Unmanned aerial vehicle real-time position and visibility calculation method based on beidou grid space-time intelligence

Abstract

The application discloses a Beidou grid space-time intelligent-based unmanned aerial vehicle real-time position line-of-sight calculation method, and particularly relates to the field of airborne station-to-station wireless transmission link line-of-sight calculation technology based on Beidou high-precision positioning, which comprises the following steps: in a satellite navigation application system edge computing node, a satellite positioning high-precision positioning message is converted into a horizontal grid position by latitude and longitude, an elevation layer position is converted by height, and a time slice position is converted by Beidou time service, a Beidou space-time grid position number is synthesized, and a station-to-station link position number table is generated according to a wireless transmission relationship in which at least one of a space station or an airborne station moves; the unmanned aerial vehicle high-precision positioning message is converted into a Beidou space-time grid position number, a station-to-station link grid-penetration order record is generated, and a same time slice flip link is identified according to a link response word sequence; a hidden shielding candidate grid is reversely determined from a grid coverage relationship between the flip link and a non-flip link by using a hypergraph cut; and finally, an edge incremental rewriting is performed on an affected line-of-sight matrix.

Description

Technical Field

[0001] This invention relates to the field of line-of-sight calculation technology for airborne inter-station wireless transmission links in BeiDou high-precision positioning, and more specifically, to a method for calculating the real-time location line-of-sight of unmanned aerial vehicles based on BeiDou grid spatiotemporal intelligence. Background Technology

[0002] In low-altitude UAV collaborative flight based on BeiDou satellite positioning, existing processing methods mostly integrate the real-time latitude, longitude, altitude and time information of UAVs into the satellite navigation application platform, and then overlay 3D maps, point cloud models or fixed obstacle models to determine the occlusion of the wireless transmission path between UAVs, and generate link maintenance, collision warning or trajectory adjustment results accordingly. However, in urban building cluster inspections, low-altitude delivery in industrial parks, and emergency rescue formation flights, the communication links between drones need to be updated rapidly with the BeiDou time slice. At the same time, there are also crane boom rotation, temporary scaffolding erection, lifting platform movement, flocks of birds crossing, glass curtain wall reflection, tree canopy penetration, and local rain and fog attenuation. Some of the obstruction factors are not pre-written into the 3D model and are difficult to obtain completely by the airborne perception of a single drone. In this situation, relying solely on static models will continue to classify links that are actually affected by obstruction as visible. Furthermore, it is difficult to distinguish the sources of signal strength changes in a single link from obstruction, reflection, weather attenuation, and electromagnetic interference. If all links of all UAVs are recalculated pairwise, it will increase the update burden at the edge end and delay the rewriting of link status. In practice, this manifests as the wireless links between multiple UAVs decreasing synchronously within the same BeiDou time slice. Existing processing methods still cannot deduce the spatiotemporal grid location of the unmodeled obstruction based on this, resulting in inconsistency between the visibility matrix and the actual link status. The technical problem to be solved by this application is: how to determine the spatiotemporal grid corresponding to the unmodeled hidden occlusion when multiple UAV wireless links experience simultaneous time slice line-of-sight reversal based on BeiDou high-precision spatiotemporal positioning, and to perform edge incremental updates on the line-of-sight relationship of the affected UAVs. Summary of the Invention

[0003] To overcome the aforementioned deficiencies in the prior art, embodiments of the present invention provide a method for calculating the real-time location and line-of-sight of unmanned aerial vehicles (UAVs) based on BeiDou grid spatiotemporal intelligence. This method converts high-precision UAV positioning messages into BeiDou spatiotemporal grid reference numbers, generates inter-station link grid-crossing sequence records, identifies links that flip in the same time slice based on the link response word sequence, and then uses hypergraph cut to reverse-determine hidden occlusion candidate grids from the grid coverage relationship between flipped and non-flipped links. Finally, it performs edge incremental rewriting on the affected line-of-sight matrix to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for calculating the real-time location and line-of-sight of unmanned aerial vehicles (UAVs) based on BeiDou grid spatiotemporal intelligence, comprising: S1. At the edge computing node of the satellite navigation application system, the high-precision positioning message of satellite positioning is converted from latitude and longitude to horizontal grid position, from altitude to elevation layer position, and from BeiDou time synchronization to time synchronization segment position, synthesizing BeiDou spatiotemporal grid position number, and generating an inter-station link position number table according to the wireless transmission relationship of at least one of the space station or airborne station moving. S2. For inter-station links, read the BeiDou spatiotemporal grid position numbers at both ends, project the direct path from both ends onto the subdivided grid, and recursively traverse along the grid interface to form a grid-crossing sequence record. S3. Drive the detection frame from the transmitting airborne station to the receiving airborne station by the grid sequence record, so that the acknowledgment is written back according to the time slice, the received power is subtracted between slices to get the first response word, the retransmission count bit difference is entered into the first response word, and the arrival delay is rearranged according to the grid sequence and then entered to form the link response word sequence. S4. Based on the link response word sequence, execute the sparse group fusion lasso change point detection algorithm. The single link change word is formed by the inter-chip response difference, and the group change word is formed by the difference of the same source link. The fusion lasso term compresses the common change segment and outputs the set of flipped links and the set of unflipped links. S5. In the hypergraph cut algorithm, the grid sequence record is converted into a path hypergraph, the flipped path hyperedge is entered into the cut side, and the unflipped path hyperedge is entered into the retention side. First, the grid that does not cover the flipped path hyperedge is removed, and then the grid is removed again according to the cut result of the retention side. The hidden occlusion candidate grid is output.

[0005] In a preferred embodiment, it further includes: S6. Based on the hidden occlusion candidate grid, trigger the inter-station link to retransmit the probe frame, so that the new response word sequence is written back to the time slice, and the affected link is located by the common grid number, the line-of-sight state is rewritten, and the line-of-sight matrix is ​​output.

[0006] In a preferred embodiment, S1 includes: S1-1. Within the edge computing node of the satellite navigation application system, the source station number of the high-precision positioning message of the satellite positioning is used as the writing index. The positioning longitude is subtracted from the grid origin longitude and then divided by the longitudinal grid distance to obtain the longitudinal integer part. The positioning latitude is subtracted from the grid origin latitude and then divided by the latitudinal grid distance to obtain the latitudinal integer part. The longitudinal integer part and the latitudinal integer part are concatenated to generate the horizontal grid part. S1-2. Based on the horizontal grid position, subtract the elevation datum from the positioning height and divide by the floor height to obtain the elevation integer position. Subtract the start time of the time synchronization segment from the BeiDou time synchronization and divide by the segment length to obtain the time synchronization integer position. Then, write the horizontal grid position, elevation integer position and time synchronization integer position under the source station number name in the order of horizontal grid position, and output the BeiDou spatiotemporal grid position number. S1-3. Based on the BeiDou spatiotemporal grid position number, read the source station number position of the current time slice and the adjacent previous time slice. If the position of the source station number is inconsistent, write it into the mobile station position. Pair the source station number of the mobile station position belonging to the space station or airborne station with another space station or airborne station to generate an inter-station link position number table.

[0007] In a preferred embodiment, S2 includes: S2-1. For inter-station links in the inter-station link tag table, the grid center coordinates of the transmitting end are solved from the BeiDou spatiotemporal grid tag number of the transmitting end, and the grid center coordinates of the receiving end are solved from the BeiDou spatiotemporal grid tag number of the receiving end. The link direction difference word is obtained by subtracting the grid center coordinates of the transmitting end from the grid center coordinates of the receiving end. The positive and negative signs of each coordinate component in the link direction difference word are written into the recursive direction word of the corresponding interface, and the link starting grid record is output. S2-2. Based on the link starting grid record, take the current grid center coordinates as the starting point of this round. First, use the recursive direction word to obtain the coordinates of the next sub-interface. Then, subtract the coordinates of the starting point of this round from the coordinates of the next sub-interface and divide by the same coordinate component in the link direction difference word to obtain the sub-interface arrival step word of this round. When the coordinate component is zero, write the corresponding sub-interface arrival step word as an empty step and stop the corresponding direction recursion. Output the sub-interface exit of this round. S2-3. Rewrite the current grid position number according to the current grid exit interface, so that the current grid position number enters the next grid along the corresponding recursive direction word, and write the next grid position number into the grid exit sequence record under the inter-station link position number until the next grid position number is consistent with the BeiDou spatiotemporal grid position number of the receiving end, and then output the grid exit sequence record.

