Method and system for rapidly monitoring radiation risk of unmanned aerial vehicle

Abstract

The invention discloses a method and system for rapidly monitoring the radiation risk of an unmanned aerial vehicle, and relates to the technical field of unmanned aerial vehicle radiation monitoring. The method is used for solving the problems of high-altitude detection attenuation distortion, difficult standard-exceeding source accurate positioning and weak network data loss. Firstly, an unmanned aerial vehicle is controlled to execute gridding high-altitude scanning in a terrain following mode, and the large-range screening efficiency is ensured; then, earth surface equivalent source intensity data are dynamically reconstructed in combination with elevation fluctuation characteristics and distance attenuation rules, and abnormal nodes are accurately locked by comparing dynamic thresholds; the locked node triggers height reduction retest, the radiation extreme value is accurately anchored in a target abnormal pixel area through visual kernel feature correlation analysis, and physical pollution source visual consolidation is achieved; finally, data transmission is guaranteed by adopting a self-adaptive cache and multi-terminal distribution mechanism based on network delay, a linkage closed loop from air discovery to field rectification is constructed, and reliable support is provided for radiation safety monitoring of a complex environment.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) radiation monitoring technology, specifically to a method and system for rapid monitoring of UAV radiation risk. Background Technology

[0002] With the deepening development of nuclear energy and related mining and metallurgical industries, environmental radiation monitoring of work areas has become a core component in ensuring occupational health and environmental safety. Traditional ground-based manual surveying methods have limitations such as high exposure risks and susceptibility to complex terrain. In recent years, integrating radiation detection modules into multi-rotor UAV platforms for large-scale, rapid screening has become an important trend in the industry. UAVs possess excellent high mobility, enabling them to replace manual labor in conducting grid-based patrol measurements deep into hazardous areas, greatly improving the efficiency and coverage of radiation environmental monitoring.

[0003] However, existing UAV radiation monitoring methods have significant inherent limitations in complex real-world scenarios. First, because radiation naturally attenuates with distance in the air, existing methods suffer from severe attenuation and distortion of the radiation signal received by the detector during high-altitude flights, failing to accurately reflect the intensity of the surface source. Furthermore, continuously lowering the flight altitude not only causes severe dust interference with monitoring due to the strong airflow generated by the UAV rotors, but also poses a risk of collision and crash in areas with significant terrain undulations. Second, when existing monitoring systems detect radiation exceedances, they can only record abstract data points, making it difficult to visually link invisible radiation peaks with actual pollution source entities in the terrain, leading to difficulties in subsequent remediation and location. Finally, in long-distance, complex monitoring scenarios, communication links are easily disrupted by terrain obstruction. Existing methods are prone to alarm data loss in weak network environments, and warning information typically only travels one-way back to the main control unit, failing to immediately connect with on-site personnel, resulting in a severe disconnect between warning detection and on-site hazard mitigation. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method and system for rapid monitoring of radiation risks from unmanned aerial vehicles (UAVs), thus solving the problems mentioned in the background.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for rapid monitoring of radiation risk from unmanned aerial vehicles (UAVs), comprising: S1. controlling a UAV in terrain-following mode to perform gridded flight at a first preset altitude, and acquiring three-dimensional position coordinates and timestamped first radiation dose rate data through the low-range channel of a dual-range GM detector; S2. extracting elevation features of the three-dimensional position coordinates, calculating the elevation fluctuation variation coefficient, dynamically updating the spatial compensation weights by combining the inverse square law of distance attenuation, and compensating and mapping the first radiation dose rate data to reconstruct the equivalent source strength data of the ground surface; if the reconstructed equivalent source strength data of the ground surface exceeds the dynamic warning threshold constructed by fusing historical background and preset standard deviation, then marking the corresponding coordinates as abnormal nodes and generating a target area retest command; S3. responding to the retest command, The UAV is controlled to descend to the second preset altitude and hover at the abnormal node. The detector is switched to the high-range channel to acquire the second radiation dose rate data sequence. The timestamps representing radiation extrema in the sequence are extracted and associated image frames are captured. The nuclear feature correlation analysis is performed to anchor the radiation extrema space to the target abnormal pixel region within the associated image frame. The anchoring result, associated image frame, and current three-dimensional position coordinates are packaged into a retest data packet. S4. The network delay feature parameters are acquired. If the delay exceeds the preset threshold, the retest data packet is written to the onboard asynchronous cache. The upload is triggered when the network delay feature parameters fall back to the preset delay threshold. If the preset delay threshold is met, the data packet is synchronously distributed to the ground control terminal and the field terminal. In response to the rectification confirmation signal from the field terminal, the UAV is controlled to return to the interrupted position and resume gridded flight.

[0006] Furthermore, the specific process of controlling the UAV entering terrain-following mode to perform gridded flight at a first preset altitude and collecting three-dimensional position coordinates and timestamped first radiation dose rate data through the low-range channel of the dual-range GM detector is as follows: acquire three-dimensional terrain reference profile data of the target monitoring area, and generate a terrain-following guidance trajectory in conjunction with the real-time centimeter-level positioning signal output by the airborne RTK; control the airborne flight control unit to drive the UAV along the terrain-following guidance trajectory and maintain the first preset altitude to perform gridded cruise; configure the dual-range GM detector to lock the low-range detection channel and initialize a fixed sampling period; extract the absolute timestamp of the airborne time reference within each fixed sampling period; spatiotemporally bind the absolute timestamp, the first radiation dose rate data synchronously output by the low-range detection channel, and the three-dimensional position coordinates synchronously output by the airborne RTK, and encapsulate them into a low-level raw data packet.

[0007] Furthermore, the elevation features of the three-dimensional position coordinates are extracted to calculate the elevation fluctuation variation coefficient. Combined with the inverse square law of distance attenuation, the spatial compensation weights are dynamically updated. The specific process of compensating and mapping the first radiation dose rate data to reconstruct the equivalent source strength data of the Earth's surface is as follows: The elevation features of the three-dimensional position coordinates in multiple consecutive bottom-level raw data packets within a preset time window are extracted to construct a dynamic elevation floating sequence. The deviation of the dynamic elevation floating sequence from the first preset altitude is extracted, and the preset steady-state parameters of the UAV's flight attitude are fused to determine the elevation fluctuation variation coefficient. Based on the instantaneous ground-free altitude contained in the dynamic elevation floating sequence, a preset inverse square law of distance attenuation is mapped to extract the corresponding benchmark attenuation factor. The nonlinear feature correction of the benchmark attenuation factor is performed using the elevation fluctuation variation coefficient, and the spatial compensation weights are output. The spatial compensation weights are then fused with the first radiation dose rate data at the data level, and the mapping is reversed to reconstruct the equivalent source strength data of the Earth's surface.

