, including consistency reports, inspection indicator lists, and archived operation records, forming a traceable record as an audit basis. The process uses a blockchain hash chain algorithm to link the track nodes, preventing tampering and supporting review. After the traceable record is established, the termination judgment of the entire process is solidified, ensuring the comparability of results under unattended conditions.

[0074] Through verification, step S5 uses seam inspection as the final check to ensure that the splicing results are transformed into reliable final strips and re-entry benchmarks, avoiding the accumulation of seam misalignment, and providing a complete audit chain for intelligent terrain scanning and 3D modeling, thereby simultaneously improving the accuracy and consistency of the model in the UAV airport scenario.

[0075] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0076] It should be noted that the system of the present invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting a variety of hardware environments and usage requirements.

[0077] Accordingly, this application also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the intelligent terrain scanning and 3D modeling method based on UAV airports and LiDAR as described above.

[0078] Accordingly, this application also provides an electronic device, comprising: one or more processors; a memory for storing one or more programs; and, when the one or more programs are executed by the one or more processors, causing the one or more processors to implement the intelligent terrain scanning and 3D modeling method based on UAV airports and LiDAR as described above. Figure 2 The diagram shown is a hardware structure diagram of any data processing-capable device, including an intelligent terrain scanning and 3D modeling device based on UAV airport and LiDAR, provided in an embodiment of the present invention. Except for... Figure 2 In addition to the processor, memory, and network interface shown, any data processing device in the embodiment may also include other hardware depending on the actual function of the data processing device, which will not be described in detail here.

[0079] Accordingly, this application also provides a computer-readable storage medium storing computer instructions, which, when executed by a processor, implement the intelligent terrain scanning and 3D modeling method based on UAV airports and LiDAR as described above. The computer-readable storage medium can be an internal storage unit of any data processing device as described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be an external storage device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units of any data processing device and external storage devices. The computer-readable storage medium is used to store the computer program and other programs and data required by the data processing device, and can also be used to temporarily store data that has been output or will be output.

[0080] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein.

Claims

1. A method for intelligent terrain scanning and 3D modeling based on UAV airports and lidar, characterized in that, include: S1: Read the inertial navigation attitude reference when the scan starts and generate an anchor record as the starting point for cross-end tracking; S2: During the airport transfer phase, generate an arrival sequence signature for each collected segment and bind the signature to the anchor record to form a tracking list; S3: The ground side performs verification based on the tracking chain, and divides the available interval and breakpoint interval according to the sequential signature and attitude reference continuity to form an interval index; S4: Calculate the attitude angle difference and sequence misalignment rate in the breakpoint interval, generate splicing decision using the evidence synthesis method, and select the sequential replay path or attitude inheritance path to complete the splicing. S5: Conduct joint inspection on the splicing results, determine the final strip and re-entry benchmark based on the inspection results, and establish traceable records.

2. The intelligent terrain scanning and 3D modeling method based on UAV airport and lidar according to claim 1, characterized in that, Step S1 includes: At the start of the scan, read the attitude reference and start time, and embed the timestamp into the attitude data structure; The attitude reference is converted into a three-dimensional vector form represented by Euler angles, the noise is optimized, the attitude orientation is obtained, and a start frame marker containing frame serialization and initial point cloud index is generated for the first point cloud data frame generated by the lidar. Using a hash function, a unique identifier is calculated for the start time, attitude reference, attitude orientation, and start frame marker, forming an anchor record as the starting point for cross-end tracking.

3. The intelligent terrain scanning and 3D modeling method based on UAV airport and lidar according to claim 1, characterized in that, Step S2 includes: Each acquisition segment is processed according to the segment reception order. The acquisition sequence tag in each acquisition segment is read, and the difference between the acquisition sequence tag and the actual arrival position is compared to generate the sequence signature basic data, including the difference sequence and the out-of-order position list. The acquisition segment is a data unit formed by packaging on the machine according to a preset time window or a preset number of frames, and includes at least a continuous point cloud frame sequence, segment identifier, acquisition sequence tag, start and end timestamps, and attitude / position information corresponding to the point cloud frame. Based on the difference sequence, out-of-order features and main out-of-order patterns are extracted, and the main out-of-order patterns are encoded into fragment-level sequential signatures. The fragment-level sequential signature is bound to the anchor record, and the linked list nodes are appended to the tracking linked list in the order of fragment arrival to form a linear link. The tracking linked list includes fragment representations and fragment-level sequential signature-level anchor record references. During the appending process, the reference consistency is verified. The reference consistency includes at least the following: the anchor IDs of all linked list nodes are consistent and match the anchor record hash identifier transmitted with the linked list.

4. The intelligent terrain scanning and 3D modeling method based on UAV airport and lidar according to claim 3, characterized in that, Based on the difference sequence, disorder features and main disorder patterns are extracted, including: The feature extraction method is applied to the difference sequence of the collected segments. Specifically, the disordered features are characterized by calculating the cumulative sum of continuous differences. Principal component analysis (PCA) is used to isolate the main disordered patterns. Specifically, the covariance matrix of the difference sequence is constructed, where each element is the covariance value of each pair of elements in the sequence. Eigenvalue decomposition is performed to solve for the eigenvectors and eigenvalues ​​of the covariance matrix. The eigenvalues ​​are then sorted in descending order to identify the dominant components. The difference sequence is projected onto the first few eigenvectors to separate the main disordered patterns.