[0008] In a preferred embodiment, S3 includes: S3-1. Based on the grid sequence record, the transmitting airborne station writes the inter-station link number into the first segment of the probe frame, writes the current time synchronization segment into the second segment of the probe frame, and XORs the first grid number of the grid sequence record bit by bit with the last grid number and writes it into the last segment of the probe frame, and outputs a probe frame carrying the path check word. S3-2. After receiving the probe frame, the receiving airborne station subtracts the received power sample value of the inter-station link in the previous time slot from the received power sample value to form the first response word, and subtracts the retransmission count readback value of the inter-station link in the previous time slot from the retransmission count readback value to form the second response word. The second response word is then shifted to the left according to the bit length of the probe frame tail segment and added to the first response word. The link differential response word is then output. S3-3. Based on the writing order of the path grid in the grid sequence record, the arrival time of the probe frame is reduced by the transmission time of the probe frame to form the arrival delay word. The arrival delay word is then cyclically shifted according to the writing order of the path grid and connected to the link differential response word. The link response word sequence is then output.

[0009] In a preferred embodiment, S4 includes: S4-1. Based on the link response word sequence, locate the response row by the inter-station link position number and the response column by the BeiDou time synchronization position. Subtract the previous response word from the next response word in the same response row to obtain the inter-segment difference word. Subtract the previous link response word from the next link response word under the same airborne station name to obtain the same source difference word. Write the inter-segment difference word and the same source difference word back according to the response row to form a change detection word table. S4-2. In the sparse group fusion lasso change point detection algorithm, candidate change bits are generated column by column starting from the first response column. When the candidate change bit is zero, the inter-slice difference word is retained as a residual word. When the candidate change bit is one, the inter-slice difference word is written as an interpreted word. Then, the same source difference word is subtracted from the candidate change bit in the same column to generate the same source residual word. The candidate change bits in two adjacent columns are XORed to generate the break word. The detection cost word is generated by splicing the residual word with the number of non-zero bits, the same source residual word with the number of non-zero bits, the number of candidate change bits, and the number of non-zero bits.

[0010] In a preferred embodiment, S4 further includes: S4-3. Rewrite the candidate change bits line by line for the change detection word table. After each rewrite, regenerate the detection cost word and retain the candidate change bits with the dictionary order of the detection cost word first. Continue until the candidate change bits are consistent after one round of line-by-line rewriting. Write the inter-station links with a candidate change bit of 1 into the flipped link set and write the inter-station links with a candidate change bit of 0 into the unflipped link set.

[0011] In a preferred embodiment, S5 includes: S5-1. Based on the flipped link set and the non-flipped link set, read the grid order record, write the path grid position number of each inter-station link into the placeholder word according to the global grid order, and write the sequence number of the path grid position number in the grid order record into the same position order word. The placeholder word of the flipped link forms the cutting page row, and the placeholder word of the non-flipped link forms the reserved page row. The grid position with a value of one in the placeholder word generates the path hypergraph vertex. S5-2. In the hypergraph cut algorithm, the candidate cut word is bitwise ANDed with the cut page row to obtain the cut hit word. When the cut hit word is all zero, write 1 in the flipped debt word. When the cut hit word is non-zero, write 0 in the flipped debt word. The candidate cut word is bitwise ANDed with the retained page row to obtain the retained hit word. When the retained hit word is non-zero, write 1 in the retained intrusion word. When the retained hit word is all zero, write 0 in the retained intrusion word. Then, the number of 1s in the flipped debt word, the number of 1s in the retained intrusion word, the number of 1s in the candidate cut word, and the sum of the corresponding order words are concatenated to form the cut cost word.

[0012] In a preferred embodiment, S5 further includes: S5-3. Starting from an empty candidate cut word, execute the candidate bit writing one and recalculate the cut cost word bit by bit according to the vertex position order of the path hypergraph. If the new cut cost word is in the dictionary order first, retain the corresponding candidate cut word. If the new cut cost word is in the dictionary order later, restore the original candidate cut word. Continue until the vertex position order of the path hypergraph is completed in one round and the candidate cut word has not been rewritten. Output the grid bits with a value of one in the candidate cut word as hidden occlusion candidate grids.

[0013] In a preferred embodiment, S6 includes: S6-1. Generate candidate grid placeholders based on hidden occlusion candidate grids. Perform bitwise AND operations between the candidate grid placeholders and the path placeholders of each inter-station link to obtain the link hit word. If there is a bit with a value of 1 in the link hit word, write it into the verification transmission bit and write the first grid bit with a value of 1 into the common grid bit. Output the verification link record. S6-2. Based on the verification link record, the sending airborne station sends a probe frame according to the verification transmission bit, so that the receiving airborne station subtracts the original received power from the new received power to form a verification power difference word, subtracts the original retransmission count from the new retransmission count to form a verification retransmission difference word, subtracts the original arrival delay from the new arrival delay to form a verification delay difference word, and forms a verification response word in the order of writing the verification power difference word, the verification retransmission difference word, and the verification delay difference word; S6-3. Rewrite the line-of-sight matrix according to the verification response word, and generate the verification rewritten word by bitwise XORing the verification response word with the original link response word. If there are inter-station links with a value of 1 in the verification rewritten word, write them into the affected link bits according to the common grid bits. The matrix cells corresponding to the affected link bits are rewritten to the non-line-of-sight state. The matrix cells that are not written into the affected link bits retain the original line-of-sight state. Output the line-of-sight matrix.

[0014] The technical effects and advantages of this invention are as follows: This solution uses cross-grid inversion between flipped and unflipped links in the same time-slice to locate unmodeled hidden occlusion meshes, making the visibility matrix relatively consistent with the actual link state. Binding the BeiDou spatiotemporal grid position number to the positions at both ends of the inter-station link makes the path grid sequence have a common calculation benchmark, reducing the burden of repeated calculation of link path; Link response word compression is performed on received power, retransmission count and arrival delay to more fully express the source of link changes and relatively alleviate misjudgment of single signal strength. In the sparse group fusion lasso change point detection, the common change segment of the same source link is extracted, so that the synchronous flip link is centrally identified and the interference of isolated fluctuations on the occlusion judgment is reduced. The hypergraph cut algorithm is used to distinguish between flipped path coverage and non-flipped path intrusion, making the selection of hidden occlusion candidate meshes more consistent with the common changes of multiple links; By writing back new response words based on the verification probe frame and rewriting the affected links, the line-of-sight matrix is ​​updated incrementally only for the hit links, which relatively reduces the update burden at the edge. Attached Figure Description

