A risk grid updating method and system for low-altitude control of a UAV

Abstract

The application discloses a risk grid updating method and system for unmanned aerial vehicle low-altitude control, and belongs to the technical field of unmanned aerial vehicle low-altitude airspace control. The method obtains multi-source risk data of a risk grid, analyzes to obtain risk source identification, collection batch identification, action grid, effective time period, credibility and source dependency relationship, generates a risk source contribution chain containing direct contribution value and derived contribution value and writes into a contribution ledger, and generates a current risk level based on residual contribution items. When a review state indicates that a risk source is corrected or denied, homologous direct contribution and homologous derived contribution are located, deleted or attenuated, and heterologous contribution is retained to obtain a corrected contribution ledger, generate a corrected risk level and output a control instruction. The method can reduce the situation that a heterologous real risk is incorrectly removed when a false positive source is revoked.

Description

Technical Field

[0001] This application belongs to the field of low-altitude airspace management and control technology for unmanned aerial vehicles (UAVs), specifically a risk grid update method and system for low-altitude UAV management and control. Background Technology

[0002] Low-altitude drone control typically requires dividing urban low-altitude airspace, industrial park low-altitude airspace, or airspace surrounding logistics routes into multiple risk grids. Based on weather conditions, communication quality, sensitive ground areas, pedestrian density, obstacle changes, drone identification results, and temporary control information, each risk grid is assigned a low-risk, medium-risk, or high-risk status. Existing low-altitude control systems can dynamically update the risk grids based on multi-source data and output control commands such as detour, speed limit, hovering, return to base, or prohibition of entry.

[0003] However, when multiple sources of risk data simultaneously impact the same risk grid, existing methods typically only save the final risk level of that risk grid, or only record some of the risk events that triggered that risk level. For the risk impacts generated by different data sources within the same risk grid at different collection periods, existing methods usually lack records that can be used to distinguish the scope, duration, and cross-grid correlations of these impacts. Therefore, when a risk source is subsequently determined to be a false alarm, invalid, or have reduced credibility by manual review, cross-sensor verification, or data correction, the system struggles to determine which grids, time batches, and associated risk assessments should be affected by this review result.

[0004] In the above situation, if the risk level of the entire risk grid is directly reduced, other risks that still exist, such as strong winds, communication obstruction, and crowds gathering on the ground, may also be eliminated, leading to premature relaxation of low-altitude airspace control instructions. If the original risk level is maintained, false alarm sources that have already been rejected may continue to affect flight path control, causing drones to detour, be speed-limited, or be prohibited from entering for extended periods. If all multi-source data is re-integrated, multiple grids, multiple data collection periods, and cross-grid correlation effects need to be reprocessed, resulting in a large amount of computation, and it is still difficult to accurately distinguish the impact caused by rejected risk sources from the impact caused by other real risk sources.

[0005] For example, during the operation of low-altitude logistics routes in urban commercial areas, a certain risk grid may be simultaneously affected by short-term strong winds, degraded communication quality, increased ground pedestrian density, and a suspected unauthorized flight target identified by a visual camera. Based on this, the system updates the risk grid to high risk and issues detour or exclusion instructions to drones passing through the grid. Subsequently, manual review or radar data comparison reveals that the suspected unauthorized flight target identified by the visual camera was actually a reflection from a crane boom or a floating object at high altitude, not a real drone target. If the existing system only saves the final high-risk level of the grid, without saving data that can be used to limit the impact range of the visual recognition result, the collection period, and subsequent related effects, the system will find it difficult to simply reverse the risk impact caused by the false alarm. In this case, if the grid is directly downgraded to low risk, the remaining risks of strong winds and degraded communication quality may be ignored; if it remains at high risk, it will lead to long-term detours or exclusions for logistics drones, reducing airspace efficiency.

[0006] Therefore, the existing risk grid update method has problems such as coarse risk level correction granularity, inaccurate false alarm rollback, and easy erroneous removal of other still valid risks after the risk source has been reviewed, corrected or denied. Summary of the Invention

[0007] This application provides a risk grid update method and system for low-altitude management of unmanned aerial vehicles (UAVs). When multiple risk data sources work together on the same risk grid, the system records the direct contribution, derived contribution, and source dependency of different risk sources in the risk level formation process. After a certain risk source is corrected or negated, the system locally deletes or attenuates the homogeneous contribution generated by that risk source, while retaining the heterogeneous contribution that is still within the valid time period. This reduces the problems of incorrect overall risk level downgrading, incorrect maintenance, and incorrect removal of heterogeneous real risks.

[0008] In a first aspect, this application provides a risk grid update method for low-altitude drone management, comprising: acquiring multi-source risk data of the risk grid; parsing the multi-source risk data to obtain a risk source identifier, a collection batch identifier, an action grid, an effective time period, a credibility level, and a source dependency relationship; generating a risk source contribution chain containing direct contribution values ​​and derived contribution values ​​based on the action grid, the credibility level, and the source dependency relationship; writing the risk source contribution chain into the contribution ledger of the action grid according to the risk source identifier and the collection batch identifier, filtering out invalid contribution items according to the effective time period, and generating a current risk level based on the remaining contribution items; receiving a review status matching the risk source identifier and the collection batch identifier; when the review status indicates that the risk source has been corrected or denied, locating the same-source direct contribution and the same-source derived contribution from the contribution ledger according to the source dependency relationship, deleting or attenuating the same-source direct contribution and the same-source derived contribution while retaining the heterogeneous contribution, to obtain a corrected contribution ledger; generating a corrected risk level based on the corrected contribution ledger, writing the current risk level or the corrected risk level into the risk grid to obtain a risk grid update result, and outputting management instructions based on the risk grid update result.

[0009] Using the above method, this application does not only save the final risk level of the risk grid, but also simultaneously records the risk source identifier, data collection batch identifier, direct contribution value, derived contribution value, effective time period, credibility, and source dependency relationship during the risk level formation process. When the subsequent review status indicates that a certain risk source has been corrected or denied, the direct contribution corresponding to that risk source and the derived contribution formed by the propagation or triggering of that risk source can be located based on the risk source identifier, data collection batch identifier, and source dependency relationship, and only the contribution from the same source is deleted or attenuated. Therefore, it avoids the simultaneous removal of real risks from other sources such as strong winds, communication obstruction, and ground crowds due to a false alarm from a visual recognition, a single sensor failure, or correction of a single batch of data, making the risk grid update result closer to the actual risk state of low-altitude airspace.

[0010] In one possible implementation, acquiring multi-source risk data for the risk grid includes: acquiring at least two of the following: meteorological data, communication data, ground environment data, and target identification data corresponding to the risk grid; wherein the meteorological data includes wind speed, wind direction, precipitation, or visibility; the communication data includes link quality or communication obstruction status; the ground environment data includes population density or sensitive area information; and the target identification data includes identification results of suspected UAV targets, obstacle targets, or temporary floating object targets. This implementation allows the risk level of the risk grid to be supported by multiple types of low-altitude control data, avoiding frequent erroneous updates to the risk level due to fluctuations from a single data source.

[0011] In one possible implementation, parsing the multi-source risk data includes: reading the data source field and collection time field carried by the multi-source risk data; converting the data source field into a risk source identifier; writing data from the same source within the same collection window into the same collection batch identifier; determining the action grid based on the location coordinates, coverage area, and spatial boundary of the risk grid corresponding to the multi-source risk data; and determining the effective time period based on the collection time and the effective duration of the source. This implementation ensures that each piece of risk data has a traceable source, batch, spatial scope, and effective time range, providing a data foundation for subsequent contribution ledger writing, invalid contribution item screening, and review status matching.

[0012] In one possible implementation, the confidence level is determined by the sensor source level, the freshness of the data acquisition time, and the cross-validation results. The confidence level is obtained by weighting the base confidence value corresponding to the sensor source level, the time decay value from the acquisition time to the current time, and the validation values ​​from other sources for the same risk event. When the cross-validation results indicate that the same risk event is supported by at least one heterogeneous data source, the confidence level is increased by a preset confidence increment according to a preset confidence correction rule, ensuring that the increased confidence level does not exceed a preset confidence upper limit. This confidence level is used to generate direct contribution values ​​and derived contribution values. This implementation allows risk contributions to not only depend on the risk event itself but also incorporate source reliability, data freshness, and heterogeneous validation results in the calculation, thereby reducing the excessive influence of low-confidence sources on the risk level.

[0013] In one possible implementation, the source dependency includes the spatial propagation relationship of the same risk source to adjacent risk grids and the triggering relationship between different risk events. When generating the risk source contribution chain, the risk impact within the action grid is recorded as a direct contribution value, and the risk impact transmitted to the action grid by the spatial propagation relationship or the triggering relationship is recorded as a derived contribution value. This implementation distinguishes between the direct impact of a risk source within its own action grid and the impact transmitted through adjacent grids or related events, enabling the rollback of both direct and propagated impacts from the same source during review and correction.