5. The intelligent terrain scanning and 3D modeling method based on UAV airport and lidar according to claim 1, characterized in that, Step S3 includes: The ground station receives the tracking list, reads the segment identifier, segment-level sequential signature, and anchor record reference for each node, confirms the segment timing relationship according to the segment-level sequential signature, and generates a timing consistency tag list. Based on the temporal consistency label list, the attitude continuity of adjacent segments is examined by the attitude reference recorded by anchor and the attitude jump position list is recorded. The attitude jump position list is merged with the temporal consistency label list to form a joint consistency evaluation dataset. Using the joint consistency evaluation dataset, the fragment sequences in the tracking chain are divided. A cluster of continuous fragments that simultaneously satisfies the sequential signature temporal consistency and attitude continuity is divided into a usable interval; otherwise, it is divided into a breakpoint interval. For each interval, the start and end positions, the segment to which it belongs, the interval ID, and the consistency type are assigned to form an interval index.

6. The intelligent terrain scanning and 3D modeling method based on UAV airport and lidar according to claim 1, characterized in that, Step S4 includes: The start and end positions and the segments to which a specific breakpoint interval belongs are read from the interval index. The attitude orientation in the anchor record is used as a reference. The attitude orientation of adjacent segments before and after the breakpoint is extracted. The angular distance between vectors is calculated pair by pair to obtain the attitude angular difference of all breakpoint positions in each breakpoint interval. The segment identifier list formed based on the actual arrival sequence within the breakpoint interval is selected and compared with the collection sequence marker list to obtain the sequence misalignment rate; The attitude angle difference is mapped to the initial confidence of the attitude inheritance path, and the sequence misalignment rate is mapped to the initial confidence of the sequence replay path. The two initial confidences are fused and the confidence of each path is calculated. The path with the highest confidence is used as the splicing decision. If the splicing decision is a sequential replay path, the breakpoint interval segment sequence is reconstructed according to the acquisition order. If the decision is an attitude inheritance path, the attitude reference of the anchor record is locked at the breakpoint position as the inheritance benchmark, the attitude orientation of subsequent segments is adjusted to match the previous segment, and the splicing decision record is retained, including attitude angle difference, sequence misalignment rate, initial confidence level, fusion result and path selection. Among them, the sequential replay path refers to reconstructing and replaying the segment sequence of the breakpoint interval according to the original acquisition sequence marking on the machine, and the attitude inheritance path refers to locking the anchor record attitude reference at the breakpoint as the inheritance benchmark, and uniformly correcting the attitude orientation of the segment after the breakpoint to make it continuous with the strip base plane before the breakpoint.

7. The intelligent terrain scanning and 3D modeling method based on UAV airport and lidar according to claim 6, characterized in that, The attitude angle difference is mapped to the initial confidence level of the attitude inheritance path, and the sequence misalignment rate is mapped to the initial confidence level of the sequence replay path. The two initial confidence levels are fused and the confidence level of each path is calculated. The path with the highest confidence level is used as the concatenation decision. Specifically: The attitude angle difference is passed through an exponential decay function. The inverse mapping is the initial trust level of the pose inheritance path, and the order misalignment rate is reduced by an exponential decay function. The inverse mapping is the initial trust level of the sequential replay path; The Dempster-Shafer evidence theory is used to fuse two initial confidence levels to assign a basic probability mass function to the pose inheritance path. Uncertainty ,in The framework includes pose inheritance paths and sequential replay paths; a basic probability mass function is assigned to the sequential replay paths. Uncertainty ; The Dempster combinatorial rule is applied to compute the fusion quality function for each non-empty subset A. Where K is the conflict factor To resolve conflicting evidence and normalize it, where A, B, and C are all... A subset of; The confidence level of each path is calculated using pignamistic probability. , ,in This represents the confidence / probabilistic support of selecting the pose inheritance path as the correct splicing path; This represents the confidence / probabilistic support level of selecting the sequential replay path as the correct concatenation path. , , ; The path with higher credibility is selected as the sole concatenation decision.

8. The intelligent terrain scanning and 3D modeling method based on UAV airport and lidar according to claim 1, characterized in that, Step S5 includes: Read the spliced ​​strip data from the splicing decision record, identify the overlapping areas between adjacent strips, perform geometric consistency inspection on the point cloud within the overlapping areas, and generate a consistency report; Based on the consistency report, local deformation is analyzed in the overlapping area, abnormal deformation locations are marked, and the list of geometric consistency deviation locations and abnormal deformation locations are merged to form a comprehensive inspection index list. Using the comprehensive inspection index list, the spliced ​​strips are inspected as a whole. If the inspection passes, the current strip is determined to be the final strip. The strip base plane and attitude reference are extracted from the final strip as the re-navigation benchmark and archived. If the inspection fails, the problem section is marked and the entry into the database is stopped. The specific location and anomaly type of the problem section are marked in the task directory. Save the judgment trajectory, that is, the entire process log from the start of the joint inspection to the conclusion, including the consistency report, the list of inspection indicators, and the archived operation record.

9. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-8.

10. A computer-readable storage medium storing computer instructions thereon, characterized in that, When executed by the processor, this instruction implements the steps of the method as described in any one of claims 1-8.

Images