[0015] Figure 1 This is a flowchart outlining the method steps of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Refer to the instruction manual appendix Figure 1 The present invention provides a method for calculating the real-time location and line-of-sight of unmanned aerial vehicles (UAVs) based on BeiDou grid spatiotemporal intelligence, comprising: S1. At the edge computing node of the satellite navigation application system, the high-precision positioning message of satellite positioning is converted from latitude and longitude to horizontal grid position, from altitude to elevation layer position, and from BeiDou time synchronization to time synchronization segment position, synthesizing BeiDou spatiotemporal grid position number, and generating an inter-station link position number table according to the wireless transmission relationship of at least one of the space station or airborne station moving. In this embodiment, S1 is used to write the high-precision positioning message generated by satellite positioning into the BeiDou spatiotemporal grid position number under the source station number, and to make the high-precision positioning calculation status participate in the generation of mobile station positions, so as to prevent the mixing of ordinary positioning messages and high-precision positioning messages from affecting the generation of inter-station wireless transmission links; the implementation process includes the following steps: S1-1 is used to convert the planar position in the high-precision positioning message into horizontal grid bits. After receiving the high-precision positioning message, the edge computing node of the satellite navigation application system reads the source station number, positioning solution status, positioning longitude, and positioning latitude, using the source station number as the writing index. When the positioning solution status has a high-precision positioning valid flag, the positioning longitude is subtracted from the grid origin longitude and then divided by the meridional grid distance. The resulting integer quotient is written as the meridional integer digit. The positioning latitude is subtracted from the grid origin latitude and then divided by the latitudinal grid distance. The resulting integer quotient is written as the latitudinal integer digit. The meridional integer digit is concatenated with the latitudinal integer digit first and the latitudinal integer digit last to generate the horizontal grid bits, and the horizontal grid bits are written into the current message buffer corresponding to the source station number. When the positioning solution status does not have a high-precision positioning valid flag, or when the positioning longitude and positioning latitude are missing, the edge computing node reads the horizontal grid bits of the previous time slice of the same source station number and writes them into the high-precision positioning carry-over bits. Then, the high-precision positioning carry-over bits along with the horizontal grid bits are passed to S1-2. S1-2 is used to continuously write the horizontal grid bits into the height layer and time slice, generating the BeiDou spatiotemporal grid bit number under the source station name; the edge computing node uses the horizontal grid bits output by S1-1 as the starting field to read the positioning altitude and BeiDou timing from the high-precision positioning message; the positioning altitude is subtracted from the elevation datum and divided by the layer height, and the resulting integer is written as the elevation integer; the BeiDou timing is subtracted from the timing slice start time and divided by the slice length, and the resulting integer is written as the timing integer. The timing slice start time adopts the mission start timing or the BeiDou cycle second start time. The segment length is written into the configuration table based on the inter-station link detection cycle; when crossing BeiDou cycles, the edge computing node first connects the BeiDou cycle number to the BeiDou timing, and then performs timing subtraction; the horizontal grid bits, elevation integer bits, timing integer bits, and high-precision positioning carry-over bits are written into the source station number name in a fixed order to form the BeiDou spatiotemporal grid bit number; when the positioning height is missing, the elevation integer bits of the previous timing segment of the same source station number are used and the height carry-over bits are written; when BeiDou timing is missing, the generation of the BeiDou spatiotemporal grid bit number corresponding to the current message is stopped and a timing missing marker is written. S1-3 is used to identify mobile stations and generate an inter-station link tag table; edge computing nodes read the BeiDou spatiotemporal grid tag number of each source station number in the current time slice, and read the BeiDou spatiotemporal grid tag number of the same source station number in an adjacent previous time slice; if any field of the horizontal grid tag number, elevation layer tag number, or time slice tag number in the current time slice and the adjacent previous time slice is different, and the current time slice has not written the high-precision positioning carry-over tag number, the source station number is written into the mobile station tag number; if the current time slice has written the high-precision positioning carry-over tag number, only the mobile station tag number of the previous time slice is retained; the first occurrence of the source station number is written into the initial mobile station tag number; and so on. Then, the station type registration record is read, the source station number corresponding to the UAV is identified as the airborne station, and the source station number participating in the wireless relay or telemetry and control link is identified as the space station. When the mobile station belongs to the space station or the airborne station, it is concatenated with the source station number of another space station or airborne station in a fixed order of the transmitting end source station number, the receiving end source station number, and the timing slice number to generate the inter-station link number, and written into the inter-station link number table. When the same pair of source station numbers has reverse wireless transmission, the inter-station link number is generated separately according to the transmission direction. When the same transmission direction is repeatedly generated in the same timing slice, the old record is overwritten and the latest Beidou spatiotemporal grid number is retained. The above implementation process enables high-precision positioning to not only participate in grid conversion as a location source, but also participate in the generation of the mobile station position in the current time slice; when the high-precision positioning is valid, a new BeiDou spatiotemporal grid position number is generated; when the high-precision positioning is invalid, the position of the previous time slice is used and the source is marked. Subsequently, S2 reads the BeiDou spatiotemporal grid position numbers at both ends of the link and calculates the wireless direct path accordingly. In practical applications: Three drones in the park perform inspections as airborne stations. After the edge computing node receives three high-precision positioning messages in the same BeiDou time slice, it converts the messages with the high-precision positioning validity mark into new horizontal grid bits, elevation integer bits, and time integer bits. One of the drones enters the adjacent horizontal grid relative to the previous time slice, and the source station number is written into the mobile station position. Then, it generates an inter-station link bit number with the source station number of another airborne station according to the sending direction. S2 reads the BeiDou spatiotemporal grid bit numbers at both ends and calculates the wireless direct path.

[0018] S2. For inter-station links, read the BeiDou spatiotemporal grid position numbers at both ends, project the direct path from both ends onto the subdivided grid, and recursively traverse along the grid interface to form a grid-crossing sequence record. In this embodiment, S2 is used to convert the BeiDou spatiotemporal grid reference numbers at both ends of the inter-station link into grid center coordinates in the partitioned grid, and to form a grid crossing sequence record according to the order in which the wireless direct path passes through the grid interface, so that subsequent detection frames can generate path check words and link response word sequences along the actual grid crossing path of the inter-station link; the partitioned grid is formed by removing the timing slice position from the BeiDou spatiotemporal grid reference number, and the horizontal grid position and the elevation layer position jointly determine the three-dimensional spatial grid position, and the timing slice position is limited only to this batch of inter-station link calculations; the implementation process includes the following steps: S2-1 is used to inversely solve the BeiDou spatiotemporal grid reference numbers at both ends of the inter-station link into the grid center coordinates in the same local rectangular coordinate system, and generate the link direction difference word that controls the subsequent recursive direction; the edge computing node reads the inter-station link reference number from the inter-station link reference number table, and reads the BeiDou spatiotemporal grid reference number of the transmitting end and the BeiDou spatiotemporal grid reference number of the receiving end according to the inter-station link reference number; for the BeiDou spatiotemporal grid reference number of the transmitting end, the edge computing node reads the meridional integer bits, the latitudinal integer bits, and the elevation integer bits, so that... The longitude of the transmitting end grid center is obtained by multiplying the integer part of the meridional direction by the meridional grid distance and adding half the meridional grid distance. The latitude of the transmitting end grid center is obtained by multiplying the integer part of the latitudinal direction by the latitudinal grid distance and adding half the latitudinal grid distance. The height of the transmitting end grid center is obtained by multiplying the integer part of the elevation by the layer height and adding half the layer height. The longitude, latitude, and height of the transmitting end grid center are then converted into the coordinates of the transmitting end grid center. For the BeiDou spatiotemporal grid position number of the receiving end, the edge computing nodes obtain the coordinates of the receiving end grid center using the same conversion method. Subtracting the transmitter's grid center coordinates from the receiver's grid center coordinates yields the link direction difference word. When the meridional component in the link direction difference word is positive, a positive meridional recursive direction word is written; when the meridional component is negative, a negative meridional recursive direction word is written; and when the meridional component is zero, a zero meridional recursive direction word is written. The latitudinal and elevation components are written into the latitudinal and elevation recursive direction words according to the same rule. The final output is a link start grid record containing the transmitter's grid number, the receiver's grid number, the transmitter's grid center coordinates, the receiver's grid center coordinates, and the link direction difference word. If the transmitter's BeiDou spatiotemporal grid number or the receiver's BeiDou spatiotemporal grid number has a timing missing marker, the edge computing node writes the link start missing marker and stops the grid recursion of this inter-station link. S2-2 is used to calculate the step position of the wireless direct path to the next grid interface within the current grid and determine the interface crossing point in this round. The edge computing node uses the transmitter grid number in the link starting grid record as the current grid number, and uses the current grid number to solve for the current grid center coordinates, and uses the current grid center coordinates as the starting point of this round. When the recursion direction word is a positive recursion direction word, the coordinates of the next interface are taken from the positive boundary of the current grid. When the recursion direction word is a negative recursion direction word, the coordinates of the next interface are taken from the negative boundary of the current grid. When the recursion direction word is a zero recursion direction word, the interface arrival step word of the corresponding direction is written as a blank step and the direction stops participating in the comparison in this round. For directions not marked as empty steps, the edge computing node subtracts the coordinates of the starting point of the current round from the coordinates of the next interface, and then divides by the same coordinate component in the link direction difference word to obtain the interface arrival step word for the corresponding direction. The edge computing node compares the arrival step words of the meridional interface, the arrival step words of the latitudinal interface, and the arrival step words of the elevation interface, and takes the interface with the first value as the exit interface of the current round. If the arrival step words of the interfaces of two or three directions are the same, the edge computing node writes multiple interfaces corresponding to the same arrival step word together as the exit interface of the current round, so that multiple directions are recursively advanced when the direct path crosses the grid edge or grid corner. If all interface arrival step words are empty steps, the edge computing node writes a path stagnation mark and stops the grid-crossing recursion of this inter-station link. S2-3 is used to rewrite the current grid position number based on the current exit interface and write the recursively obtained path grid into the grid sequence record; the edge computing node reads the recursive direction word corresponding to the current exit interface and rewrites the current grid position number to the next grid position number along the corresponding recursive direction word; when the current exit interface contains multiple directions, the current grid position number is rewritten simultaneously in multiple directions to prevent the path from being repeatedly written to the same grid when passing through grid edges or grid corners; the edge computing node writes the next grid position number, entry surface, exit surface, and link sequence into the inter-station link position number name. The next grid position is recorded, and the next grid position is solved to obtain the next grid center coordinates, which is used as the starting point for the next round of calculation to continue S2-2; when the next grid position is the same as the BeiDou spatiotemporal grid position of the receiving end, the edge computing node writes the recursion termination bit and outputs the grid position record formed bit by bit from the sending end grid to the receiving end grid; when the next grid position exceeds the spatial enclosed range formed by the sending end grid center coordinates and the receiving end grid center coordinates, the edge computing node writes the path cross-boundary mark and stops the recursion. The inter-station link with the path cross-boundary mark does not generate a probe frame in S3; Through the above implementation process, the BeiDou spatiotemporal grid position numbers at both ends of the inter-station link are converted into wireless direct path grid traversal sequence records in the subdivided grid. The transmitting end grid, the intermediate grid of the path, and the receiving end grid are written bit by bit according to the wireless transmission direction. Zero components, grid edge crossings, grid corner crossings, path stalls, and path boundary crossings are all marked with corresponding tags for subsequent steps to read. In practical applications: Two airborne stations are located in different BeiDou spatiotemporal grid positions on both sides of the building. The edge computing node first deduces the grid center coordinates of the transmitting end from the BeiDou spatiotemporal grid position, and then deduces the grid center coordinates of the receiving end from the BeiDou spatiotemporal grid position. Subsequently, it calculates the step word of the direct path to the grid interface near the building according to the link direction difference word. When the path passes through the grid edge of the building corner, the longitudinal and latitudinal directions are simultaneously recursively deduced. Finally, the grid crossing sequence record is read by S3 and used to generate the path check word of the detection frame.