[0008] Furthermore, if the reconstructed surface equivalent source strength data exceeds the dynamic warning threshold constructed by fusing historical baseline and preset standard deviation, the specific process of marking the corresponding coordinates as abnormal nodes and generating target area retesting instructions is as follows: retrieve historical radiation baseline data of the target monitoring area under normal conditions, extract the dispersion characteristics of the historical radiation baseline data to generate a standard deviation base, introduce a preset multiplier parameter to scale the standard deviation base, perform fusion reconstruction with the scaled standard deviation base and historical radiation baseline data, and output the dynamic warning threshold; establish a data comparison logic determiner, input the surface equivalent source strength data into the data comparison logic determiner and perform exceedance comparison detection with the dynamic warning threshold; when the data comparison logic determiner outputs an exceedance determination signal, mark the three-dimensional position coordinates bound to the surface equivalent source strength data as abnormal nodes, write the abnormal node coordinates into a preset flight control instruction message format, and generate a target area retesting instruction containing a second preset altitude reduction instruction and a detector range switching trigger code.

[0009] Furthermore, in response to the retest command, the specific process of controlling the UAV to descend to the second preset altitude and hover at the abnormal node, and switching the detector to the high-range channel to obtain the second radiation dose rate data sequence is as follows: The retest command for the target area is analyzed to extract the coordinates of the abnormal node and the altitude reduction trigger code. The three-dimensional coordinate breakpoint features of the current high-altitude gridded flight interruption position of the UAV are recorded, and the airborne multi-directional visual obstacle avoidance sensor array is activated simultaneously. Combining the coordinates of the abnormal node and the spatial obstacle distribution features output by the airborne multi-directional visual obstacle avoidance sensor array, an obstacle avoidance vertical descent trajectory is generated, and the UAV is driven to descend to the second preset altitude along the obstacle avoidance vertical descent trajectory and lock the hovering attitude. A hardware channel switching control signal is sent to the dual-range GM detector to cut off the data stream of the low-range measurement channel and activate the high-range measurement channel, and the high-frequency continuous sampling period is initialized to extract the second radiation dose rate data sequence.

[0010] Further, the process of extracting timestamps representing radiation extrema from the sequence and extracting associated image frames, performing kernel feature association analysis, spatially anchoring radiation extrema to the target anomalous pixel region within the associated image frame, and packaging the anchoring result, associated image frame, and current 3D position coordinates into a retest data package is as follows: Parse the second radiation dose rate data sequence, extract radiation extrema with local maximum features, retrieve the absolute timestamp bound to the radiation extrema in the hardware clock register, input the absolute timestamp as the retrieval key into the video stream buffer queue of the airborne camera component, extract video frames with temporal consistency features to construct associated image frames; perform terrain anomaly visual recognition analysis on the associated image frames, output the target anomalous pixel region, perform kernel feature association analysis, perform 2D spatial mapping and superposition of the radiation extrema and the target anomalous pixel region to generate spatial anchoring results; perform multi-source data fusion encapsulation on the spatial anchoring results, associated image frames, radiation extrema, and the current 3D position coordinates output by the airborne RTK at the second preset altitude to generate a retest data package.

[0011] Further, the network latency characteristic parameters are obtained. If they exceed a preset latency threshold, the retest data packets are written to the onboard asynchronous storage cache. The specific process of triggering the upload when the network latency characteristic parameters fall back to within the preset latency threshold is as follows: Heartbeat detection messages are continuously sent to the ground control terminal through the onboard communication unit. The round-trip response time of the heartbeat detection messages is extracted and reconstructed into network latency characteristic parameters. State machine logic judgment rules are introduced, and the network latency characteristic parameters are input into the state machine logic judgment rules for interval comparison. When the state machine logic judgment rules output a network disconnection status signal that exceeds the preset latency threshold, the local file system write permission of the onboard storage unit is activated, and the retest data packets are appended to the tail of the preset asynchronous cache queue. A background continuous monitoring mechanism for network latency characteristic parameters is maintained. When the network latency characteristic parameters are detected to be within the preset latency threshold range within a continuous preset time window, historical retest data packets are sequentially extracted from the head of the asynchronous cache queue to trigger the upload.

[0012] Furthermore, if the preset delay threshold is met, the data is simultaneously distributed to the ground control terminal and the field terminal. The specific process of controlling the UAV to return to the interrupted position and resume gridded flight in response to the rectification confirmation signal from the field terminal is as follows: When the state machine logic judgment rule outputs a normal state signal that meets the preset delay threshold, a dual-link communication channel is established, and the retest data packet is synchronously pushed to the ground control terminal database and the construction site terminal interaction interface through the dual-link communication channel; the UAV is kept in a hovering monitoring state at the second preset altitude until the airborne communication unit parses the rectification confirmation signal triggered and reported by the construction site terminal based on physical manual interaction operation; the three-dimensional coordinate features of the stored gridded flight interruption position are extracted to generate a return recovery trajectory, and the UAV is driven to climb along the return recovery trajectory to the spatial position where the interruption position is located, and the UAV is controlled to return to the interruption position to resume gridded flight.