[0014] In one possible implementation, the contribution ledger includes multiple contribution items, each including a risk source identifier, a collection batch identifier, a direct contribution value, a derived contribution value, a credibility level, a valid time period, and a source dependency. When filtering out invalid contribution items according to the valid time period, contribution items that do not belong to the corresponding valid time period at the current moment are marked as invalid contribution items and excluded from the current risk level calculation, while contribution items that belong to the corresponding valid time period at the current moment are treated as remaining contribution items. This implementation method can prevent historical risk data that has exceeded its valid time period from continuing to participate in the current risk level calculation, reducing the risk level misrepresentation caused by expired risk contributions.

[0015] In one possible implementation, generating the current risk level based on remaining contribution items includes: reading the direct contribution value, derived contribution value, and confidence level from the remaining contribution items; summing the direct contribution value and the derived contribution value to obtain the total source contribution; multiplying the total source contribution by the corresponding confidence level to obtain the effective source contribution; summing the effective source contributions within the same action grid to obtain the grid risk contribution; and mapping the grid risk contribution to the current risk level according to a preset risk level mapping table. Through this implementation, risk contributions from different sources, batches, and propagation paths can be uniformly converted into comparable grid risk contribution amounts, and the current risk level can be formed accordingly.

[0016] In one possible implementation, the review status is generated by manual review results, cross-sensor verification results, or data correction results, and includes a confirmed status, a corrected status, and a negative status. When the review status is confirmed, the corresponding contribution item in the contribution ledger remains unchanged. When the review status is corrected, the post-review credibility or review attenuation ratio carried by the review status is read. When the post-review credibility is read, the direct contribution and the derived contribution are multiplied by the post-review credibility to obtain the attenuated contribution. When the review attenuation ratio is read, the direct contribution and the derived contribution are multiplied by the review attenuation ratio to obtain the attenuated contribution. When the review status is negative, the corresponding direct contribution and the derived contribution are deleted from the contribution ledger. Through this implementation, the retention, attenuation, or deletion processing can be performed according to different types of review conclusions, giving the risk source correction process a clear processing path.

[0017] In one possible implementation, retaining heterogeneous contributions includes: reading contribution items in the contribution ledger whose risk source identifiers differ from those of the corrected or negated risk sources, and retaining contribution items that are within their valid time period and do not have a negative status review record matching the risk source identifier and collection batch identifier of the contribution item as heterogeneous contributions in the corrected contribution ledger; when the effective contribution amount of the heterogeneous contributions is not less than a preset mandatory retention threshold, or the number of risk sources corresponding to the heterogeneous contributions is not less than a preset source quantity threshold, ensuring that the corrected risk level is not lower than a preset minimum control level. This implementation allows for the continued retention of other valid and unnegated risk sources after a false alarm source is revoked, and maintains the necessary minimum control level while heterogeneous contributions still meet control requirements.

[0018] Secondly, this application provides a risk grid update system for low-altitude drone management, including a data acquisition module, a data parsing module, a contribution chain generation module, a ledger update module, a review and correction module, and an instruction output module. The data acquisition module acquires multi-source risk data for the risk grid. The data parsing module parses the multi-source risk data to obtain risk source identifiers, collection batch identifiers, action grids, effective time periods, credibility, and source dependencies. The contribution chain generation module generates a risk source contribution chain containing direct contribution values ​​and derived contribution values ​​based on the action grids, credibility, and source dependencies. The ledger update module writes the risk source contribution chain into the data according to the risk source identifier and collection batch identifier. The contribution ledger of the action grid filters out invalid contributions according to the effective time period and generates the current risk level based on the remaining contributions; the review and correction module is used to receive the review status matching the risk source identifier and the collection batch identifier; when the review status indicates that the risk source has been corrected or denied, the same-source direct contributions and same-source derived contributions are located from the contribution ledger according to the source dependency relationship, the same-source direct contributions and same-source derived contributions are deleted or attenuated and the heterogeneous contributions are retained to obtain the corrected contribution ledger; the instruction output module is used to generate a corrected risk level based on the corrected contribution ledger, write the current risk level or the corrected risk level into the risk grid to obtain the risk grid update result, and output the control instruction according to the risk grid update result.

[0019] Compared with related technologies, this application has the following advantages: By acquiring multi-source risk data from the risk grid and parsing it to obtain risk source identifiers, collection batch identifiers, action grids, effective time periods, credibility, and source dependencies, a risk source contribution chain containing direct and derived contribution values ​​is generated based on the action grids, credibility, and source dependencies. The risk source contribution chain is then written into the contribution ledger of the action grid according to the risk source identifier and collection batch identifier. This allows the data involved in risk level generation to be distinguished and saved according to source, batch, action grid, effective time period, and impact mode, thus providing a traceable and calculable data foundation for subsequent risk source verification, contribution location, and local correction.

[0020] By filtering out failed contributions according to the effective time period and generating the current risk level based on the remaining contributions, the impact of risks that have exceeded the effective time period is no longer included in the calculation of the current risk level, thereby improving the matching degree between the current state of the risk grid and the actual changes in the low-altitude environment. At the same time, by receiving the review status that matches the risk source identifier and the collection batch identifier, and when the review status indicates that the risk source has been corrected or denied, the same-source direct contributions and same-source derived contributions are located from the contribution ledger according to the source dependency relationship. The same-source direct contributions and same-source derived contributions are deleted or attenuated, so that the impact of the risk source that has been reviewed, corrected or denied can be partially rolled back in the contribution ledger, thereby reducing the amount of calculation required to re-integrate all multi-source risk data and reducing the probability that false alarm sources will continue to affect the risk level.

[0021] By deleting or attenuating direct and derived contributions from the same source while retaining contributions from different sources, other risks such as strong winds, communication obstruction, and ground crowd gatherings that are within the effective time period and have not been negated can still participate in the calculation of the corrected risk level. This reduces the possibility of other effective risks being eliminated due to the revocation of a single false alarm source. Furthermore, by generating a corrected risk level based on the corrected contribution ledger, the current risk level or the corrected risk level is written into the risk grid to obtain the risk grid update result. Then, control instructions are output based on the risk grid update result, so that instructions such as detour, speed limit, hovering, return to base, no-entry, or lifting control can be updated synchronously with the risk level correction result. This improves the matching and continuity between low-altitude control instructions and the actual state of the risk grid. Attached Figure Description

[0022] Figure 1 A general flowchart of a risk grid update method for low-altitude management of unmanned aerial vehicles provided in this application embodiment; Figure 2 A flowchart of multi-source risk data acquisition and analysis provided in this application embodiment; Figure 3 A flowchart for generating a risk source contribution chain is provided for an embodiment of this application; Figure 4 A flowchart for writing contribution ledgers and generating current risk levels is provided for embodiments of this application; Figure 5 A flowchart of a verification state-driven homogeneous contribution correction is provided for an embodiment of this application; Figure 6 A flowchart of heterogeneous contribution retention and control instruction output provided in this application embodiment; Figure 7 A schematic diagram of the module structure of a risk grid update system for low-altitude management of unmanned aerial vehicles provided in this application embodiment; Figure 8 This is a schematic diagram illustrating an application scenario of a risk grid update method for low-altitude management of unmanned aerial vehicles (UAVs) provided in an embodiment of this application. Detailed Implementation

[0023] The embodiments of this application will now be described with reference to the accompanying drawings. The described embodiments are only for explaining the technical solutions of this application and are not intended to limit the scope of protection of this application. Where there is no conflict, the technical features in the following embodiments can be combined with each other.

[0024] In this embodiment, the risk grid refers to a spatial management unit formed by dividing low-altitude airspace according to latitude and longitude ranges, altitude ranges, and time control windows. Each risk grid has a grid number, spatial boundaries, altitude range, current risk level, and control status. Risk levels include low risk, medium risk, and high risk, and can also be divided into more levels according to the control requirements of the low-altitude control platform. The risk level division rules are pre-configured by the low-altitude control platform and stored as a risk level mapping table.

[0025] like Figure 8 As shown, in one application scenario, the low-altitude airspace surrounding urban commercial areas, industrial parks, or logistics routes is divided into multiple risk grids. Logistics drones pass through the target risk grid and its adjacent risk grids along a preset route. Micro-weather stations, communication base stations, visual cameras, radar, and ground-based pedestrian monitoring equipment are set up around the target risk grid. The micro-weather stations send meteorological data such as wind speed, wind direction, precipitation, or visibility to the low-altitude control platform. The communication base stations or communication monitoring equipment send communication data such as link quality, packet loss rate, latency, or communication obstruction status to the low-altitude control platform. The ground-based pedestrian monitoring equipment or industrial park management system sends ground environmental data such as pedestrian density, sensitive area status, or temporary activity areas to the low-altitude control platform. The visual cameras or radar send the identification results of suspected drone targets, obstacle targets, or temporary floating targets to the low-altitude control platform.