[0019] S3. Drive the detection frame from the transmitting airborne station to the receiving airborne station by the grid sequence record, so that the acknowledgment is written back according to the time slice, the received power is subtracted between slices to get the first response word, the retransmission count bit difference is entered into the first response word, and the arrival delay is rearranged according to the grid sequence and then entered to form the link response word sequence. In this embodiment, S3 is used to convert the grid sequence record formed in S2 into an inter-station link probe frame, and rewrite the wireless transmission result written back by the receiver's airborne station into a link response word sequence, so that S4 can read the changes in received power, retransmission count, and arrival delay from the same inter-station link position and the same timing slot position. The field word length of the probe frame is given by the frame format configuration table in the edge computing node of the satellite navigation application system. The inter-station link position, the current timing slot position, and the path check word are all written with a fixed word length. If the field is less than the fixed word length, zeros are added from the high bits; if the field exceeds the fixed word length, the low bits are truncated. This implementation process includes the following steps: S3-1 is used to generate a probe frame with a path check word based on the grid sequence record, so that the probe frame can carry the endpoint grid relationship of the wireless direct path. The transmitting airborne station reads the inter-station link number from the inter-station link number table and reads the first grid number and the last grid number from the grid sequence record. Before the first grid number and the last grid number are XORed, they are first aligned according to the length of the path check field. The short side is padded with zeros from the high bits, and the long side retains the low bits. Then, they are XORed bit by bit to form the path check word. The transmitting airborne station writes the inter-station link number into the first segment of the probe frame, writes the current time slot number into the second segment of the probe frame, writes the path check word into the last segment of the probe frame, and writes the probe frame transmission time into the frame header buffer. When the grid sequence record has a path out-of-bounds mark or a path stall mark, the transmitting airborne station does not generate a probe frame and writes a probe skip mark under the inter-station link number name so that S4 can ignore the corresponding response position of this time slot. S3-2 is used to convert the radio reception results written back by the receiving airborne station into a link differential response word, so that the changes in received power and retransmission count are included under the same inter-station link tag name; after the receiving airborne station completes the demodulation of the probe frame, it reads the received power sample value from the RF receiving channel and writes the received power sample value into the acknowledgment buffer according to the inter-station link tag name; the transmitting airborne station reads the retransmission count readback value of the same inter-station link tag name from the retransmission control buffer and writes it back to the edge computing node along with the probe acknowledgment; the edge computing node reads the received power sample value of the same inter-station link in the previous timing slot, so that the current receiving power changes and retransmission count changes are included under the same inter-station link tag name. The first response word is formed by subtracting the received power sample value from the received power sample value of the previous time slot. Then, the retransmission count readback value of the inter-site link of the previous time slot is read, and the current retransmission count readback value is subtracted from the retransmission count readback value of the previous time slot to form the second response word. The second response word is then shifted to the left by the bit length of the probe frame tail segment and added to the first response word, and the link differential response word is output. When the received power sample value or retransmission count readback value of the inter-site link of the previous time slot is missing, the edge computing node writes the current time slot data into the initialization buffer and writes the first response word or the second response word to zero, and writes the initialization flag at the same time. S3-3 is used to rewrite the arrival delay of the probe frame into a sequence-related response according to the grid order record, so that the link differential response word carries the influence of path length and grid order. When the probe frame demodulation is completed, the receiving airborne station writes the arrival time of the probe frame according to BeiDou time synchronization. The edge computing node reads the transmission time of the probe frame from the frame header buffer of the sending end, and subtracts the transmission time of the probe frame from the arrival time to form the arrival delay word. The edge computing node reads the number of path grids in the grid order record, divides the number of path grids by the length of the link differential response word field, and takes the remainder as the path grid number. The arrival delay word is cyclically shifted, and the shifted-out bits are filled in from the other end. The shifted arrival delay word is connected to the link differential response word, the link response word sequence is output, and it is written to the response buffer according to the inter-station link number and the current timing chip. When the probe receipt is not returned, the edge computing node writes the received power sample value into the received null value, writes the retransmission count readback value into the actual retransmission readback value of the transmitting end, writes the arrival delay word into the non-arrival code, and still generates the link response word sequence according to the same field order for S4 to identify link changes. Through the above implementation process, the probe frame is not only used to confirm whether the transmitting airborne station and the receiving airborne station have received the acknowledgment, but also writes the inter-station link number, the current time slot number, the path check word, the received power change, the retransmission count change and the arrival delay change into the same link response word sequence, so that S4 can read the wireless transmission changes of the same time slot under the inter-station link number name. In practical applications: the grid-crossing sequence record between two airborne stations passes through the building edge grid. The transmitting airborne station writes the inter-station link number into the first segment of the probe frame, writes the current time synchronization segment into the second segment of the probe frame, and XORs the first and last grid numbers to generate a path check word, which is then written into the last segment of the probe frame. The receiving airborne station writes back the received power sample value and arrival time, and the transmitting airborne station writes back the retransmission count readback value. The edge computing node then forms a link response word sequence, which S4 uses to determine whether the inter-station link in this time synchronization segment has flipped.