[0013] A system for rapid monitoring of radiation risk from unmanned aerial vehicles (UAVs), used to execute the aforementioned method for rapid monitoring of radiation risk from UAVs, includes: a high-altitude initial screening module for controlling a UAV entering terrain-following mode to perform gridded flight at a first preset altitude, acquiring three-dimensional position coordinates and timestamped first radiation dose rate data through the low-range channel of a dual-range GM detector; a dynamic compensation and early warning module for extracting elevation features of the three-dimensional position coordinates to calculate the elevation fluctuation variation coefficient, dynamically updating the spatial compensation weights based on the inverse square law of distance attenuation, and compensating and mapping the first radiation dose rate data to reconstruct the equivalent source strength data of the ground surface; if the reconstructed equivalent source strength data of the ground surface exceeds the dynamic early warning threshold constructed by fusing historical background and preset standard deviation, the corresponding coordinates are marked as abnormal nodes and a target area retest command is generated; a visual nucleus fusion retest module, using... In response to the retest command, the UAV is controlled to descend to the second preset altitude and hover at the abnormal node. The detector is switched to the high-range channel to acquire the second radiation dose rate data sequence. The timestamps representing radiation extrema in the sequence are extracted and associated image frames are captured. The nuclear feature correlation analysis is performed to anchor the radiation extrema space to the target abnormal pixel region within the associated image frame. The anchoring result, associated image frame, and current three-dimensional position coordinates are packaged into a retest data packet. The task linkage closed-loop module is used to acquire network latency characteristic parameters. If the latency exceeds the preset threshold, the retest data packet is written to the onboard storage asynchronous cache. The upload is triggered when the network latency characteristic parameters fall back to the preset latency threshold. If the preset latency threshold is met, it is synchronously distributed to the ground control terminal and the field terminal. In response to the rectification confirmation signal from the field terminal, the UAV is controlled to return to the interrupted position and resume gridded flight.

[0014] The present invention has the following beneficial effects: (1) A method for rapid monitoring of radiation risk by unmanned aerial vehicles (UAVs) involves controlling the UAV to enter terrain-following mode, performing gridded flight at a first preset altitude, and opening a low-range detection channel. The extracted elevation fluctuation variation coefficient and the inverse square law of distance attenuation are used to dynamically update the spatial compensation weights to reconstruct the equivalent source strength data of the ground surface. This completely overcomes the measurement distortion problem caused by the natural attenuation of radiation from high-altitude detection, restoring weak airborne radiation readings to the true source strength of the ground surface, and comparing them with a dynamic early warning threshold generated by fusing historical background data. This mechanism, while ensuring flight safety, greatly improves the accuracy of large-scale early radiation risk screening and avoids false alarms and missed alarms caused by terrain undulations.

[0015] (2) A system for rapid monitoring of radiation risk from unmanned aerial vehicles (UAVs), which, upon responding to a target area retest command, automatically lowers the UAV to a second preset altitude and switches to a high-range detection channel. It extracts radiation extreme value timestamps and extracts associated image frames to perform visual kernel feature correlation analysis, anchoring the radiation extreme value space to the target's abnormal pixel area. Simultaneously, it utilizes an asynchronous caching and multi-terminal distribution mechanism with a network latency state machine. This architecture achieves visual locking of physical pollution sources and accurate evidence consolidation of multi-source data at the system detection level, while ensuring the integrity of early warning data in complex weak network environments at the communication transmission level. It completely establishes a closed-loop system-level business management system, from aerial discovery of potential hazards, multi-terminal data distribution, on-site terminal guidance for hazard mitigation, to the automatic resumption of UAV patrols.

[0016] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0017] Figure 1 This is a flowchart of a method for rapid monitoring of radiation risk from unmanned aerial vehicles (UAVs) according to the present invention.

[0018] Figure 2 This is a system flowchart for rapid monitoring of radiation risks from unmanned aerial vehicles (UAVs) according to the present invention. Detailed Implementation

[0019] This application provides a method and system for rapid monitoring of radiation risks from unmanned aerial vehicles (UAVs), which solves the problems of attenuation and distortion in high-altitude detection in complex terrain, difficulty in accurately locating points exceeding the standard, and fragmentation of early warning services in weak network environments.

[0020] The overall concept of the solution in this application embodiment is as follows: The drone, controlled to enter terrain-following mode, performs gridded flight at a first preset altitude, acquiring three-dimensional position coordinates and timestamped first radiation dose rate data through the low-range channel of the dual-range GM detector. The elevation features of the three-dimensional position coordinates are extracted to calculate the elevation fluctuation variation coefficient. Combined with the inverse square law of distance attenuation, the spatial compensation weight is dynamically updated to compensate and map the first radiation dose rate data to reconstruct the equivalent source strength data of the ground surface. If the reconstructed equivalent source strength data of the ground surface exceeds the dynamic warning threshold constructed by fusing historical background and preset standard deviation, the corresponding coordinates are marked as an abnormal node, and a target area retest command is generated. In response to the retest command, the drone is controlled to descend to a second preset altitude and hover at the abnormal node, switching detection... The device acquires the second radiation dose rate data sequence through the high-range channel; it extracts the timestamps representing radiation extrema from the sequence and extracts associated image frames, performs nuclear feature correlation analysis, anchors the radiation extrema space to the target abnormal pixel region within the associated image frame, and packages the anchoring result, associated image frame, and current 3D position coordinates into a retest data packet; it acquires network delay characteristic parameters, and if the delay exceeds a preset threshold, it writes the retest data packet to the onboard asynchronous cache, triggering upload when the network delay characteristic parameters fall back to the preset delay threshold; if the preset delay threshold is met, it synchronously distributes the data to the ground control terminal and the field terminal; in response to the rectification confirmation signal from the field terminal, it controls the UAV to return to the interrupted position and resume gridded flight.

[0021] Example 1; please refer to Figure 1This invention provides a technical solution: a method for rapid monitoring of radiation risk from unmanned aerial vehicles (UAVs), comprising: S1. controlling the UAV, which has entered terrain-following mode, to perform gridded flight at a first preset altitude, and acquiring three-dimensional position coordinates and a first radiation dose rate data with timestamps through the low-range channel of a dual-range GM detector; S2. extracting the elevation features of the three-dimensional position coordinates, calculating the elevation fluctuation variation coefficient, dynamically updating the spatial compensation weights by combining the inverse square law of distance attenuation, and compensating and mapping the first radiation dose rate data to reconstruct the equivalent source strength data of the ground surface; if the reconstructed equivalent source strength data of the ground surface exceeds the dynamic warning threshold constructed by fusing historical background and preset standard deviation, then marking the corresponding coordinates as abnormal nodes and generating a target area retest command; S3. responding to the retest command, controlling the UAV... The drone descends to the second preset altitude and hovers at the abnormal node. It switches the detector to the high-range channel to acquire the second radiation dose rate data sequence. It extracts the timestamps representing radiation extrema from the sequence and captures associated image frames. It performs nuclear feature correlation analysis to anchor the radiation extrema spatially to the target abnormal pixel region within the associated image frame. It then packages the anchoring result, associated image frame, and current 3D position coordinates into a retest data packet. S4. It acquires network delay characteristic parameters. If the delay exceeds the preset threshold, it writes the retest data packet to the onboard asynchronous cache. It triggers the upload when the network delay characteristic parameters fall back to the preset delay threshold. If the delay threshold is met, it is synchronously distributed to the ground control terminal and the field terminal. In response to the rectification confirmation signal from the field terminal, it controls the drone to return to the interrupted position and resume gridded flight.