[0026] After receiving the aforementioned multi-source risk data, the low-altitude airspace control platform generates a risk source contribution chain according to the risk source identifier and the collection batch identifier, and writes the risk source contribution chain into the contribution ledger of the corresponding risk grid. The verification terminal is used to feed back the manual verification results, cross-sensor verification results, or data correction results to the low-altitude airspace control platform. When a risk source is indicated as corrected or denied by the verification status, the low-altitude airspace control platform locates the same-source direct contribution and the same-source derived contribution from the contribution ledger according to the source dependency relationship, deletes or attenuates the same-source direct contribution and the same-source derived contribution, and retains the heterogeneous contribution that is within the valid time period and has not been denied, generating a corrected contribution ledger. The low-altitude airspace control platform generates a corrected risk level based on the corrected contribution ledger and outputs detour, speed limit, hover, return, prohibition, or release control instructions to the logistics drone, the low-altitude traffic management platform, or the route scheduling system.

[0027] The multi-source risk data in this embodiment refers to at least two types of data that can affect the risk level of the risk grid. Multi-source risk data includes meteorological data, communication data, ground environment data, and target identification data. Meteorological data is provided by meteorological stations, micro-meteorological sensors, or meteorological service interfaces; communication data is provided by UAV communication link monitoring equipment, ground base stations, or air-to-ground link quality detection modules; ground environment data is provided by ground pedestrian monitoring equipment, park management systems, or sensitive area databases; and target identification data is provided by visual cameras, radar, acoustic detection equipment, or multi-sensor fusion identification modules.

[0028] The risk source contribution chain in this embodiment refers to a chain of data objects that record the risk contribution of risk sources to risk grids. The risk source contribution chain includes at least a risk source identifier, a collection batch identifier, an impact grid, a direct contribution value, a derived contribution value, an effective time period, credibility, and source dependency. The direct contribution value represents the amount of risk impact the risk source has on the impact grid itself. The derived contribution value represents the amount of subsequent risk impact the risk source has on other grids or the same grid through spatial propagation or event triggering relationships. The source dependency represents the generation relationship between the direct contribution value and the derived contribution value.

[0029] The contribution ledger in this application embodiment refers to a record table or database table that stores multiple risk source contribution chains according to the risk grid. Each contribution item in the contribution ledger includes a risk source identifier, a collection batch identifier, a direct contribution value, a derived contribution value, an effective time period, credibility, source dependency relationship, and a review status matching field. The contribution ledger participates in the calculation of the current risk level, the revised contribution ledger, and the revised risk level, and is not only used for log recording.

[0030] like Figure 1As shown in the figure, the method for updating the risk grid for low-altitude management of unmanned aerial vehicles provided in this application includes the following steps.

[0031] Step S101: Acquire multi-source risk data for the risk grid. In one implementation, the low-altitude airspace management platform reads the multi-source risk data corresponding to each risk grid according to a preset acquisition cycle. The preset acquisition cycle is set based on the density of UAV flight routes, sensor refresh frequency, and management response time. For example, when low-altitude logistics routes are dense and a short management response time is required, the preset acquisition cycle is set to an integer multiple less than the minimum stable refresh cycle of the sensor; when the UAV traffic in the airspace is lower than a preset traffic threshold, the preset acquisition cycle is set to a sampling cycle no greater than the allowable management response time. The preset acquisition cycle is written into the acquisition configuration table during system initialization. The acquisition configuration table includes data type, data source interface, acquisition cycle, data validity duration, and spatial coverage. The low-altitude airspace management platform queries the acquisition configuration table according to the data type, reads the corresponding data source interface and acquisition cycle, and initiates a data reading request accordingly.

[0032] During the data collection process, the low-altitude control platform reads wind speed, wind direction, precipitation, and visibility from the meteorological interface; link quality, packet loss rate, latency, and communication obstruction status from the communication monitoring interface; population density, sensitive area status, and temporary activity areas from the ground environment interface; and identification results for suspected UAV targets, obstacle targets, or temporary floating objects from the target identification interface. Each piece of multi-source risk data carries a data type field, a data source field, a collection time field, a location field, a spatial coverage field, and a data content field. The data type field is used to distinguish meteorological data, communication data, ground environment data, or target identification data; the data source field is used to generate a risk source identifier; the collection time field is used to generate a collection batch identifier and an effective time period; the location field and spatial coverage field are used to determine the effective grid; and the data content field is used to convert it into a risk impact quantity.

[0033] In one optional implementation, when the collection times of data from different sources are inconsistent, the low-altitude airspace control platform uses the risk grid update time as a benchmark and reads data within a preset time window before the update time. The preset time window is determined based on the effective duration of various types of risk data and is written into the collection configuration table. For multiple data points from the same risk source within the preset time window, the platform sorts them from closest to furthest from the update time based on the collection time, and selects the data that is sorted first as the data to participate in this update. For meteorological data or communication data that requires smoothing, the platform reads the multiple data points that are sorted first and calculates the average value, using the average value as the current value of the data content field. Through the above processing, the platform generates a time-aligned multi-source risk data set. This set is used by step S102 to parse the risk source identifier, collection batch identifier, action grid, effective time period, credibility, and source dependency relationship.

[0034] Step S102: Analyze the multi-source risk data to obtain the risk source identifier, collection batch identifier, action grid, effective time period, credibility, and source dependency relationship.

[0035] like Figure 2 As shown, the low-altitude airspace control platform reads the data source field of each multi-source risk data point and converts it into a risk source identifier. The risk source identifier consists of data type, device number, and source sequence number. For example, a suspected target identification result output by a visual camera is recorded as a target identification source, and strong wind data output by a micro-weather station is recorded as a meteorological source.

[0036] The low-altitude airspace control platform reads the data collection time field and writes data generated from the same risk source within the same data collection window into the same data collection batch identifier. The data collection window is determined by the data collection cycle. For example, when the data collection cycle is ten seconds, data output by the same sensor within the same ten-second window is classified into the same data collection batch identifier. The data collection batch identifier is used for subsequent verification of status matching to avoid processing data from other time batches by only considering the risk source identifier.

[0037] For example, if the same micro-weather station M01 reports gust data three times consecutively within the acquisition window from 10:05:00 to 10:05:10, the platform will write all three gust data into the acquisition batch M01-100500 corresponding to the meteorological source M01. If the same micro-weather station M01 reports gust data in the next acquisition window from 10:05:10 to 10:05:20, then the data will be written into acquisition batch M01-100510. If subsequent review only rejects the data corresponding to acquisition batch M01-100500, the platform will only locate, delete, or attenuate the direct and derived contributions from the same source formed by that batch, without affecting the valid contribution items corresponding to acquisition batch M01-100510.

[0038] The low-altitude airspace control platform reads the location and spatial coverage fields, comparing the location coordinates corresponding to multi-source risk data with the spatial boundaries of the risk grid. For point-like risk sources, the platform identifies the risk grids into which the location coordinates fall as the effective grids. For area-like risk sources, the platform identifies grids whose spatial coverage overlaps with the spatial boundaries of the risk grids as effective grids. For moving target risk sources, the platform reads the target's current position, direction of movement, and speed, and determines the grids it is expected to pass through based on the current position, direction of movement, and preset prediction duration. The platform then identifies the current position grid and the grids it is expected to pass through as effective grids. The preset prediction duration is configured in the prediction duration configuration table by the UAV control platform based on the flight path response time and target identification refresh cycle. The platform queries the prediction duration configuration table according to the target type and reads the corresponding value.

[0039] The low-altitude air traffic control platform determines the valid time period based on the data acquisition time field and the data validity duration. The data validity duration is pre-configured by the acquisition configuration table according to the data type. For example, the validity duration of short-term strong wind data is shorter than that of temporary control information. The valid time period consists of a start time and an end time; the start time is the acquisition time, and the end time is the sum of the acquisition time and the data validity duration. The valid time period is used in step S104 to filter out invalid contribution items.

[0040] The low-altitude airspace control platform determines credibility based on sensor source level, data collection time freshness, and cross-validation results. Specifically, the platform pre-configures a credibility weight table, which includes data type, source level weight, data collection time freshness weight, cross-validation weight, lower credibility limit, and upper credibility limit. The platform reads the data type field, data source field, and data collection time field from the multi-source risk data and queries the corresponding weight in the credibility weight table according to the data type field. The platform reads the sensor ledger table corresponding to the data source field. The sensor ledger table includes the device number, device type, most recent calibration result, calibration validity period, total number of historical reviews, number of historical false alarms, and maintenance status. When the most recent calibration result is qualified and has not exceeded the calibration validity period, the platform assigns a first calibration level. When the most recent calibration result is qualified but the remaining time before the expiration of the calibration validity period does not exceed a preset expiration period, the platform assigns a second calibration level. When the calibration validity period has expired or the most recent calibration result is unqualified, the platform assigns a third calibration level. The platform divides the number of historical false alarms by the total number of historical reviews to obtain the historical false alarm rate. When the total number of historical reviews is 0, the historical false alarm rate is recorded as the preset initial false alarm rate. The platform queries the preset source level mapping table based on the calibration level, the range of the historical false alarm rate, and the maintenance status to obtain the source level value. The platform subtracts the current time from the collection time corresponding to the collection time field to obtain the collection time difference, and queries the preset freshness mapping table based on the collection time difference to obtain the time freshness value. The larger the collection time difference, the lower the corresponding time freshness value. The platform uses the risk event type, action grid, and collection time corresponding to the current multi-source risk data as a benchmark to read other source data within adjacent time windows. When the risk event type of other source data is the same as the current risk event type or there is a corresponding relationship in the preset event correspondence table, the action grid of other source data is the same as the current action grid or belongs to the adjacent risk grid of the current action grid, and the collection time of other source data is within the adjacent time window, the platform determines the other source data as data supporting the current risk event. When the number of data supporting the current risk event is not less than the preset support source number threshold, the platform determines the cross-validation result as a supported state. When the number of data supporting the current risk event is 0, the platform determines the cross-validation result as a unsupported state. When there is data that negates the current risk event within the adjacent time window, the platform determines the cross-validation result as a conflict state. The platform queries the preset cross-validation mapping table according to the supported state, unsupported state, or conflict state to obtain the cross-validation value.