[0020] S4. Based on the link response word sequence, execute the sparse group fusion lasso change point detection algorithm. The single link change word is formed by the inter-chip response difference, and the group change word is formed by the difference of the same source link. The fusion lasso term compresses the common change segment and outputs the set of flipped links and the set of unflipped links. In this embodiment, S4 is used to convert the link response word sequence output by S3 into a change detection word table that can represent the changes in the state of inter-station wireless transmission links. Then, the sparse group fusion lasso change point detection algorithm is used to compress single link changes and common changes of the same source links into a shared change segment, thereby generating a set of flipped links and a set of non-flipped links. This implementation process includes the following steps: S4-1 is used to arrange the link response word sequence into a change detection word table, so that the changes of each inter-station link within adjacent BeiDou timing slices and the changes between links under the same airborne station name form fields that can participate in the detection cost calculation. Specifically, the edge computing node reads the link response word sequence written in S3, locates the response row by the inter-station link position number, locates the response column by the BeiDou timing slice position, and subtracts the link response word of the previous response column from the link response word of the subsequent response column in the same response row to obtain the inter-slice difference word; then it reads the inter-station link... For the transmitting airborne station positions in the path position number, the inter-station links under the same transmitting airborne station name are arranged in lexicographical order of the inter-station link position number, and the response words of the links arranged later are subtracted from the response words of the links arranged earlier to obtain the same source difference word; when there is only one inter-station link under the same transmitting airborne station name, the same source difference word is written as zero; when there is an initialization flag in the current response column, the inter-chip difference word is written as zero and the initialization flag is retained; when there is a probe skip flag in the current response column, the current inter-station link does not participate in this round of detection, and the change detection word table with skip bit is output; S4-2 is used to interpret changes in the change detection word table as candidate change bits, and write uninterpreted changes, inconsistent changes from the same source, change quantity, and change break as detection cost words. Specifically, the edge computing node generates candidate change bits column by column starting from the first response column. When a candidate change bit is zero, the inter-slice difference word of the current response column is not interpreted and is written into the residual word. When a candidate change bit is one, the inter-slice difference word of the current response column is written into the interpreted word, and the corresponding position of the residual word is written as zero. Then, the candidate change bits in the same response column are subtracted from the same source difference word to obtain the same source residual. The word is used to represent the remaining part of the link change under the same transmitting end airborne station name that has not been jointly interpreted; then the candidate change bits of two adjacent columns are XORed bit by bit to obtain the break word; the edge computing nodes respectively count the number of non-zero bits of the residual word, the number of non-zero bits of the same source residual word, the number of one in the candidate change bits, and the number of non-zero bits of the break word, and concatenate them into the detection cost word in this order; when there is an unreachable code in the current response column, the unreachable code is used as the link response word to participate in the inter-slice difference word calculation; when there is an initialization flag in the current response column, the candidate change bit corresponding to the initialization flag is fixed to zero; S4-3 is used to converge the detection cost word by rewriting the candidate change bits line by line, and output the flipped link set and the unflipped link set accordingly. Specifically, the edge computing node takes the first response row of the change detection word table as the starting row, performs zero-to-one or one-to-zero operations on the candidate change bits of the current response row, and regenerates the residual word, homologous residual word, broken word and detection cost word after rewriting. When the new detection cost word is arranged in lexicographical order before the original detection cost word, the candidate change bit rewriting result of the current response row is retained. When the new detection cost word is arranged in lexicographical order after the original detection cost word, the candidate change bits before the current response row is rewritten are restored. After the response row is executed line by line, if the candidate change bits in this round are the same as the candidate change bits in the previous round, the rewriting is stopped and the shared change segment is output. When the candidate change bit is one and the corresponding inter-station link has a link response word in the current response column, the inter-station link is written into the flipped link set. When the candidate change bit is zero or the candidate change bit is one but the current response column has an initialization mark, the inter-station link is written into the unflipped link set. Through the above implementation process, the changes in received power, retransmission count, and arrival delay in the link response word sequence are converted into inter-segment difference words and same-source difference words. The sparse group fusion lasso change point detection algorithm does not rely on the instantaneous fluctuations of a single wireless transmission link, but compresses the inter-station link changes into shared change segments by detecting the round-by-round rewriting of the cost words, providing the flipped link set and the non-flipped link set for S5 to construct the path hypergraph. In practical applications: When the same transmitting airborne station sends probe frames to two receiving airborne stations at the same time, building edge occlusion causes the two inter-station links to experience a decrease in receiving power and an increase in arrival delay in the same BeiDou timing slice. The edge computing node first generates a change detection word table from the link response word sequence, then obtains the shared change segment by rewriting the candidate change bits, and finally writes the two synchronously changing inter-station links into the flipped link set, and writes the inter-station links that do not show synchronous changes into the non-flipped link set.

[0021] S5. In the hypergraph cut algorithm, the grid sequence record is transcribed into the path hypergraph, the flipped path hyperedge is brought into the cutting side, and the unflipped path hyperedge is brought into the retention side. First, the grid that does not cover the flipped path hyperedge is removed, and then the grid is removed again according to the cutting result of the retention side. The hidden occlusion candidate grid is output. In this embodiment, S5 is used to convert the flipped link set and the non-flipped link set output by S4 into a path hypergraph, and to perform bit-by-bit convergence between the flipped link coverage relationship and the non-flipped link intrusion relationship through candidate cut words to obtain hidden occlusion candidate grids that trigger the synchronous flipping of multiple inter-station wireless transmission links; S5 does not recalculate the wireless direct path, but reads the grid passing sequence record formed by S2 and the link grouping result formed by S4, writes the link passing through the grid as placeholder words, and then uses the hypergraph cut algorithm to screen out the grid positions that simultaneously interpret the flipped link and do not intrude on the non-flipped link; this implementation process includes the following steps: S5-1 is used to convert grid-based sequence records into a path hypergraph, ensuring that inter-station links have the same bit order representation across the grid. Edge computing nodes read the inter-station link positions from both the flipped and unflipped link sets, and then read back the corresponding grid-based sequence records according to these positions. After deduplicating all path grid positions that have appeared in the current time slice, the edge computing nodes generate a global grid order based on the lexicographical order of the BeiDou spatiotemporal grid positions. When the path grid position of each inter-station link matches the global grid order, a placeholder word corresponds to the grid. Write 1 for each position and 0 for each position that is not hit. The length of the placeholder is equal to the number of grid positions in the global grid order. The path grid position number is written in the same position order word in the grid order record. The placeholder for the flipped link is written in the cutting page line according to the inter-station link position number. The placeholder for the non-flipped link is written in the reserved page line according to the inter-station link position number. The grid position with a value of 1 in the placeholder generates the path hypergraph vertex. When the grid order record has a path out-of-bounds marker, the corresponding inter-station link is not written in the cutting page line and the reserved page line, and an out-of-bounds removal marker is written next to the path hypergraph. S5-2 is used to calculate the interpretability of candidate cut words for flipped links and the degree of intrusion into non-flipped links, and writes both as cut cost words; edge computing nodes generate candidate cut words in global grid order, and the grid bits with a value of 1 in the candidate cut word represent the candidate occlusion grid selected in this round; the candidate cut word is bitwise ANDed with each cut page row to obtain the cut hit word for the corresponding flipped link. When the cut hit word is all zeros, it means that the candidate cut word in this round did not hit the grid passed by the corresponding flipped link, and the corresponding bit of the flipped page row is written as 1; when the cut hit word has a value of 1, the corresponding bit of the flipped page row is written as 0; the candidate cut word is ANDed with each page row to obtain the cut hit word for the corresponding flipped link. The corresponding unflipped link is obtained by performing a bitwise AND operation on each row of the remaining ledger page. If a single retained hit word exists, it means that the candidate cutting word has entered the unflipped link through the grid in this round, and the corresponding bit of the retained intrusion word is written as 1. If all bits of the retained hit word are zero, the corresponding bit of the retained intrusion word is written as zero. The edge computing node counts the number of 1s in the flipped ledger words, the number of 1s in the retained intrusion words, and the number of 1s in the candidate cutting words. Then, it reads the corresponding sequential word of the grid bit of the candidate cutting word and accumulates them. The four results are concatenated in order to form the cutting cost word. When the cutting ledger page is empty, an empty flip mark is written. When the remaining ledger page is empty, all bits of the retained intrusion word are written as zero. S5-3 is used to compare the candidate cutting words with the cutting cost words to ensure that the hidden occlusion candidate mesh converges and is output within a finite number of rounds. The edge computing node starts from the empty candidate cutting word and writes the candidate bits one by one according to the vertex position order of the path hypergraph. After each candidate bit is written, S5-2 is executed again to generate a new cutting cost word. When the new cutting cost word is arranged in lexicographical order before the original cutting cost word, the candidate cutting word after writing is retained. When the new cutting cost word is arranged in lexicographical order after the original cutting cost word, the candidate cutting word before writing is restored. After the global mesh order completes a round of trial writing, if the candidate cutting word has not been rewritten relative to the beginning of the round, the trial writing stops and the mesh bits with a value of one in the candidate cutting word are output as hidden occlusion candidate meshes. If the candidate cutting word has been rewritten, the rewritten candidate cutting word is used to enter the next round of trial writing. When the candidate cutting word is all zero and has not been rewritten after a round of trial writing, the edge computing node writes an empty candidate mark. The empty candidate mark is used by S6 to skip the verification sending bit generation and continue the original line-of-sight state. Through the above implementation process, the flipped link set and the non-flipped link set are transformed into cutting constraints in the path hypergraph. The hidden occlusion candidate grid is generated by the combined constraints of the flipped debt word, the retained intrusive word, the number of candidate cutting words, and the same position order word, so that S6 can directly read the hidden occlusion candidate grid and trigger the verification detection frame of the hit inter-station link. In practical applications: the inter-station links between the three airborne stations flip during the same timing slice. Two of the flipped links pass through the grid near the building corner, while one non-flipped link bypasses the grid near the building corner. After the edge computing node writes the grid crossing sequence record of the three inter-station links as placeholder words, when the candidate cutting word hits the grid near the building corner, the number of flipped debt words decreases, the number of retained intrusion words remains zero, the cutting cost words retain the grid near the building corner in lexicographical order, and finally the grid near the building corner is output as the hidden occlusion candidate grid.