[0022] In this implementation plan, step S1 mainly achieves the preliminary screening and basic data acquisition functions for high-altitude, large-scale radiation signals. The terrain-following mode refers to the UAV's ability to use its onboard positioning module to sense ground undulations and automatically adjust its flight attitude to maintain a constant relative altitude. The first preset altitude refers to a relatively high cruising reference altitude set for large-area, rapid, and safe scanning. The dual-range GM detector refers to a Geiger-Müller radiation sensor that integrates two independent measurement channels: one with high sensitivity and the other with a wide range. The technical role of this step is to utilize the UAV's high maneuverability combined with the detector's high signal-capturing capability for weak signals in its low-range channel, while ensuring the safety of the aircraft and preventing collisions with complex terrain such as mines, to quickly and continuously acquire weak radiation signals with absolute time and spatial three-dimensional coordinates, providing spatiotemporally aligned raw data support for subsequent algorithm analysis.

[0023] Step S2 primarily achieves accurate compensation and restoration of spatial radiation attenuation data and dynamic early warning of target areas exceeding the standard. The elevation fluctuation variation coefficient refers to a characteristic parameter that quantitatively assesses the deviation of the actual instantaneous altitude from the ideal cruising altitude when the UAV is flying over undulating terrain; the distance attenuation inverse square law refers to the inherent physical law that the radiation dose rate decreases rapidly inversely with the square of the distance from the radiation source; the surface equivalent source strength data refers to the ground-based radiation source intensity value derived by the system after removing the high-altitude air attenuation factor through algorithms; the dynamic early warning threshold refers to the adaptive alarm limit calculated in real time by fusing the historical background radiation value of a specific area with environmental dispersion, abandoning the traditional fixed and rigid alarm values. The technical role of this step is to completely eliminate the radiation attenuation error caused by UAV high-altitude operations and drastic terrain undulations at the algorithm level, accurately restoring the distorted weak measurement values ​​in the air to the true pollution values ​​on the ground, and, combined with flexible alarm thresholds, significantly reducing the radiation false alarm and missed alarm rates in complex terrain environments.

[0024] Step S3 primarily achieves low-altitude precise fixed-point re-measurement and multi-source data visual locking of physical pollution sources. The second preset altitude refers to the extremely low hovering altitude at which the UAV, after responding to commands, descends close to the ground for precise detection and photography. Visual nuclear feature correlation analysis refers to the algorithmic process of aligning and mapping the extreme values ​​of invisible nuclear radiation data sensed by the detector with the visible high-definition images captured by the camera component, based on the same timestamp. Target abnormal pixel regions refer to the specific image block boundaries in the captured image frames that the system analyzes and determines directly correspond to real physical pollution sources. The technical significance of this step lies in solving the industry pain point that existing conventional technologies can only provide abstract coordinates of excessive levels but cannot accurately pinpoint which specific rock or puddle is exceeding the standard. Through low-altitude, close-range, high-range, overload-resistant detection and the fusion and matching of multi-source heterogeneous data, the radiation peak is directly anchored to the physical target, providing intuitive and conclusive visual evidence for subsequent manual ground cleanup and hazard mitigation.

[0025] Step S4 primarily implements a closed-loop function that integrates data loss prevention and security strategies with on-site rectification in weak network environments. Network latency characteristic parameters are evaluation indicators used to characterize the signal quality and congestion status of the current airborne communication module and external network link in real time. Asynchronous caching refers to a loss prevention mechanism that, in the event of communication interruption or extreme congestion, does not block the current control thread of the UAV but temporarily stores the early warning and evidence collection data in the airborne local memory and automatically resumes transmission after the network is restored. The rectification confirmation signal refers to the task completion feedback instruction sent by on-site personnel to the UAV main control system via the terminal after completing the physical treatment of the pollution source based on the received retest data packet. The technical role of this step is to fully ensure the absolute safety and non-loss of important evidence collection data when signal disconnection occurs due to complex terrain such as mining areas. It also breaks down the barrier of traditional UAVs only reporting unidirectionally to the main control station, directly distributing excessive data to the on-site terminal to guide personnel in hazard mitigation, and automatically instructing the UAV to resume the interrupted grid scanning task after safety is confirmed. This forms a closed-loop business process of unmanned and human collaboration from aerial hazard discovery to ground hazard elimination.

[0026] Specifically, the process of controlling the UAV to enter terrain-following mode to perform gridded flight at a first preset altitude and to collect three-dimensional position coordinates and timestamped first radiation dose rate data through the low-range channel of the dual-range GM detector is as follows: Three-dimensional terrain reference profile data of the target monitoring area is acquired, and a terrain-following guidance trajectory is generated in conjunction with the real-time centimeter-level positioning signal output by the airborne RTK; the airborne flight control unit is controlled to drive the UAV along the terrain-following guidance trajectory and maintain the first preset altitude to perform gridded cruise; the dual-range GM detector is configured to lock the low-range detection channel and initialize a fixed sampling period; within each fixed sampling period, the absolute timestamp of the airborne time reference is extracted, and the absolute timestamp, the first radiation dose rate data synchronously output by the low-range detection channel, and the three-dimensional position coordinates synchronously output by the airborne RTK are spatiotemporally bound and encapsulated into a low-level raw data packet.