[0041] The platform multiplies the source level value by the source level weight to obtain the source level weighted value; multiplies the time freshness value by the time freshness weight to obtain the time freshness weighted value; multiplies the cross-validation value by the cross-validation weight to obtain the cross-validation weighted value; and adds the source level weighted value, time freshness weighted value, and cross-validation weighted value together to obtain the initial credibility.

[0042] The platform also invokes preset credibility correction rules to correct the initial credibility based on heterogeneous support. These preset credibility correction rules include a heterogeneous support quantity range, a preset credibility increment, a preset credibility reduction amount, a credibility lower limit, and a credibility upper limit. The heterogeneous support quantity range is set based on the number of times the same risk event can be cross-validated by different data sources. The preset credibility increment is set based on the correspondence between the number of heterogeneous supports and the risk event confirmation ratio in historical manual review records. The preset credibility reduction amount is set based on the correspondence between the heterogeneous data conflict status and the false alarm ratio in historical manual review records. When the platform invokes the preset credibility correction rules, it uses the cross-validation results and the number of heterogeneous data supports as query conditions to read the corresponding preset credibility increment or preset credibility reduction amount.

[0043] When the cross-validation result indicates that the same risk event is supported by at least one heterogeneous data source, the platform reads the preset confidence increment corresponding to the number of heterogeneous data sources according to the preset confidence correction rule, adds the initial confidence to the preset confidence increment, and obtains the corrected confidence. When the corrected confidence is greater than the preset confidence upper limit, the preset confidence upper limit is determined as the confidence. When the corrected confidence is not greater than the preset confidence upper limit, the corrected confidence is determined as the confidence.

[0044] For example, in risk grid G08 at the park boundary, visual camera V02 identifies a suspected drone target, and radar R01 detects a target point with a matching trajectory within the same risk grid or its adjacent risk grids in an adjacent time window. The platform determines that the visual recognition result and the radar point belong to the same type of support relationship based on a preset event correspondence table, and determines the cross-validation result as a support status. If the most recent calibration result of visual camera V02 is qualified but close to its calibration validity period, while the source level corresponding to radar R01 is higher, the platform reads the preset confidence increment corresponding to a heterogeneous support source based on the initial confidence level, adjusts the confidence level of the visual recognition source upwards, and ensures that the adjusted confidence level does not exceed the preset confidence level upper limit.

[0045] When the cross-validation result is unsupported, the platform does not read the preset confidence increment and uses the initial confidence as the confidence to be limited; when the confidence to be limited is greater than the preset confidence upper limit, the preset confidence upper limit is determined as the confidence; when the confidence to be limited is less than the preset confidence lower limit, the preset confidence lower limit is determined as the confidence; when the confidence to be limited is between the preset confidence lower limit and the preset confidence upper limit, the confidence to be limited is determined as the confidence.

[0046] When the cross-validation result is in a conflict state, the platform reads the preset confidence reduction amount corresponding to the conflict state according to the preset confidence correction rule, subtracts the preset confidence reduction amount from the initial confidence, and obtains the conflict-corrected confidence. When the conflict-corrected confidence is less than the preset confidence lower limit, the preset confidence lower limit is determined as the confidence. When the conflict-corrected confidence is not less than the preset confidence lower limit, the conflict-corrected confidence is determined as the confidence.

[0047] The credibility is bound to the risk source identifier, the collection batch identifier, the action grid, and the effective time period, and is used for subsequent generation of the risk source contribution chain and calculation of the effective contribution of the source.

[0048] The preset freshness mapping table can be pre-set according to the effective duration, collection and refresh cycle, and allowable response time of low-altitude air traffic control for different types of multi-source risk data. Specifically, the platform sets time difference intervals for meteorological data, communication data, ground environment data, and target identification data, and configures corresponding time freshness values ​​for each time difference interval. The upper limit of the time difference interval is no greater than the effective duration of the corresponding data type, and the smaller the collection time difference, the higher the corresponding time freshness value. For target identification data and communication data that change rapidly, the time difference interval is set relatively short; for sensitive area information or temporary activity area information that changes slowly, the time difference interval is set relatively long. When the platform calls the preset freshness mapping table, it uses the data type and collection time difference as query conditions to read the corresponding time freshness value.

[0049] The preset event correspondence table can be pre-set based on low-altitude airspace control business rules, historical risk event records, and manual review records. The table includes the current risk event type, corresponding risk event types, correspondence relationship types, action grid relationships, and effective time windows. The correspondence relationship types include similar support relationships and related support relationships. Similar support relationships indicate that the same risk event is detected from different sources, such as both visual cameras and radar detecting suspected drone targets. Related support relationships indicate that there is a risk correlation in low-altitude airspace control between different risk events, such as the correspondence between communication obstruction and drone control risk, strong winds and flight path deviation risk, and increased ground pedestrian density and restricted access control risk. When the platform calls the preset event correspondence table, it uses the current risk event type and risk event types from other sources as query conditions to determine whether there is a similar support relationship or a related support relationship between them, and uses the determination result to determine the cross-validation result.

[0050] The low-altitude airspace control platform also determines source dependencies based on spatial propagation relationships and event triggering relationships. These source dependencies record whether the direct contribution value corresponding to a risk source will generate a derived contribution value in adjacent risk grids, or whether it will trigger a derived contribution value corresponding to another type of risk event. Specifically, the platform uses the risk source identifier, collection batch identifier, risk source type, risk event type, action grid, collection time, location field, spatial coverage field, and data content field corresponding to the current multi-source risk data as inputs to determine the spatial propagation relationship and event triggering relationship, and combines the determined results into a source dependency relationship.

[0051] For spatial propagation relationships, the platform pre-configures a spatial propagation rule table, which includes risk source type, propagation direction field, influence radius, adjacent grid range, distance interval, direction matching condition, and propagation attenuation coefficient. The platform reads the current risk source type and queries the corresponding influence radius and direction matching condition in the spatial propagation rule table. For meteorological data, the platform reads wind direction and wind speed, and determines adjacent risk grids located within the influence radius along the wind direction as propagable grids; for communication data, the platform reads the communication obstruction direction, base station coverage area, and link quality degradation area, and determines adjacent risk grids located within the obstruction direction or coverage boundary as propagable grids; for target identification data, the platform reads the position, movement direction, and target speed of suspected UAV targets, obstacle targets, or temporary floating targets, and determines the grids to be passed through based on a preset prediction duration, determining the current position grid and the grids to be passed through as propagable grids; for ground environment data, the platform reads the coverage area of ​​pedestrian flow density, sensitive area range, or temporary activity area range, and determines risk grids that spatially overlap with the coverage area as propagable grids. The platform uses the current action grid as the upstream action grid and the propagable grid as the downstream action grid, and writes the risk source identifier, collection batch identifier, upstream action grid, downstream action grid, propagation direction, propagation distance, propagation attenuation coefficient and effective time period into the spatial propagation relationship.

[0052] For event triggering relationships, the platform pre-configures a preset event triggering table. This table includes trigger source type, trigger risk event type, triggered risk type, applicable grid relationship, trigger distance condition, trigger time condition, trigger value condition, trigger intensity level, and derived base score. The platform uses the current risk source type and current risk event type as query conditions to read candidate triggered risk types from the preset event triggering table. Then, based on the relationship between the current acting grid and the corresponding grid of the candidate triggered risk type, it determines whether the applicable grid relationship and trigger distance conditions are satisfied. Based on the time difference between the current data collection time and the event time corresponding to the candidate triggered risk type, it determines whether the trigger time condition is satisfied. Based on the risk value corresponding to the current data content field, it determines whether the trigger value condition is satisfied. When the applicable grid relationship, trigger distance condition, trigger time condition, and trigger value condition are all satisfied, the platform determines that an event triggering relationship exists and writes the trigger source type, trigger risk event type, triggered risk type, acting grid, triggered acting grid, trigger intensity level, derived base score, and effective time period into the event triggering relationship. When any condition is not satisfied, the platform does not generate the corresponding event triggering relationship.