[0022] S6. Based on the hidden occlusion candidate grid, trigger the inter-station link to retransmit the probe frame, so that the new response word sequence is written back to the time slice, and the affected link is located by the common grid number, the line-of-sight status is rewritten, and the line-of-sight matrix is ​​output. In this embodiment, S6 is used to perform inter-station link verification on the hidden occlusion candidate grid output by S5, and write the verification result into the line-of-sight matrix, so that the hidden occlusion candidate grid does not directly replace the wireless link state, but completes the location of the affected link and the rewriting of the matrix unit by verifying the new response word sequence corresponding to the probe frame; the implementation process includes the following steps: S6-1 is used to convert hidden occlusion candidate grids into verification transmission ranges and determine the common grid bits between inter-station links and hidden occlusion candidate grids. The edge computing node reads the hidden occlusion candidate grids output by S5 and generates candidate grid placeholders using the global grid order formed by S5. When a hidden occlusion candidate grid is hit in the global grid order, the corresponding bit is written as 1, and the remaining bits are written as zero. The edge computing node reads the path placeholders of each inter-station link from S5, does not regenerate the bit order, and performs a bitwise AND operation between the candidate grid placeholders and the path placeholders to obtain the link hit word. When there are bits with a value of 1 in the link hit word, a verification transmission bit is written under the inter-station link bit name, and the first grid bit with a value of 1 in the global grid order is written to the common grid bit, and the remaining grid bits with a value of 1 are written to the verification auxiliary bit. When the link hit word is all zeros, no verification transmission bit is written. When a hidden occlusion candidate grid has an empty candidate marker, the edge computing node stops generating verification link records and makes the line-of-sight matrix retain the original line-of-sight state of the current time slice. S6-2 is used to re-obtain the inter-station wireless transmission results based on the verification link record and form a verification response word that can be compared with the original link response word sequence; the edge computing node reads the verification transmission bit in the verification link record and drives the transmitting airborne station to resend the probe frame according to the inter-station link bit number; after the receiving airborne station completes the demodulation of the verification probe frame, it writes back the new received power and new arrival delay, and the transmitting airborne station writes back the new retransmission count; the edge computing node reads the original received power under the same inter-station link bit number and the same timing chip bit from the corresponding buffer of S3. The original retransmission count and original arrival delay are used to subtract the original received power from the new received power to form a verification power difference word, subtract the original retransmission count from the new retransmission count to form a verification retransmission difference word, and subtract the original arrival delay from the new arrival delay to form a verification delay difference word. The verification response word is formed in the order of writing the verification power difference word, verification retransmission difference word, and verification delay difference word. If the inter-station link bit number or timing chip bit in the verification probe frame receipt is different from the original link response word sequence buffer, the edge computing node writes a verification misalignment mark and stops the matrix rewriting of this inter-station link. S6-3 is used to locate the affected link position based on the verification response word and rewrite the line-of-sight matrix according to the inter-station link direction. The edge computing node reads the verification response word and the original link response word formed by S3, and performs a bitwise XOR operation between the verification response word and the original link response word to generate the verification rewritten word. When the verification rewritten word has a value of 1 and a common grid position is written under the same inter-station link position name, the edge computing node writes the inter-station link position number into the affected link position and locates the line-of-sight matrix unit according to the direction from the transmitting airborne station to the receiving airborne station, and rewrites the corresponding matrix unit to a non-line-of-sight state. For inter-station links with all zeros in the verification rewritten word, all zeros in the link hit word, or with verification misalignment marks, no affected link position is written, and the corresponding matrix unit maintains its original line-of-sight state. For reverse wireless transmission, the other matrix unit is located by the reverse inter-station link position number. If it is necessary to output bidirectional results to the display interface, the edge computing node generates a symmetrical display table according to the same source station pairing after the line-of-sight matrix is ​​output, without changing the original direction record in the line-of-sight matrix. Through the above implementation process, the hidden occlusion candidate grid is first converted into candidate grid placeholders, and then compared with the path placeholders. The common grid position limits the verification detection range. The bit-by-bit difference between the verification response word and the original link response word is used to determine the affected link position. Finally, it is written into the line-of-sight matrix according to the inter-station link direction. In practical applications: After the building edge mesh is output as a hidden occlusion candidate mesh by S5, the edge computing node writes the mesh into the candidate mesh placeholder word and performs a bitwise AND operation with the path placeholder words of multiple inter-station links; the two inter-station links that pass through the building edge mesh are written into the verification transmission bit. After the transmitting airborne station retransmits the probe frame, the new received power and new arrival delay written back by the receiving airborne station produce bitwise differences with the original link response word. Based on this, the edge computing node writes the two inter-station links into the affected link bit and rewrites the corresponding line-of-sight matrix unit to the non-line-of-sight state.