[0027] In this implementation plan, the three-dimensional terrain reference profile data refers to digital elevation model data containing information on the undulations and slope elevation of the target monitoring area, acquired in advance through surveying and mapping methods. Airborne RTK, or Real-Time Dynamic Carrier Phase Differential Kinematics, is a satellite navigation and positioning module capable of providing centimeter-level high-precision three-dimensional positioning results. By performing real-time calculation and joint analysis of the preset static terrain profile data and the dynamic high-precision three-dimensional coordinates output by the airborne RTK, the system can calculate a three-dimensional spatial curve that perfectly matches the surface undulations—the terrain-following guidance trajectory. The technical advantage of this process is that it endows the UAV with the ability to anticipate terrain undulation trends, thereby completely avoiding the risk of UAV collisions and crashes due to sudden terrain changes when flying in non-flat monitoring areas with complex mountain slopes or deep depressions, effectively ensuring the physical safety of the flight platform. Secondly, it performs automated cruise operations at a constant relative altitude. In this step, the airborne flight control unit refers to the core control center inside the UAV responsible for controlling rotor speed, flight attitude, and heading changes; the first preset altitude refers to the constant vertical distance between the UAV and the ground surface set under the premise of balancing radiation detection range and flight collision avoidance safety; grid-based cruise refers to the standardized operation mode of controlling the UAV to conduct a full-coverage scan of the target area without blind spots according to the preset crisscrossing routes. The technical function of this step is that, based on the terrain-following guidance trajectory generated above, the flight control unit drives the UAV to maintain a constant relative distance from the ground surface while scanning over complex terrain. This avoids drastic fluctuations in flight altitude and provides a stable and consistent spatial physical reference for overcoming the air attenuation error of radiation due to distance changes. Next, the high-sensitivity operating mode of the radiation detection sensor is initialized. In this step, the dual-range GM detector is a Geiger-Miller counting device that integrates two independent hardware sensing channels: a high-sensitivity low-range detector and an overload-resistant high-range detector; the fixed sampling period refers to the extremely short fixed time interval between two consecutive output measurement results from the detector, which is forcibly set by the system. The technical purpose of this step is to address the physical phenomenon that the long-distance air obstruction during the initial preset altitude cruise of the UAV causes significant attenuation of radiation. It forcibly activates the detector's low-range channel to obtain the maximum possible capture capability for weak radiation signals, and uses a fixed sampling frequency to ensure the uniformity and continuity of the collected radiation signals along the time axis, preventing localized missed reports or data gaps caused by the UAV's rapid flight. Finally, it performs low-level precise spatiotemporal fusion and encapsulation of multi-source heterogeneous data. In this step, the absolute timestamp of the airborne time reference refers to the unified and unique time scale identifier provided by the airborne high-precision hardware clock source; spatiotemporal binding refers to the forced association of feature data acquired by different independent sensors at the same physical instant at the low-level data structure; and the low-level raw data packet refers to the initial data set containing time, space, and radiation intensity without any upper-level algorithm compensation or filtering smoothing processing.The technical role of this step is to introduce a unified absolute time reference to accurately match and fuse the instantaneous nuclear radiation intensity data captured by the detector with the high-precision latitude, longitude and elevation coordinates synchronously output by the airborne RTK at the millisecond level. This completely eliminates the problem of misalignment between the measured value and the actual spatial position caused by the data acquisition delay in the high-speed movement of the UAV. It provides a reliable underlying data source with ultimate spatiotemporal synchronization characteristics for subsequent attenuation compensation equivalent deduction and precise spatial positioning of the target area beyond the standard.

[0028] Specifically, the process of extracting elevation features of three-dimensional position coordinates to calculate the elevation fluctuation variation coefficient, dynamically updating spatial compensation weights in conjunction with the inverse square law of distance attenuation, and compensating and mapping the first radiation dose rate data to reconstruct the equivalent source strength data of the Earth's surface is as follows: The elevation features of three-dimensional position coordinates are extracted from multiple consecutive bottom-level raw data packets within a preset time window to construct a dynamic elevation floating sequence. The deviation of the dynamic elevation floating sequence from the first preset altitude is extracted, and the preset steady-state parameters of the UAV's flight attitude are fused to determine the elevation fluctuation variation coefficient. Based on the instantaneous ground-free altitude contained in the dynamic elevation floating sequence, a preset inverse square law of distance attenuation is mapped to extract the corresponding benchmark attenuation factor. The nonlinear feature correction of the benchmark attenuation factor is performed using the elevation fluctuation variation coefficient, and the spatial compensation weights are output. The spatial compensation weights are then fused with the first radiation dose rate data at the data level, and the mapping is reversed to reconstruct the equivalent source strength data of the Earth's surface.

[0029] In this implementation scheme, firstly, the system extracts the height features of the three-dimensional position coordinates from multiple consecutive underlying raw data packets within a preset time window to construct a dynamic elevation fluctuation sequence. Because in actual operation, even with the onboard positioning module, the actual flight altitude of the UAV fluctuates slightly around a first preset altitude due to interference from complex terrain and the turbulence of the rotor's underlying airflow. Therefore, the system extracts the deviation of the dynamic elevation fluctuation sequence from the first preset altitude, integrates the UAV's preset steady-state flight attitude parameters, and determines the elevation fluctuation variation coefficient using the following formula: ;in, : Coefficient of variation of elevation fluctuation; The total amount of underlying raw data packets included within the preset time window; Time series index; The first in the dynamic elevation floating sequence The instantaneous altitude above ground contained in each underlying raw data packet; First preset height; The system uses preset steady-state parameters for the UAV's flight attitude to characterize the baseline attitude damping attenuation factor when the UAV is in an ideal, undisturbed level flight state. Through the above calculations, the system transforms a single instantaneous vertical position error into a continuous fluctuation state characteristic over a period of time, thereby scientifically quantifying the real-time physical stability of the UAV platform in complex spatial environments. Next, based on the instantaneous ground clearance contained in the dynamic elevation fluctuation sequence, a preset inverse square law rule library for distance attenuation is mapped. The attenuation of radiation dose rate in the air strictly follows the inverse square law of spatial distance; however, in dynamic cruise accompanied by continuous elevation fluctuations, using a traditional static single-point compensation constant will introduce a large error. Therefore, the system extracts the corresponding baseline attenuation factor using the following formula and performs nonlinear characteristic correction on the baseline attenuation factor using the elevation fluctuation variation coefficient, outputting the spatial compensation weight: ; ;in, : No. The baseline decay factor corresponding to each time point; The system's pre-calibrated surface benchmark detection height; : No. Spatial compensation weights output at each time step; : An empirical correction adjustment factor is used to constrain the compensatory divergence drift caused by drastic changes in the elevation of the UAV; : Natural constant. The above-mentioned nonlinear characteristic correction process can effectively suppress the abnormal amplification of compensation weights caused by the sudden and severe turbulence of the UAV, ensuring the mathematical stability of the compensation mechanism. Finally, the output spatial compensation weights are fused with the first radiation dose rate data at the data level, that is, the weight coefficients are used to compensate and map the weak data at the underlying logic, and the equivalent source strength data of the ground surface is reconstructed by reverse deduction. This step realizes the accurate restoration from the weak radiation readings at high altitude to the actual pollution source intensity on the ground surface, completely eliminating the measurement errors caused by the dual factors of flight altitude fluctuations and air distance attenuation, laying the core data foundation for subsequent high-confidence anomaly early warning.