[0053] The platform combines the spatial propagation relationship and the event triggering relationship into a source dependency relationship. The source dependency relationship includes a risk source identifier, a collection batch identifier, a relationship type, an upstream action grid, a downstream action grid, a triggered risk type, a propagation attenuation coefficient, a trigger intensity level, a derived base score, and an effective time period. The relationship type includes spatial propagation relationships and event triggering relationships. After being bound to the direct contribution value, the source dependency relationship is used in step S103 to generate derived contribution values. Furthermore, when the review status indicates that the risk source has been corrected or negated, the platform can locate homogeneous derived contributions from the contribution ledger that originate from the same risk source and the same collection batch.

[0054] The spatial propagation rule table can be pre-set according to the risk source type, risk mechanism, risk grid size, and low-altitude control response range. Specifically, for meteorological data, the platform sets the propagation direction field, influence radius, and propagation attenuation coefficient based on wind speed level, wind direction stability, and meteorological influence radius; for communication data, the platform sets the adjacent grid range and direction matching conditions based on communication base station coverage, communication obstruction direction, link quality degradation range, and historical communication interruption records; for target identification data, the platform sets the propagable grid range based on target movement speed, movement direction, preset prediction duration, and UAV flight path safety distance; for ground environment data, the platform sets the propagable range based on population density coverage, sensitive area boundaries, temporary activity area range, and grid space boundaries. The propagation attenuation coefficient is set according to the propagation distance range; the larger the propagation distance, the smaller the corresponding propagation attenuation coefficient. When the platform calls the spatial propagation rule table, it uses the risk source type, influence grid, propagation direction, and adjacent grid distance as query conditions to read the corresponding influence radius, direction matching conditions, and propagation attenuation coefficient to generate spatial propagation relationships.

[0055] The preset event trigger table can be pre-set based on low-altitude air traffic control business rules, historical risk event records, manual review records, and UAV operation safety constraints. This table records whether a risk event, when certain distance, time, and numerical conditions are met, will trigger another type of risk. Specifically, communication obstruction corresponds to increased UAV control risk, strong winds correspond to increased flight path deviation risk, a suspected target approaching a logistics route corresponds to increased flight path conflict risk, and increased ground pedestrian density corresponds to increased restricted access control risk. The platform sets trigger distance conditions based on the distance range that triggered subsequent risks in historical events, trigger time conditions based on risk duration and control response time, and trigger numerical conditions based on wind speed, link quality, target identification confidence, or the pedestrian density range. It also configures derived base scores for different trigger intensity levels. When the platform calls the preset event trigger table, it uses the current risk source type, current risk event type, action grid, collection time, and data content fields as query conditions to determine whether the corresponding trigger distance, trigger time, and trigger numerical conditions are met. If the conditions are met, it reads the corresponding trigger intensity level and derived base score to generate event trigger relationships.

[0056] In a specific example, in a mountainous logistics route, the micro-weather station in the upwind risk grid G41 detects a crosswind speed exceeding a preset wind speed range. The platform reads the influence radius, propagation direction field, and propagation attenuation coefficient corresponding to this meteorological source according to the spatial propagation rule table. Since the downwind risk grid G42 is located within the influence radius, and the direction of G42 relative to G41 matches the current wind direction, the platform identifies G41 as the upstream acting grid and G42 as the downstream acting grid. The platform then generates the derived meteorological contribution value in G42 from the direct contribution value of the meteorological source in G41 according to the propagation attenuation coefficient. Furthermore, when the crosswind speed and the UAV flight path direction meet the flight path deviation risk triggering conditions in the preset event triggering table, the platform records the triggered risk type as flight path deviation risk in the source dependency relationship and writes the corresponding derived basic score into the risk source contribution chain.

[0057] Step S103: Generate a risk source contribution chain containing direct contribution values ​​and derived contribution values ​​based on the action grid, the credibility, and the source dependency relationship.

[0058] like Figure 3As shown, the low-altitude airspace control platform first reads the risk source type and risk impact quantity corresponding to the operational grid. The risk impact quantity is obtained through a risk scoring table. The risk scoring table records the basic risk scores corresponding to different risk source types. For example, different meteorological risk scores correspond to different wind speed ranges, different communication risk scores correspond to different communication link quality ranges, and different target identification risk scores correspond to different target identification confidence ranges. The risk scoring table is configured from historical low-altitude operation records, manually labeled samples, and airspace management rules, and includes the risk source type, input value range, basic risk score, and applicable control scenario. When the platform calls the risk scoring table, it uses the risk source type and data content field as input, finds the input value range where the data content field is located, and reads the corresponding basic risk score.

[0059] The low-altitude airspace control platform records the amount of risk impact directly caused by risk sources within the operational grid as the direct contribution value. The direct contribution value is determined by the basic risk score and the spatial overlap ratio. The spatial overlap ratio refers to the proportion of overlap between the coverage area of ​​the risk source and the spatial range of the operational grid. When the risk source covers the entire operational grid, the spatial overlap ratio is 1; when the risk source only covers a part of the operational grid, the spatial overlap ratio is determined according to the proportion of the overlapping area or volume to the spatial range of the operational grid. The platform multiplies the basic risk score by the spatial overlap ratio to obtain the direct contribution value. The direct contribution value is the contribution value before multiplying by the confidence level. The confidence level is bound to the direct contribution value and written into the risk source contribution chain, and is used in step S104 to calculate the effective contribution amount of the source.

[0060] The low-altitude airspace control platform then reads the source dependency relationship. When the source dependency relationship indicates that a risk source propagates to an adjacent risk grid, or triggers other risk events in the same action grid, the platform determines the derived contribution value based on the propagation distance, propagation direction, and trigger type. The derived contribution value is obtained by multiplying the direct contribution value by a preset propagation attenuation coefficient. The preset propagation attenuation coefficient is configured in the propagation attenuation table according to the risk source type and spatial distance range. The propagation attenuation table includes the risk source type, distance range, propagation direction matching status, and attenuation coefficient. When the platform calls the propagation attenuation table, it first determines the risk source type and the distance range of the target action grid, and then determines whether the propagation direction matches the direction of the target action grid. If the direction matches, the corresponding attenuation coefficient is read to calculate the derived contribution value; if the direction does not match, the attenuation coefficient under the direction mismatch state is read to calculate the derived contribution value.

[0061] In one optional implementation, the derived contribution value can also be determined through a preset event trigger table. The preset event trigger table includes trigger source type, triggered risk type, trigger distance condition, trigger time condition, trigger intensity level, and derived base score. The platform reads the trigger source type and triggered risk type from the source dependency relationship, queries the corresponding derived base score from the preset event trigger table, and uses the derived base score as the derived contribution value. The derived contribution value is the contribution value without multiplying by the credibility level. The credibility level is bound to the derived contribution value and written into the risk source contribution chain, and used in step S104 to calculate the effective contribution amount of the source. This derived contribution value establishes a dependency relationship with the corresponding direct contribution from the same source, for use in step S105 to locate the derived contribution from the same source.

[0062] The low-altitude airspace control platform combines risk source identifiers, data collection batch identifiers, action grids, effective time periods, credibility, source dependencies, direct contribution values, and derived contribution values ​​to generate a risk source contribution chain. This risk source contribution chain is then written into the contribution ledger in step S104.

[0063] Step S104: Write the risk source contribution chain into the contribution ledger of the action grid, filter out invalid contribution items according to the effective time period, and generate the current risk level based on the remaining contribution items.

[0064] like Figure 4 As shown, the low-altitude airspace control platform uses the grid number of the operational grid as an index to read the contribution ledger corresponding to that operational grid. The platform writes the risk source contribution chain into the contribution ledger according to the risk source identifier and the collection batch identifier. If there is no contribution item with the same risk source identifier and collection batch identifier in the contribution ledger, a new contribution item is added; if a contribution item with the same risk source identifier and collection batch identifier already exists, the direct contribution value, derived contribution value, effective period, credibility, and source dependency relationship in that contribution item are updated.

[0065] In one optional implementation, when contribution items with the same risk source identifier and collection batch identifier already exist in the contribution ledger, the platform overwrites the old contribution item with the new one and writes the old contribution item to the historical version area. When contribution items with the same risk source identifier but different collection batch identifiers exist in the contribution ledger, the platform saves the contribution items from different batches separately and participates in risk level generation according to their respective effective time periods. Through this process, contribution items generated by the same camera or the same weather station in different time batches will not be deleted due to subsequent review status matching. The historical version area is only used to record the risk source identifier, collection batch identifier, contribution value before coverage, and coverage time of the covered contribution item, and does not participate in the calculation of the current risk level and the corrected risk level. When the subsequent review status requires verification of the covered contribution item, the platform reads the historical version area to generate a traceability prompt, without changing the contribution items in the current contribution ledger based on the historical version area.

[0066] The low-altitude airspace control platform reads each contribution item from the contribution ledger according to the current time, identifies contribution items that do not belong to the corresponding valid time period as invalid contributions, and excludes these invalid contributions from the current risk level calculation. Contribution items that are still within the valid time period are considered remaining contributions.