[0023] Furthermore, the present invention also includes, taking low-altitude inspection of urban parks as an example, the edge computing node of the satellite navigation application system manages three UAVs A, B, and C and a space station R participating in wireless telemetry and control forwarding. All three UAVs are registered as airborne stations, and space station R is registered as a space station. The task configuration table includes the grid origin longitude, grid origin latitude, meridional grid distance, latitudinal grid distance, elevation datum, floor height, time slice start time, and slice length, wherein the slice length is the same as the inter-station link detection period. UAV A moves from the east side of the building to the corner of the building, UAV B is located on the west side of the building, and UAV C is located in an open area. The crane arm temporarily raised near the corner of the building is not included in the 3D model, but it has entered the wireless direct path between A and B. Within the same BeiDou time synchronization slice, edge computing nodes receive high-precision positioning messages from UAVs A, B, and C respectively. Using the source station number as the write index, they subtract the grid origin longitude from the positioning longitude and divide by the meridional grid distance to obtain the meridional integer part; they subtract the grid origin latitude from the positioning latitude and divide by the latitudinal grid distance to obtain the latitudinal integer part, and concatenate them to generate the horizontal grid position. Subsequently, they subtract the elevation datum from the positioning altitude and divide by the layer height to obtain the elevation integer part; and they subtract the start time of the BeiDou time synchronization slice from the time synchronization slice start time and divide by the slice length to obtain the position. Once the timing integer bits are obtained, they are written into a BeiDou spatiotemporal grid position number according to the horizontal grid position, elevation integer bits, and timing integer bits. After the edge computing node reads the BeiDou spatiotemporal grid position numbers of A, B, and C in the previous timing slice, it finds that the horizontal grid position of UAV A has changed, while the positions of UAV B and UAV C have not changed. Therefore, the source station number of UAV A is written into the rover position, and the inter-station link position numbers from A to B, A to C, and R to A are generated in the order of "sender source station number, receiver source station number, timing slice position". For the link between stations A and B, the edge computing node solves the transmitter grid center coordinates from the BeiDou spatiotemporal grid reference number of station A and the receiver grid center coordinates from the BeiDou spatiotemporal grid reference number of station B. The link direction difference word is obtained by subtracting the transmitter grid center coordinates from the receiver grid center coordinates. The meridional, latitudinal, and elevation components in the link direction difference word are written into the recursive direction word. Starting from the transmitter grid, the edge computing node uses the current grid center coordinates as the starting point of this round and calculates the direct path to the meridional boundary and latitudinal boundary, respectively. The arrival step word of the surface and elevation interface is taken as the interface that exits in this round, and the next grid is entered along the corresponding recursive direction. When the direct path passes through the corner grid edge, the arrival step word of the meridional interface is the same as the arrival step word of the latitudinal interface, and the edge calculation node is rewritten with both the meridional and latitudinal grid positions. After multiple rounds of recursion, the link between stations A and B forms a grid crossing sequence record that includes the grid near the corner; the link between stations A and C forms a grid crossing sequence record that bypasses the grid near the corner. Edge computing nodes generate probe frames based on the A-to-B grid sequence record. The transmitting airborne station A writes the link number between stations A and B into the first segment of the probe frame, writes the current time slot number into the second segment, aligns the first and last grid numbers of the grid sequence record with the same word length, and then performs a bitwise XOR operation to form a path check word, which is then written into the last segment of the probe frame. Upon receiving the probe frame, the receiving airborne station B writes back the received power sample value and the arrival time of the probe frame. The transmitting airborne station A writes back the retransmission count readback value. The edge computing node subtracts the received power sample value of the previous time slot from the current received power sample value to form the first response word, subtracts the retransmission count readback value of the previous time slot from the current retransmission count readback value to form the second response word, and subtracts the transmission time of the probe frame from the arrival time of the probe frame to form the arrival delay word. The arrival delay word is then cyclically shifted according to the number of path grids in the grid sequence record, ultimately forming the A-to-B link response word sequence. The A-to-C link response word sequence is formed in the same way. Subsequently, the edge computing node locates the response row using the inter-station link position number and the response column using the BeiDou timing segment position. It subtracts the previous response word from the next response word from A to B to obtain the inter-segment difference word. It also subtracts the two source links A to B and A to C under A according to the lexicographical order of the inter-station link position number to obtain the source difference word, forming a change detection word table. During the sparse group fusion lasso change point detection process, when the candidate change bit is zero, the inter-segment difference word is retained as a residual word. When the candidate change bit is one, the inter-segment difference word is written as an interpreted word. Then, the source difference word is subtracted from the candidate change bit in the same column to obtain the source residual word. The XOR of the candidate change bits in two adjacent columns is used to obtain the break word. The edge computing node concatenates the non-zero number of residual words, the non-zero number of source residual words, the number of candidate change bits, and the non-zero number of break words in sequence to form the detection cost word, and rewrites the candidate change bits row by row. Finally, the candidate change bits of the A to B link are one and written into the flipped link set, and the candidate change bits of the A to C link are zero and written into the unflipped link set. During the hypergraph cut processing stage, edge computing nodes read the A-to-B grid crossing sequence records from the flipped link set and the A-to-C grid crossing sequence records from the unflipped link set. After deduplicating the path grid positions that have appeared in the current time slice, a global grid order is generated according to the lexicographical order of the BeiDou spatiotemporal grid positions. The grid positions traversed by A-to-B are written in the placeholder words to form the cutting ledger line, and the grid positions traversed by A-to-C are written in the placeholder words to form the reserved ledger line. Starting from the empty candidate cutting word, the edge computing node attempts to write the candidate positions one by one according to the vertex position order of the path hypergraph, and performs a bitwise AND operation between the candidate cutting word and the cutting ledger line to obtain the cutting hit word, and performs a bitwise AND operation between the candidate cutting word and the reserved ledger line to obtain the reserved hit word. When the candidate cutting word hits a grid near the corner, the flipped debt word corresponding to A-to-B changes from one to zero, the reserved intrusion word corresponding to A-to-C remains zero, and the cutting cost word is retained in the lexicographical order comparison. After the global grid order completes one round of trial writing, the grid near the corner is output as a hidden occlusion candidate grid. Finally, the edge computing node generates candidate grid placeholders based on the hidden occlusion candidate grids and performs a bitwise AND operation with the path placeholders from A to B and A to C. The A to B link hit word contains bits with a value of 1, the link bit number between A and B stations is written into the verification transmission bit, the grid bits near the building corner are written into the common grid bit, and the A to C link hit word is all zeros and is not written into the verification transmission bit. Drone A sends a probe frame to Drone B again according to the verification transmission bit. Drone B writes back the new received power and new arrival delay, and Drone A writes back the new retransmission count. The edge computing node reads A from the S3 cache. The original received power, original retransmission count, and original arrival delay to B are subtracted respectively to obtain the verification power difference word, verification retransmission difference word, and verification delay difference word, forming the verification response word. After bitwise XORing the verification response word with the original link response word, there is a bit with a value of one, and the A to B link hit word hits the common grid bit. The edge computing node writes A to B into the affected link bit and rewrites the matrix cells in the A to B direction in the line-of-sight matrix to the non-line-of-sight state. The A to C matrix cells maintain the original line-of-sight state, and the R to A link does not hit the hidden occlusion candidate grid, and also maintains the original line-of-sight state.

[0024] Working principle: This scheme first converts the high-precision positioning message of the UAV into BeiDou spatiotemporal grid reference number, and then generates an inter-station link reference number table based on the moving side of the space station or airborne station; then it reads the grid reference numbers at both ends of each inter-station link, calculates which subdivided grids the wireless direct path passes through, and forms a grid crossing sequence record; then the grid crossing sequence record drives the transmission of probe frames, and a link response word sequence is formed based on the received power, retransmission count, and arrival delay; then, the sparse group fusion lasso change point detection algorithm is used to identify which inter-station links have experienced line-of-sight flips and which inter-station links remain in their original state; finally, the hypergraph cut algorithm is used to deduce hidden occlusion candidate grids from the grid crossing relationship of flipped and unflipped links, and the affected links are confirmed by verifying the probe frames, and the line-of-sight matrix is ​​rewritten. During a multi-drone inspection in an urban park, one drone flew from the east side of a building to its corner, while another drone was located on the west side. A temporarily raised crane arm, though not included in the 3D model, blocked the direct wireless path between the two drones. The edge computing node calculated the link passing through the grid near the building corner based on the BeiDou spatiotemporal grid positions of the two drones. After sending a detection frame, it found that the received power of this link decreased, the retransmission count increased, and the arrival delay changed, while other links that bypassed the building corner did not change synchronously. Therefore, the grid near the building corner was identified as a candidate grid for hidden obstruction. After the detection frame was reviewed and confirmed again that the link was affected, the corresponding direction in the line-of-sight matrix was rewritten to a non-line-of-sight state, thus providing a basis for subsequent link switching, early warning display, and drone cooperative flight.

[0025] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for calculating the real-time location and line-of-sight of unmanned aerial vehicles (UAVs) based on BeiDou grid spatiotemporal intelligence, characterized in that, include: S1. At the edge computing node of the satellite navigation application system, the high-precision positioning message of satellite positioning is converted from latitude and longitude to horizontal grid position, from altitude to elevation layer position, and from BeiDou time synchronization to time synchronization segment position, synthesizing BeiDou spatiotemporal grid position number, and generating an inter-station link position number table according to the wireless transmission relationship of at least one of the space station or airborne station moving. S2. For inter-station links, read the BeiDou spatiotemporal grid position numbers at both ends, project the direct path from both ends onto the subdivided grid, and recursively traverse along the grid interface to form a grid-crossing sequence record. S3. Drive the detection frame from the transmitting airborne station to the receiving airborne station by the grid sequence record, so that the acknowledgment is written back according to the time slice, the received power is subtracted between slices to get the first response word, the retransmission count bit difference is entered into the first response word, and the arrival delay is rearranged according to the grid sequence and then entered to form the link response word sequence. S4. Based on the link response word sequence, execute the sparse group fusion lasso change point detection algorithm. The single link change word is formed by the inter-chip response difference, and the group change word is formed by the difference of the same source link. The fusion lasso term compresses the common change segment and outputs the set of flipped links and the set of unflipped links. S5. In the hypergraph cut algorithm, the grid sequence record is converted into a path hypergraph, the flipped path hyperedge is entered into the cut side, and the unflipped path hyperedge is entered into the retention side. First, the grid that does not cover the flipped path hyperedge is removed, and then the grid is removed again according to the cut result of the retention side. The hidden occlusion candidate grid is output.

2. The method for calculating the real-time location and line-of-sight of unmanned aerial vehicles based on BeiDou grid spatiotemporal intelligence according to claim 1, characterized in that: Also includes: S6. Based on the hidden occlusion candidate grid, trigger the inter-station link to retransmit the probe frame, so that the new response word sequence is written back to the time slice, and the affected link is located by the common grid number, the line-of-sight state is rewritten, and the line-of-sight matrix is ​​output.