[0030] Specifically, if the reconstructed surface equivalent source strength data exceeds the dynamic warning threshold constructed by fusing historical baseline and preset standard deviation, the process of marking the corresponding coordinates as anomaly nodes and generating target area retesting instructions is as follows: retrieve historical radiation baseline data of the target monitoring area under normal conditions, extract the dispersion characteristics of the historical radiation baseline data to generate a standard deviation base, introduce a preset multiplier parameter to scale the standard deviation base, perform fusion reconstruction with the scaled standard deviation base and historical radiation baseline data, and output the dynamic warning threshold; establish a data comparison logic determiner, input the surface equivalent source strength data into the data comparison logic determiner and perform exceedance comparison detection with the dynamic warning threshold; when the data comparison logic determiner outputs an exceedance determination signal, mark the three-dimensional position coordinates bound to the surface equivalent source strength data as anomaly nodes, write the coordinates of the anomaly node into a preset flight control instruction message format, and generate a target area retesting instruction containing a second preset altitude reduction instruction and a detector range switching trigger code.

[0031] In this implementation scheme, to avoid the regional underreporting or frequent false alarms that are easily generated by traditional fixed numerical alarm thresholds when facing differences in background radiation across different geological regions, a dynamic early warning and target area locking action triggering mechanism based on environmental adaptation is designed. First, the system retrieves historical radiation background baseline data of the target monitoring area under normal conditions. Due to objective geographical differences in natural radioactivity levels across the monitoring area, the system extracts the dispersion characteristics of the historical radiation background baseline data to generate a standard deviation base. Then, a preset multiplier parameter is introduced to scale the standard deviation base. The scaled standard deviation base is then fused and reconstructed with the historical radiation background baseline data, and the dynamic early warning threshold is output through the following adaptive evolution formula: ;in, Dynamic early warning threshold; : The statistical average of historical radiation background baseline data for the target monitoring area; : Preset multiplier parameter, used to linearly adjust the sensitivity tolerance boundary of the early warning response mechanism; The standard deviation of historical radiation background baseline data is used. This step endows the system with strong environmental adaptive detection capabilities, enabling the anomaly judgment criteria to be dynamically adjusted according to the natural background fluctuations in different operating areas, perfectly eliminating the interference of normal geological environmental fluctuations on the detection results. Subsequently, the system establishes a low-level data comparison logic determiner, inputting the reconstructed surface equivalent source strength data into the data comparison logic determiner, and performing strict exceedance comparison detection logic with the real-time generated dynamic warning threshold. When the determiner continuously identifies surface equivalent source strength data greater than the dynamic warning threshold within the internal clock cycle, it immediately outputs an exceedance judgment signal to the main control system. Once the exceedance judgment signal is intercepted, the system immediately extracts the three-dimensional position coordinates bound to the low-level surface equivalent source strength data and marks it as a physically meaningful anomaly node. Then, the system directly encapsulates the three-dimensional coordinates of the anomaly node into the data payload segment of the preset flight control command message format, and simultaneously compiles and generates a target area retest command in conjunction with the additional second preset altitude reduction command and the detector range switching trigger code. This step completely streamlines the decision-making process in the system control logic, from obtaining equivalent values ​​through large-area aerial macroscopic scanning to automatically changing the UAV flight mission queue after an anomaly is detected. This enables the UAV system to have the ability to autonomously descend and accurately verify suspected contaminated target areas immediately upon detection, greatly improving the targeting and focusing characteristics and emergency response speed of the UAV radiation monitoring workflow.

[0032] Specifically, in response to the retest command, the process of controlling the UAV to descend to the second preset altitude and hover at the abnormal node, and switching the detector to the high-range channel to obtain the second radiation dose rate data sequence is as follows: The retest command for the target area is analyzed to extract the coordinates of the abnormal node and the altitude reduction trigger code. The three-dimensional coordinate breakpoint features of the current high-altitude gridded flight interruption position of the UAV are recorded, and the airborne multi-directional visual obstacle avoidance sensor array is activated simultaneously. Combining the coordinates of the abnormal node with the spatial obstacle distribution features output by the airborne multi-directional visual obstacle avoidance sensor array, an obstacle avoidance vertical descent trajectory is generated, and the UAV is driven to descend to the second preset altitude along the obstacle avoidance vertical descent trajectory and lock the hovering attitude. A hardware channel switching control signal is sent to the dual-range GM detector to cut off the data stream of the low-range measurement channel and activate the high-range measurement channel, and the high-frequency continuous sampling period is initialized to extract the second radiation dose rate data sequence.

[0033] In this implementation plan, firstly, the target area retest command is parsed to accurately extract the three-dimensional spatial coordinates of the abnormal nodes, and the three-dimensional coordinate breakpoint features of the current high-altitude gridded flight interruption position are recorded in real time. This action is to provide an absolute regression benchmark for the UAV to accurately resume flight from the breakpoint after completing the low-altitude retest. Subsequently, the system synchronously activates the onboard multi-directional visual obstacle avoidance sensor array, which is a hardware perception module that can scan the surrounding space in real time through multiple sets of depth cameras and construct a three-dimensional spatial point cloud. Considering that actual working sites such as uranium mines are prone to low-altitude obstacles such as trees, towering rocks, or large construction machinery, the system does not adopt a rigid, mechanical vertical straight-line descent strategy. Instead, it combines the coordinates of the abnormal nodes with the distribution characteristics of spatial obstacles to construct a comprehensive obstacle avoidance descent cost function, thereby generating a safe and optimal obstacle avoidance vertical descent trajectory. The specific calculation logic of the comprehensive obstacle avoidance descent cost function is as follows: ;in, : Overall cost parameters for obstacle avoidance descent; Target gravity weighting coefficient; The three-dimensional Euclidean distance between the drone's current spatial position and the anomaly node; The total number of discrete obstacles detected by the airborne multi-directional visual obstacle avoidance sensor array within the current field of view; Spatial sequence labels of discrete obstacles; Obstacle repulsion force weighting coefficient; The distance from the edge of the drone's current outer shell to the [number]th [unit] The shortest spatial distance between discrete obstacles; This prevents division by zero and represents an extremely small constant indicating the physical safety buffer distance. By solving the aforementioned cost function in real time at the flight control layer to obtain the minimum spatial displacement vector, the system drives the UAV to safely land at the second preset altitude along the optimal obstacle avoidance vertical descent trajectory and lock into a stable hovering attitude. After hovering, the system immediately sends a hardware channel switching control signal to the dual-range GM detector, forcibly cutting off the low-range measurement channel, which is prone to signal saturation or even blockage and crashing in close-range strong radiation fields, while activating the high-range measurement channel with overload resistance. With the initial high-frequency continuous sampling period, the detector can extract a high-temporal-resolution second radiation dose rate data sequence in a very short time. This completely solves the technical pain point of detector paralysis when facing strong radiation sources at close range, ensuring the integrity and authenticity of close-range detection data.