[0067] For example, within risk grid G15, the effective period for the short-term visibility reduction contribution is from 10:20 to 10:23, the effective period for the communication obstruction contribution is from 10:20 to 10:25, and the effective period for the temporary activity area contribution is from 10:20 to 10:50. When the platform generates the current risk level at 10:26, it marks the short-term visibility reduction and communication obstruction contributions as invalid contributions and excludes them from the current risk level calculation. Meanwhile, since the temporary activity area contribution is still within its effective period, the platform includes this contribution as a remaining contribution in the grid risk contribution calculation.

[0068] The platform reads the direct contribution value, derived contribution value, and credibility of each remaining contribution item, sums the direct and derived contribution values ​​to obtain the total source contribution, and then multiplies the total source contribution by the credibility to obtain the effective source contribution. The platform sums the effective source contributions of multiple sources in the same action grid to obtain the grid risk contribution.

[0069] The low-altitude airspace control platform invokes a preset risk level mapping table to map the grid risk contribution to the current risk level. The preset risk level mapping table includes a lower limit for risk contribution, an upper limit for risk contribution, a risk level, and the corresponding control status. The lower and upper limits for risk contribution are determined by the contribution distribution of historical low-altitude operational events that triggered speed limits, detours, and entry restrictions. When the platform invokes the preset risk level mapping table, it compares the grid risk contribution with each risk contribution interval. When the grid risk contribution is not less than a certain lower limit and less than the corresponding upper limit, the risk level corresponding to that interval is determined as the current risk level. The current risk level is used in step S106 to write to the risk grid, or directly for outputting control commands if no review or correction occurs.

[0070] The preset risk level mapping table can be pre-set based on low-altitude airspace control rules, historical flight risk event records, UAV mission types, and control command triggering conditions. The preset risk level mapping table includes a lower limit for risk contribution, an upper limit for risk contribution, risk level, control status, and corresponding control command type. Specifically, the platform statistically analyzes the grid risk contribution distribution when triggering control actions such as routine passage, speed limit, detour, hovering, return to base, or prohibition of entry during historical low-altitude operations. It then combines this with airspace sensitivity, flight path density, UAV payload type, and manual review results to determine the contribution range corresponding to different risk levels. A higher risk contribution corresponds to a higher risk level and a stricter control status. For sensitive areas such as schools, hospitals, airport airspace protection zones, or large event areas, the platform can set the lower limit for the corresponding risk level lower than that of ordinary areas. For areas with low airspace traffic or only performing inspection tasks, the platform can set the lower limit for the corresponding risk level higher than that of high-density logistics flight path areas. When the platform calls the preset risk level mapping table, it compares the grid risk contribution with each risk contribution range. When the grid risk contribution is not less than a certain risk contribution lower limit and less than the corresponding risk contribution upper limit, the risk level corresponding to that range is determined as the current risk level or the corrected risk level.

[0071] Step S105: Receive a review status that matches the risk source identifier and the collection batch identifier; when the review status indicates that the risk source has been corrected or denied, locate the same-source direct contribution and same-source derived contribution from the contribution ledger according to the source dependency relationship, delete or attenuate the same-source direct contribution and same-source derived contribution and retain the heterogeneous contribution to obtain the corrected contribution ledger.

[0072] like Figure 5As shown, the low-altitude airspace control platform receives the review status; the review status can be generated according to preset review status generation rules. These preset review status generation rules include manual review rules, cross-sensor verification rules, and data correction rules. For manual review rules, the low-altitude airspace control platform pushes the risk source identifier, collection batch identifier, action grid, collection time, image frame, radar spot, on-site feedback information, and current risk level corresponding to the contribution item to be reviewed to the review terminal. Control personnel select to confirm, correct, or deny on the review terminal and enter the review description, the credibility after review, or the review attenuation ratio. The platform combines the above content to generate the manual review status. For cross-sensor verification rules, the platform uses the risk source identifier, collection batch identifier, risk event type, action grid, and collection time of the contribution item to be verified as query conditions to read radar data, visual recognition data, communication monitoring data, and meteorological data within adjacent time windows. When no fewer than the preset number of heterogeneous data sources support the same risk event within the same action grid or adjacent risk grids, the platform generates a confirmation status. When heterogeneous data supports the existence of a risk event but the risk value, target location, or credibility is lower than the corresponding value of the contribution item to be verified, the platform generates a correction status and determines the credibility after verification based on the average or minimum credibility of the heterogeneous data. When none of the heterogeneous data supports the risk event, or when there is heterogeneous data that meets the preset rejection conditions, the platform generates a rejection status. For data correction rules, the platform reads sensor delay data, quality correction data, or false alarm correction data. When the corrected data content field still meets the preset trigger conditions corresponding to the original risk event, the platform generates a confirmation status. When the corrected data content field still meets the preset trigger conditions but the risk value is reduced, the platform generates a correction status and determines the review attenuation ratio based on the ratio of the risk values ​​before and after correction. When the corrected data content field does not meet the preset trigger conditions, the platform generates a rejection status. The review status includes risk source identifier, collection batch identifier, review conclusion, review basis type, review time, and post-review credibility or review attenuation ratio, which are used to subsequently locate direct contributions and derived contributions from the same source and perform maintenance, attenuation, or deletion processing.

[0073] The review status includes at least three states: confirmation, correction, and negation. Confirmation indicates the corresponding risk source remains valid; correction indicates the corresponding risk source still exists but its credibility or contribution value needs to be reduced; negation indicates the corresponding risk source is identified as a false alarm or is invalid. The review status carries the risk source identifier, collection batch identifier, review conclusion, review time, and the credibility or review attenuation ratio after review. The platform first filters contribution items in the contribution ledger by risk source identifier, then locks the target batch by collection batch identifier, and finally searches for contribution items derived from the target batch based on source dependencies. Contribution items directly formed from the risk source corresponding to the target batch are direct contributions from the same source, and derived contribution items formed from the direct contributions from the same source through source dependencies are derived contributions from the same source.

[0074] When the review status is confirmed, the platform keeps the corresponding contribution item in the contribution ledger unchanged. When the review status is corrected, the platform reads the post-review confidence level or review attenuation ratio carried by the review status. When the post-review confidence level is read, the platform multiplies the direct contribution and derived contribution from the same source by the post-review confidence level to obtain the attenuated direct contribution and derived contribution from the same source. When the review attenuation ratio is read, the platform multiplies the direct contribution and derived contribution from the same source by the review attenuation ratio to obtain the attenuated direct contribution and derived contribution from the same source. The platform writes the attenuated direct contribution and derived contribution from the same source into the corrected contribution ledger. For example, if the visual recognition target is confirmed to exist after review but the confidence level is reduced, the platform reads the post-review confidence level or review attenuation ratio carried by the review status and attenuates the direct contribution value corresponding to the visual recognition source and the derived contribution value formed by its propagation.

[0075] For example, during port area inspection routes, the communication monitoring module B04 reported a decline in link quality in data collection batch B04-142000, generating both a direct contribution from the communication source and a derivative contribution from UAV control risk triggered by communication obstruction. Subsequently, the base station maintenance system transmitted data correction results, indicating that the link quality in this batch was not due to communication interruption, but rather link jitter caused by a short-term increase in load. The corrected link quality still met the risk triggering conditions, but the risk value was reduced. Based on this, the platform generated a correction status, carrying a verification attenuation ratio of 0.4. The platform located the direct and derivative contributions from the same communication source according to the risk source identifier B04 and data collection batch B04-142000, multiplied these contributions by 0.4, and wrote them into the correction contribution ledger, while retaining the meteorological and ground environment contributions that were not negated during the same period.

[0076] When the review status is negative, the platform deletes the direct contributions from the same source under the corresponding risk source identifier and collection batch identifier from the contribution ledger, and finds the derived contributions generated from the direct contributions based on the source dependency relationship, and deletes the derived contributions simultaneously. The deleted direct and derived contributions are not written to the correction contribution ledger.

[0077] When deleting or attenuating contributions from the same source, the platform reads contribution items in the contribution ledger whose risk source identifier differs from the corrected or negated risk source, and determines whether the contribution item is within a valid time period and whether there is a negative status review record matching the risk source identifier and collection batch identifier of the contribution item. If the contribution item is within a valid time period and there is no negative status review record matching the risk source identifier and collection batch identifier of the contribution item, then the contribution item is written as a heterogeneous contribution into the correction contribution ledger. The correction contribution ledger includes attenuated contributions from the same source and retained heterogeneous contributions; when the review status is negative, the correction contribution ledger includes retained heterogeneous contributions. The correction contribution ledger is used in step S106 to generate the corrected risk level.

[0078] Step S106: Generate a revised risk level based on the revised contribution ledger, write the current risk level or the revised risk level into the risk grid, obtain the risk grid update result, and output control instructions based on the risk grid update result.

[0079] like Figure 6 As shown, the low-altitude airspace control platform reads each contribution item from the correction contribution ledger. For each contribution item in the correction contribution ledger, the platform reads the direct contribution value, derived contribution value, and confidence level. It sums the direct and derived contribution values ​​to obtain the total correction source contribution, and then multiplies this total correction source contribution by the corresponding confidence level to obtain the effective correction source contribution. The platform sums the effective correction source contributions from multiple sources within the same action grid to obtain the correction grid risk contribution. The platform then calls a preset risk level mapping table to map the correction grid risk contribution to a correction risk level.