3. The method for calculating the real-time location and line-of-sight of unmanned aerial vehicles based on BeiDou grid spatiotemporal intelligence according to claim 2, characterized in that: S1 includes: S1-1. Within the edge computing node of the satellite navigation application system, the source station number of the high-precision positioning message of the satellite positioning is used as the writing index. The positioning longitude is subtracted from the grid origin longitude and then divided by the longitudinal grid distance to obtain the longitudinal integer part. The positioning latitude is subtracted from the grid origin latitude and then divided by the latitudinal grid distance to obtain the latitudinal integer part. The longitudinal integer part and the latitudinal integer part are concatenated to generate the horizontal grid part. S1-2. Based on the horizontal grid position, subtract the elevation datum from the positioning height and divide by the floor height to obtain the elevation integer position. Subtract the start time of the time synchronization segment from the BeiDou time synchronization and divide by the segment length to obtain the time synchronization integer position. Then, write the horizontal grid position, elevation integer position and time synchronization integer position under the source station number name in the order of horizontal grid position, and output the BeiDou spatiotemporal grid position number. S1-3. Based on the BeiDou spatiotemporal grid position number, read the source station number position of the current time slice and the adjacent previous time slice. If the position of the source station number is inconsistent, write it into the mobile station position. Pair the source station number of the mobile station position belonging to the space station or airborne station with another space station or airborne station to generate an inter-station link position number table.

4. The method for calculating the real-time location and line-of-sight of unmanned aerial vehicles based on BeiDou grid spatiotemporal intelligence according to claim 3, characterized in that: S2 includes: S2-1. For inter-station links in the inter-station link tag table, the grid center coordinates of the transmitting end are solved from the BeiDou spatiotemporal grid tag number of the transmitting end, and the grid center coordinates of the receiving end are solved from the BeiDou spatiotemporal grid tag number of the receiving end. The link direction difference word is obtained by subtracting the grid center coordinates of the transmitting end from the grid center coordinates of the receiving end. The positive and negative signs of each coordinate component in the link direction difference word are written into the recursive direction word of the corresponding interface, and the link starting grid record is output. S2-2. Based on the link starting grid record, take the current grid center coordinates as the starting point of this round. First, use the recursive direction word to obtain the coordinates of the next sub-interface. Then, subtract the coordinates of the starting point of this round from the coordinates of the next sub-interface and divide by the same coordinate component in the link direction difference word to obtain the sub-interface arrival step word of this round. When the coordinate component is zero, write the corresponding sub-interface arrival step word as an empty step and stop the corresponding direction recursion. Output the sub-interface exit of this round. S2-3. Rewrite the current grid position number according to the current grid exit interface, so that the current grid position number enters the next grid along the corresponding recursive direction word, and write the next grid position number into the grid exit sequence record under the inter-station link position number until the next grid position number is consistent with the BeiDou spatiotemporal grid position number of the receiving end, and then output the grid exit sequence record.

5. The method for calculating the real-time location and line-of-sight of unmanned aerial vehicles based on BeiDou grid spatiotemporal intelligence according to claim 4, characterized in that: S3 includes: S3-1. Based on the grid sequence record, the transmitting airborne station writes the inter-station link number into the first segment of the probe frame, writes the current time synchronization segment into the second segment of the probe frame, and XORs the first grid number of the grid sequence record bit by bit with the last grid number and writes it into the last segment of the probe frame, and outputs a probe frame carrying the path check word. S3-2. After receiving the probe frame, the receiving airborne station subtracts the received power sample value of the inter-station link in the previous time slot from the received power sample value to form the first response word, and subtracts the retransmission count readback value of the inter-station link in the previous time slot from the retransmission count readback value to form the second response word. The second response word is then shifted to the left according to the bit length of the probe frame tail segment and added to the first response word. The link differential response word is then output. S3-3. Based on the writing order of the path grid in the grid sequence record, the arrival time of the probe frame is reduced by the transmission time of the probe frame to form the arrival delay word. The arrival delay word is then cyclically shifted according to the writing order of the path grid and connected to the link differential response word. The link response word sequence is then output.

6. The method for calculating the real-time location and line-of-sight of unmanned aerial vehicles based on BeiDou grid spatiotemporal intelligence according to claim 5, characterized in that: S4 includes: S4-1. Based on the link response word sequence, locate the response row by the inter-station link position number and the response column by the BeiDou time synchronization position. Subtract the previous response word from the next response word in the same response row to obtain the inter-segment difference word. Subtract the previous link response word from the next link response word under the same airborne station name to obtain the same source difference word. Write the inter-segment difference word and the same source difference word back according to the response row to form a change detection word table. S4-2. In the sparse group fusion lasso change point detection algorithm, candidate change bits are generated column by column starting from the first response column. When the candidate change bit is zero, the inter-slice difference word is retained as a residual word. When the candidate change bit is one, the inter-slice difference word is written as an interpreted word. Then, the same source difference word is subtracted from the candidate change bit in the same column to generate the same source residual word. The candidate change bits in two adjacent columns are XORed to generate the break word. The detection cost word is generated by splicing the residual word with the number of non-zero bits, the same source residual word with the number of non-zero bits, the number of candidate change bits, and the number of non-zero bits.

7. The method for calculating the real-time location and line-of-sight of unmanned aerial vehicles based on BeiDou grid spatiotemporal intelligence according to claim 6, characterized in that: S4 also includes: S4-3. Rewrite the candidate change bits line by line for the change detection word table. After each rewrite, regenerate the detection cost word and retain the candidate change bits with the dictionary order of the detection cost word first. Continue until the candidate change bits are consistent after one round of line-by-line rewriting. Write the inter-station links with a candidate change bit of 1 into the flipped link set and write the inter-station links with a candidate change bit of 0 into the unflipped link set.

8. The method for calculating the real-time location and line-of-sight of unmanned aerial vehicles based on BeiDou grid spatiotemporal intelligence according to claim 7, characterized in that: S5 includes: S5-1. Based on the flipped link set and the non-flipped link set, read the grid order record, write the path grid position number of each inter-station link into the placeholder word according to the global grid order, and write the sequence number of the path grid position number in the grid order record into the same position order word. The placeholder word of the flipped link forms the cutting page row, and the placeholder word of the non-flipped link forms the reserved page row. The grid position with a value of one in the placeholder word generates the path hypergraph vertex. S5-2. In the hypergraph cut algorithm, the candidate cut word is bitwise ANDed with the cut page row to obtain the cut hit word. When the cut hit word is all zero, write 1 in the flipped debt word. When the cut hit word is non-zero, write 0 in the flipped debt word. The candidate cut word is bitwise ANDed with the retained page row to obtain the retained hit word. When the retained hit word is non-zero, write 1 in the retained intrusion word. When the retained hit word is all zero, write 0 in the retained intrusion word. Then, the number of 1s in the flipped debt word, the number of 1s in the retained intrusion word, the number of 1s in the candidate cut word, and the sum of the corresponding order words are concatenated to form the cut cost word.

9. The method for calculating the real-time location and line-of-sight of unmanned aerial vehicles based on BeiDou grid spatiotemporal intelligence according to claim 8, characterized in that: S5 also includes: S5-3. Starting from an empty candidate cut word, execute the candidate bit writing one and recalculate the cut cost word bit by bit according to the vertex position order of the path hypergraph. If the new cut cost word is in the dictionary order first, retain the corresponding candidate cut word. If the new cut cost word is in the dictionary order later, restore the original candidate cut word. Continue until the vertex position order of the path hypergraph is completed in one round and the candidate cut word has not been rewritten. Output the grid bits with a value of one in the candidate cut word as hidden occlusion candidate grids.

10. The method for calculating the real-time location and line-of-sight of unmanned aerial vehicles based on BeiDou grid spatiotemporal intelligence according to claim 9, characterized in that: S6 includes: S6-1. Generate candidate grid placeholders based on hidden occlusion candidate grids. Perform bitwise AND operations between the candidate grid placeholders and the path placeholders of each inter-station link to obtain the link hit word. If there is a bit with a value of 1 in the link hit word, write it into the verification transmission bit and write the first grid bit with a value of 1 into the common grid bit. Output the verification link record. S6-2. Based on the verification link record, the sending airborne station sends a probe frame according to the verification transmission bit, so that the receiving airborne station subtracts the original received power from the new received power to form a verification power difference word, subtracts the original retransmission count from the new retransmission count to form a verification retransmission difference word, subtracts the original arrival delay from the new arrival delay to form a verification delay difference word, and forms a verification response word in the order of writing the verification power difference word, the verification retransmission difference word, and the verification delay difference word; S6-3. Rewrite the line-of-sight matrix according to the verification response word, and generate the verification rewritten word by bitwise XORing the verification response word with the original link response word. If there are inter-station links with a value of 1 in the verification rewritten word, write them into the affected link bits according to the common grid bits. The matrix cells corresponding to the affected link bits are rewritten to the non-line-of-sight state. The matrix cells that are not written into the affected link bits retain the original line-of-sight state. Output the line-of-sight matrix.

Images