[0034] Specifically, the process of extracting timestamps representing radiation extrema from the sequence and extracting associated image frames, performing kernel feature association analysis, spatially anchoring radiation extrema to the target anomalous pixel region within the associated image frame, and packaging the anchoring result, associated image frame, and current 3D position coordinates into a retest data package is as follows: Parse the second radiation dose rate data sequence, extract radiation extrema with local maximum characteristics, retrieve the absolute timestamp bound to the radiation extrema in the hardware clock register, input the absolute timestamp as the retrieval key into the video stream buffer queue of the airborne camera component, extract video frames with temporal consistency characteristics to construct associated image frames; perform terrain anomaly visual recognition analysis on the associated image frames, output the target anomalous pixel region, perform kernel feature association analysis, perform 2D spatial mapping and superposition of the radiation extrema and the target anomalous pixel region to generate spatial anchoring results; perform multi-source data fusion encapsulation on the spatial anchoring results, associated image frames, radiation extrema, and the current 3D position coordinates output by the airborne RTK at the second preset altitude to generate a retest data package.

[0035] In this implementation scheme, firstly, the system performs deep analysis on the acquired second radiation dose rate data sequence and uses a peak-finding algorithm to extract the radiation extremum with local maximum characteristics. This radiation extremum directly represents the peak value of the strongest radioactive source emission captured by the UAV within the hovering area. The system then retrieves the absolute timestamp closely bound to this radiation extremum from the hardware clock register and uses it as the unique retrieval key value to input into the video stream buffer queue of the airborne camera component, accurately extracting video frames at the same microsecond-level instant in the time dimension, thereby constructing associated image frames with strict temporal consistency characteristics. Next, the system performs a terrain anomaly visual recognition analysis on the associated image frames, using a preset edge detection and color segmentation model to extract areas with terrain features such as exposed ore, wastewater accumulation, or abnormal soil dumping, and outputs the target abnormal pixel areas. To accurately map the radiation peak invisible in three-dimensional space to the two-dimensional visible image plane, the system performs visual kernel feature association analysis, and calculates the spatial anchoring fusion index to perform two-dimensional spatial mapping and superposition of the radiation extremum and the target abnormal pixel areas. The specific calculation rules for the spatial anchoring fusion index are as follows: ;in, Spatial Anchoring Integration Index; : Preset nuclear radiation characteristic contribution weight constant; Radiation extrema extracted from the sequence; The upper limit of the physical range of the high-range measurement channel is the critical value. The two-dimensional pixel offset distance between the physical projection center point of the radiation extremum on the image plane and the geometric center point of the target anomalous pixel region; : Optical spatial mapping divergence scale coefficient of airborne camera components; : Preset visual feature contribution weight constant; The image texture gradient variation value of the target anomalous pixel region output by the visual recognition and analysis of terrain anomalies. The system traverses all potential visual anomaly regions within the image and selects the region with the largest spatial anchoring fusion index as the final spatial anchoring result. This calculation process completely replaces the tedious operation of subjective comparison by human eyes in traditional methods, forcibly binding the radiation intensity peak representing danger with the physical terrain entity causing the anomaly at the pixel level from the underlying data dimension. Finally, the system performs multi-source data fusion and encapsulation of the calculated spatial anchoring result, associated image frames with on-site real-time conditions, radiation extreme values ​​representing the degree of pollution, and high-precision current three-dimensional position coordinates output by the airborne RTK at the second preset altitude. This operation not only generates a retest data package, but also provides intuitive, conclusive, and highly confident visual and data-driven dual evidence support for subsequent guidance of ground personnel in accurately locating and mitigating hazards.

[0036] Specifically, the process of acquiring network latency characteristic parameters and writing retest data packets into the onboard asynchronous cache if they exceed a preset latency threshold, and triggering upload when the network latency characteristic parameters fall back to within the preset latency threshold, is as follows: The onboard communication unit continuously sends heartbeat detection messages to the ground control terminal. The round-trip response time of the heartbeat detection messages is extracted and reconstructed into network latency characteristic parameters. A state machine logic judgment rule is introduced, and the network latency characteristic parameters are input into the state machine logic judgment rule for interval comparison. When the state machine logic judgment rule outputs a network outage signal exceeding the preset latency threshold, the local file system write permission of the onboard storage unit is activated, and the retest data packets are appended to the tail of the preset asynchronous cache queue. A background continuous monitoring mechanism for network latency characteristic parameters is maintained. When the network latency characteristic parameters are detected to be within the preset latency threshold range within a continuous preset time window, historical retest data packets are sequentially extracted from the head of the asynchronous cache queue to trigger upload.