[0080] When the correction contribution ledger is empty, the platform reads the default control status configuration table corresponding to the risk grid. This default control status configuration table includes the region type, drone mission type, default risk level, and default control status. The platform uses the region type and drone mission type of the risk grid as query conditions to read the corresponding default risk level and default control status. When the default control status is "de-control," the platform writes the corrected risk level to "low risk" and the control status to "de-control." When the default control status is "retained inspection," the platform writes the corrected risk level to "low risk" and the control status to "regular inspection." The above processing results are written into the risk grid update results.

[0081] The default control status configuration table can be pre-set according to the risk grid's area type, UAV mission type, airspace sensitivity, and low-altitude control safety requirements. The default control status configuration table includes area type, UAV mission type, default risk level, default control status, and default command type. Specifically, for sensitive areas such as schools, hospitals, airport airspace protection zones, large event areas, or temporary control zones, even if the correction contribution ledger is empty, the platform can set the default control status to routine inspection or low-level speed limit to retain basic control. For ordinary parks, low-density logistics routes, or open airspace, when the correction contribution ledger is empty and there are no valid external contributions, the platform can set the default control status to de-control. For heavy logistics, emergency rescue, or inspection missions, the platform configures different default risk levels and default command types according to the mission's risk tolerance. When the platform calls the default control status configuration table, it uses the risk grid's area type and UAV mission type as query conditions to read the corresponding default risk level, default control status, and default command type, and generates a risk grid update result when the correction contribution ledger is empty.

[0082] In one optional implementation, the platform further sets preset mandatory retention conditions. These preset mandatory retention conditions include a preset mandatory retention threshold and a preset source quantity threshold. The preset mandatory retention threshold is determined based on the lower limit of the effective contribution of sources requiring retention for speed limits, detours, or no-entry measures in historical low-altitude operation records; the preset source quantity threshold is determined based on the minimum number of risk sources requiring coordinated control according to airspace management rules. The platform reads the heterogeneous source contributions from the corrected contribution ledger, calculates the effective contribution of the corresponding source, and counts the number of risk sources corresponding to the heterogeneous source contributions. When the effective contribution of a source in a heterogeneous contribution is not less than the preset mandatory retention threshold, or the number of risk sources corresponding to a heterogeneous contribution is not less than the preset source quantity threshold, the platform compares the corrected risk level with the preset minimum control level; if the corrected risk level is lower than the preset minimum control level, the corrected risk level is raised to the preset minimum control level; if the corrected risk level is not lower than the preset minimum control level, the corrected risk level remains unchanged. The preset minimum control level is used to prevent the remaining heterogeneous source risks from being eliminated after a false alarm source is deleted.

[0083] The platform reads heterogeneous contribution items from the correction contribution ledger. A heterogeneous contribution item refers to a contribution item whose risk source identifier differs from the corrected or negated risk source, and that the contribution item is within its valid time period and has not been negated by the corresponding review status. For each heterogeneous contribution item, the platform reads its direct contribution value, derived contribution value, and credibility. It adds the direct contribution value and the derived contribution value to obtain the total heterogeneous source contribution. Then, it multiplies the total heterogeneous source contribution by the corresponding credibility to obtain the effective source contribution amount corresponding to the heterogeneous contribution item. If multiple heterogeneous contribution items within the same effective time period exist for the same risk source identifier in the same action grid, the platform calculates the effective source contribution amount for each heterogeneous contribution item separately and adds them together to obtain the effective heterogeneous source contribution amount corresponding to the risk source identifier. If only one heterogeneous contribution item exists for the same risk source identifier, the effective source contribution amount corresponding to that heterogeneous contribution item is used as the effective heterogeneous source contribution amount corresponding to the risk source identifier. The platform counts the number of different risk source identifiers in the heterogeneous contribution items to obtain the number of risk sources corresponding to the heterogeneous contribution. When the same risk source identifier has valid heterogeneous contribution items in multiple collection batches, it is counted as one risk source. The platform compares the valid heterogeneous contribution amount corresponding to each risk source identifier with a preset mandatory retention threshold. When the valid heterogeneous contribution amount corresponding to any risk source identifier is not less than the preset mandatory retention threshold, or the number of risk sources corresponding to the heterogeneous contribution is not less than the preset source quantity threshold, the heterogeneous contribution is determined to meet the preset mandatory retention condition.

[0084] The low-altitude airspace control platform determines the risk level to be written into the risk grid based on whether a review and correction have occurred. When no review status is received or the review status is confirmed, the platform writes the current risk level into the risk grid. When the review status is corrected or negative, the platform writes the corrected risk level into the risk grid. The data written into the risk grid includes the grid number, risk level, control status, update time, contribution item identifiers involved in the calculation, and review status identifier. The above data constitutes the risk grid update result.

[0085] The low-altitude airspace control platform outputs control instructions based on the risk grid update results; control instructions include detour instructions, speed limit instructions, hovering instructions, return-to-base instructions, no-entry instructions, or de-control instructions; the platform calls the control instruction mapping table to map the risk level and control status to the corresponding control instructions.

[0086] The control command mapping table can be pre-set according to low-altitude airspace control rules, UAV mission type, risk level, control status, and flight safety constraints. The control command mapping table includes risk level, control status, UAV mission type, command type, command parameters, and command validity duration. Specifically, the platform configures corresponding control actions based on different risk levels. When the risk level is low and the control status is de-controlled, the command type can be set to de-controlled. When the risk level is medium, the command type can be set to speed limit, altitude adjustment, or detour. When the risk level is high, the command type can be set to hover, return to home, or prohibition of entry. For different mission types such as heavy logistics, emergency rescue, and inspection, the platform sets command parameters such as maximum permissible speed, detour distance, hovering duration, return to home trigger conditions, and prohibition of entry range. For sensitive areas such as schools, hospitals, airport airspace protection zones, large event areas, or temporary control areas, the platform can set the control commands corresponding to the same risk level to more stringent command types. When the platform calls the control command mapping table, it uses risk level, control status and UAV mission type as query conditions to read the corresponding command type, command parameters and command validity duration, and generates control commands to be sent to the UAV flight control system, low-altitude traffic management platform or route scheduling system.

[0087] In a specific example, a risk grid above a city's commercial area receives four types of risk data within the same update cycle: strong wind data reported by a micro-weather station, link quality degradation data reported by a communication monitoring module, increased pedestrian density data reported by a ground environment interface, and suspected illegal flying targets reported by a visual camera. The platform generates contributions from meteorological sources, communication sources, pedestrian sources, and visual identification sources, and writes these contributions into the risk grid's contribution ledger according to the risk source identifier and collection batch identifier. Since multiple contributions overlap to reach a high-risk range, the platform updates the risk grid to high risk and outputs a detour or prohibition instruction. Subsequently, manual review indicates that the suspected illegal flying target identified visually is actually a reflection from a tower crane boom. Based on the review status, the platform locates the direct contribution from the same source as the visual identification source and deletes the derived contribution from the same source based on source dependency, while retaining the strong wind contribution, communication contribution, and pedestrian contribution within the valid time period. The platform regenerates a corrected risk level based on the corrected contribution ledger. If the contribution from strong winds and the contribution from communication still meet the preset mandatory retention conditions, the risk grid will not be directly downgraded to low risk, but will remain at a state no lower than the preset minimum control level, and will continue to output speed limit or detour instructions.

[0088] For a further example, the contribution ledger records that collection batch A1 from meteorological source A corresponds to a direct contribution value of 30 and a derived contribution value of 5; collection batch B1 from communication source B corresponds to a direct contribution value of 20 and a derived contribution value of 3; and collection batch C1 from visual source C corresponds to a direct contribution value of 35 and a derived contribution value of 10. When visual source C is rejected upon review, the platform locates the contribution item for that batch according to the risk source identifier C and collection batch C1, deletes the direct contribution value of 35 and the derived contribution value of 10 corresponding to collection batch C1, retains the contribution items from meteorological source A and communication source B that are within the valid time period and have not been rejected, and regenerates the revised risk level based on the retained contribution items. In this example, each contribution item in the contribution ledger participates in the calculation of the current risk level or the revised risk level, and the deletion or attenuation actions have a clear target and subsequent purpose.

[0089] like Figure 7 As shown in the illustration, this application also provides a risk grid update system for low-altitude management of unmanned aerial vehicles (UAVs). This system includes a data acquisition module, a data parsing module, a contribution chain generation module, a ledger update module, a review and correction module, and an instruction output module.

[0090] The data acquisition module is used to acquire multi-source risk data from the risk grid and send the multi-source risk data to the data parsing module. The multi-source risk data includes at least two of the following: meteorological data, communication data, ground environment data, and target identification data.

[0091] The data parsing module is used to parse the multi-source risk data to obtain the risk source identifier, collection batch identifier, action grid, effective time period, credibility and source dependency relationship, and send the parsing results to the contribution chain generation module.

[0092] The contribution chain generation module is used to generate a risk source contribution chain containing direct contribution values ​​and derived contribution values ​​based on the action grid, the credibility, and the source dependency, and send the risk source contribution chain to the ledger update module.

[0093] The ledger update module is used to write the risk source contribution chain into the contribution ledger of the action grid, filter out invalid contribution items according to the effective time period, and generate the current risk level based on the remaining contribution items.

[0094] The review and correction module is used to receive the review status that matches the risk source identifier and the collection batch identifier, and when the review status indicates that the risk source has been corrected or denied, it locates the same-source direct contribution and same-source derived contribution from the contribution ledger according to the source dependency relationship, deletes or attenuates the same-source direct contribution and same-source derived contribution, and retains the heterogeneous contribution that is within the valid time period and has not been denied, so as to obtain the corrected contribution ledger.

[0095] The instruction output module is used to generate a corrected risk level based on the corrected contribution ledger, write the current risk level or the corrected risk level into the risk grid, generate a risk grid update result, and output control instructions to the UAV flight control system, low-altitude traffic management platform or flight route scheduling system.

[0096] In one implementation, the system is deployed on a low-altitude airspace control server, or it can be deployed on a park low-altitude management platform, an urban low-altitude traffic management platform, or a logistics drone dispatch platform. Each module in the system is implemented by software programs, or by a combination of software and hardware. The modules are located on the same computer device, or distributed across multiple computer devices, and interact with each other via a network.

[0097] This application also provides a computer device, which includes a processor, a memory, a communication interface, and a bus. The processor, memory, and communication interface are connected via the bus. The memory is used to store computer programs and contribution ledger data. The communication interface is used to receive multi-source risk data, review status, and send control commands. The processor is used to execute the computer program to implement the above-described method for updating the risk grid for low-altitude drone control.

[0098] This application also provides a computer-readable storage medium, which is a read-only memory, random access memory, magnetic disk, optical disk, flash memory, or other medium capable of storing a computer program. When the computer program is executed by a processor, it implements the above-described method for updating the risk grid for low-altitude drone management.

[0099] It should be noted that, in this document, the terms "comprising," "including," and any other variations are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Specific examples have been used in this document to illustrate the principles and implementation methods of the technical solutions of this application. The above examples are only for the purpose of helping to understand the methods and core ideas of this application. The above descriptions are merely preferred embodiments of this application. It should be pointed out that, due to the limitations of written expression and the objective existence of infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of this application, and can also combine the above technical features in an appropriate manner; these improvements, modifications, changes, or combinations, or the direct application of the concept and technical solutions of this application to other situations without modification, should all be considered within the scope of protection of this application.

Claims

1. A risk grid update method for low-altitude management and control of unmanned aerial vehicles (UAVs), characterized in that, include: Acquire multi-source risk data from the risk grid; The multi-source risk data is analyzed to obtain the risk source identifier, collection batch identifier, action grid, effective time period, credibility, and source dependency relationship; Based on the action grid, the credibility, and the source dependency, a risk source contribution chain containing direct contribution values ​​and derived contribution values ​​is generated. Write the risk source contribution chain into the contribution ledger of the action grid according to the risk source identifier and the collection batch identifier, filter out the invalid contribution items according to the effective time period, and generate the current risk level based on the remaining contribution items; Receive the verification status that matches the risk source identifier and the collection batch identifier; When the review status indicates that the risk source has been corrected or denied, the same-source direct contribution and same-source derived contribution are located from the contribution ledger according to the source dependency relationship. The same-source direct contribution and same-source derived contribution are deleted or attenuated while the heterogeneous contribution is retained to obtain the corrected contribution ledger. Based on the revised contribution ledger, a revised risk level is generated. The current risk level or the revised risk level is written into the risk grid to obtain the risk grid update result. Control instructions are output based on the risk grid update result.

2. The method according to claim 1, characterized in that, Obtain multi-source risk data from the risk grid, including: Acquire at least two of the following types of data corresponding to the risk grid: meteorological data, communication data, ground environment data, and target identification data; wherein, the meteorological data includes wind speed, wind direction, precipitation, or visibility; the communication data includes link quality or communication obstruction status; the ground environment data includes population density or sensitive area information; and the target identification data includes the identification results of suspected drone targets, obstacle targets, or temporary floating object targets.

3. The method according to claim 1, characterized in that, The analysis of the multi-source risk data includes: Read the data source field and collection time field carried by the multi-source risk data, convert the data source field into a risk source identifier, and write data from the same source within the same collection window into the same collection batch identifier; The action grid is determined based on the location coordinates, coverage area, and spatial boundary of the risk grid corresponding to the multi-source risk data; The effective time period is determined based on the collection time and the effective duration of the source.

4. The method according to claim 1, characterized in that, The reliability is determined by the sensor source level, the freshness of the data acquisition time, and the cross-validation results. The confidence level is obtained by weighting the basic confidence value corresponding to the sensor source level, the time decay value from the acquisition time to the current time, and the verification values ​​of other sources for the same risk event. When the cross-validation result indicates that the same risk event is supported by at least one heterogeneous data source, the credibility is increased by a preset credibility increment according to a preset credibility correction rule, and the increased credibility is not greater than a preset credibility upper limit. The credibility is used to generate direct contribution values ​​and derived contribution values.

5. The method according to claim 1, characterized in that, The source dependency relationship includes the spatial propagation relationship of the same risk source to adjacent risk grids and the triggering relationship between different risk events; When generating the risk source contribution chain, the risk impact amount located within the action grid is recorded as the direct contribution value, and the risk impact amount transmitted to the action grid by the spatial propagation relationship or the triggering relationship is recorded as the derived contribution value.

6. The method according to claim 1, characterized in that, The contribution ledger includes multiple contribution items, each of which includes a risk source identifier, a collection batch identifier, a direct contribution value, a derived contribution value, a credibility level, an effective period, and a source dependency relationship. When filtering out invalid contributions according to the effective time period, contributions that do not belong to the corresponding effective time period at the current time are marked as invalid contributions and excluded from the current risk level calculation, while contributions that belong to the corresponding effective time period at the current time are treated as remaining contributions.

7. The method according to claim 1, characterized in that, The process of generating the current risk level based on the remaining contribution items includes: Read the direct contribution value, derived contribution value, and credibility from the remaining contribution items; sum the direct contribution value and the derived contribution value to obtain the total source contribution; multiply the total source contribution by the corresponding credibility to obtain the effective source contribution. The effective contributions from multiple sources within the same functional grid are summed to obtain the grid risk contribution, and the grid risk contribution is mapped to the current risk level according to a preset risk level mapping table.

8. The method according to claim 1, characterized in that, The verification status is generated from the results of manual verification, cross-sensor verification, or data correction, and includes a confirmation status, a correction status, and a rejection status. When the review status is confirmed, the corresponding contribution item in the contribution ledger remains unchanged; When the review status is in the correction status, read the post-review credibility or review attenuation ratio carried by the review status; When the verified credibility is read, the direct contribution and the derived contribution are multiplied by the verified credibility to obtain the attenuated contribution. When the core attenuation ratio is read, the direct contribution and the derived contribution are multiplied by the core attenuation ratio to obtain the attenuated contribution. When the review status is negative, the corresponding direct contribution and derived contribution from the same source are deleted from the contribution ledger.

9. The method according to claim 1, characterized in that, Retain heterogeneous contributions, including: Read contribution items in the contribution ledger whose risk source identifier is different from the corrected or negated risk source, and retain contribution items that are within the valid period and do not have a negative status review record that matches the risk source identifier and collection batch identifier of the contribution item as heterogeneous contributions in the correction contribution ledger; When the effective contribution amount of the heterogeneous contribution is not less than the preset mandatory retention threshold, or when the number of risk sources corresponding to the heterogeneous contribution is not less than the preset source quantity threshold, the corrected risk level is not lower than the preset minimum control level.

10. A risk grid update system for low-altitude management of unmanned aerial vehicles (UAVs), implementing the method described in any one of claims 1 to 9, characterized in that, include: The data acquisition module is used to acquire multi-source risk data from the risk grid; The data parsing module is used to parse the multi-source risk data to obtain the risk source identifier, collection batch identifier, action grid, effective time period, credibility, and source dependency relationship; The contribution chain generation module is used to generate a risk source contribution chain containing direct contribution values ​​and derived contribution values ​​based on the action grid, the credibility, and the source dependency relationship. The ledger update module is used to write the risk source contribution chain into the contribution ledger of the action grid according to the risk source identifier and the collection batch identifier, filter out invalid contribution items according to the effective time period, and generate the current risk level based on the remaining contribution items; The review and correction module is used to receive the review status that matches the risk source identifier and the collection batch identifier; When the review status indicates that the risk source has been corrected or denied, the same-source direct contribution and same-source derived contribution are located from the contribution ledger according to the source dependency relationship. The same-source direct contribution and same-source derived contribution are deleted or attenuated while the heterogeneous contribution is retained to obtain the corrected contribution ledger. The instruction output module is used to generate a revised risk level based on the revised contribution ledger, write the current risk level or the revised risk level into the risk grid, obtain the risk grid update result, and output control instructions based on the risk grid update result.

Images