[0037] In this implementation plan, to address the issue of easily interrupted communication links in complex terrain environments such as mining areas due to mountain obstructions or base station signal blind spots, a network adaptive anti-interference and asynchronous data caching mechanism based on state machine logic is designed. First, the airborne communication unit continuously sends heartbeat detection messages of minimal data volume to the ground control terminal at a fixed high-frequency interval. The system not only extracts the round-trip response time of a single heartbeat detection message, but also introduces a smoothing filter mechanism to reconstruct network delay characteristic parameters in order to filter out instantaneous jitter spikes caused by multipath effects or occasional physical interference in the wireless channel. The specific reconstruction calculation formula is as follows: ;in, Network latency characteristic parameters; Instantaneous weight adjustment factor, used to balance the proportion of the latest network state to the historical network state; The instantaneous round-trip response time of the latest heartbeat detection message; The capacity of the historical heartbeat sampling buffer pool; : Data sequence number within the historical heartbeat sampling buffer; The round-trip response time of the s-th heartbeat detection message in the historical heartbeat sampling buffer pool. Through the above calculation, the system obtains a network latency characteristic parameter that can both keenly reflect the network disconnection trend and possess strong anti-jitter robustness. Next, the system inputs the network latency characteristic parameter into the pre-configured state machine logic judgment rules to perform continuous interval comparison. When the judgment result exceeds the preset latency threshold, the state machine immediately outputs a network disconnection status signal. At this time, in order to prevent the retest data packets containing critical over-standard photos and radiation extreme values ​​from being lost due to overflow in the transmission buffer of the communication module, the system urgently activates the local file system write permission of the onboard storage unit, appending the retest data packets as independent data blocks to the tail of the preset asynchronous cache queue for secure and fixed storage. During the network disconnection, the system maintains a continuous background monitoring mechanism. Once it is detected that the network latency characteristic parameter has stably fallen back to the preset latency threshold range within a continuous preset time window, the state machine switches back to the network access state, and the system then retrieves historical retest data packets from the head of the asynchronous cache queue in a first-in-first-out order to trigger an automatic upload action. This process completely severs the strong coupling between the drone's data acquisition and transmission ends, ensuring the absolute integrity and non-loss of the radiation alarm evidence chain even in extremely harsh weak network environments.

[0038] Specifically, if the preset delay threshold is met, the data is simultaneously distributed to the ground control terminal and the field terminal. The specific process of controlling the UAV to return to the interrupted position and resume gridded flight in response to the rectification confirmation signal from the field terminal is as follows: When the state machine logic judgment rule outputs a normal state signal that meets the preset delay threshold, a dual-link communication channel is established, and the retest data packet is synchronously pushed to the ground control terminal database and the construction site terminal interaction interface through the dual-link communication channel; the UAV is kept in a hovering monitoring state at the second preset altitude until the airborne communication unit parses the rectification confirmation signal triggered and reported by the construction site terminal based on physical manual interaction operation; the three-dimensional coordinate features of the gridded flight interruption position are extracted and the return recovery trajectory is generated, and the UAV is driven to climb along the return recovery trajectory to the spatial position where the interruption position is located, and the UAV is controlled to return to the interruption position to resume gridded flight.

[0039] This implementation plan addresses the pain point of traditional radiation monitoring operations where early warning information can only be displayed in the main control room, making it difficult for on-site personnel to be aware of dangers and take immediate action to isolate and mitigate risks. A multi-terminal linkage and closed-loop task coordination mechanism is designed. When the state machine logic judgment rule outputs a normal state signal that meets a preset delay threshold, the system immediately establishes a dual-link communication channel covering both public mobile communication networks and dedicated data transmission frequency bands. Through this dual-link communication channel, the UAV streams and slices the retest data packets, pushing them in parallel and synchronously to the ground control database for overall command by the personnel in charge. Simultaneously, it pushes the packets to the interactive interface of the construction site terminal, forcibly alerting the construction personnel. During this period, the UAV maintains a hovering monitoring state at a second preset altitude, continuously using its camera components to target the pollution source area until the onboard communication unit receives and parses the rectification confirmation signal triggered and reported by the construction site terminal based on physical manual interaction. This indicates that the on-site personnel have completed the physical investigation and isolation of radiation hazards based on the precise visual coordinates in the retest data packets. After confirming that the danger has been eliminated, the system extracts the three-dimensional coordinate features of the gridded flight interruption location, which is pre-saved in the onboard memory. To enable the drone to quickly return to its original global survey task with minimal time and energy consumption, the system constructs an optimal flight functional model to generate a return-to-home recovery trajectory. The comprehensive energy consumption cost of the specific return-to-home recovery trajectory is calculated using the following formula: ;in, The overall energy cost of returning to base and restoring the trajectory; : Estimated return time; Continuous time variables during the return journey; Flight speed energy consumption conversion coefficient; : Instantaneous flight speed at time t on the return-to-home recovery trajectory; : Flight acceleration energy consumption conversion coefficient; : Instantaneous flight acceleration at time t on the return trajectory; The environmental drag coefficient for overcoming work done by gravity; The relative climb height difference between the UAV's current position and the interrupted position at time t. The system solves the velocity and acceleration boundary conditions in real time in the underlying trajectory planner to minimize the overall energy consumption cost, thereby generating a smooth return-to-home recovery trajectory with optimal energy efficiency. Subsequently, the system drives the UAV to efficiently climb along this return-to-home recovery trajectory to the spatial position of the interrupted position, resets the status flag of the control center, and controls the UAV to seamlessly resume the previously incomplete gridded flight. This design not only realizes deep information binding between the UAV monitoring system and on-site construction personnel, but also ensures at the underlying control level that the equipment can resume the continuity of automated patrol in the optimal attitude after the radiation anomaly event is handled.

[0040] Example 2; please refer to Figure 2 A system for rapid monitoring of radiation risk from unmanned aerial vehicles (UAVs) is provided for implementing a method for rapid monitoring of radiation risk from UAVs as described in the embodiments. The system includes: a high-altitude initial screening module for controlling a UAV in terrain-following mode to perform gridded flight at a first preset altitude, acquiring three-dimensional position coordinates and a first radiation dose rate data with timestamps through the low-range channel of a dual-range GM detector; a dynamic compensation and early warning module for extracting elevation features of the three-dimensional position coordinates, calculating the elevation fluctuation variation coefficient, dynamically updating the spatial compensation weights based on the inverse square law of distance attenuation, and compensating and mapping the first radiation dose rate data to reconstruct the equivalent source strength data of the ground surface; if the reconstructed equivalent source strength data of the ground surface exceeds the dynamic early warning threshold constructed by fusing historical background and a preset standard deviation, the corresponding coordinates are marked as abnormal nodes and a target area retest command is generated; and a visual nucleus fusion retest module. The system is used to respond to retest commands, control the UAV to descend to a second preset altitude and hover at the abnormal node, switch the detector to the high-range channel to acquire the second radiation dose rate data sequence; extract the timestamps representing radiation extrema in the sequence and capture associated image frames, perform visual kernel feature correlation analysis, anchor the radiation extrema space to the target abnormal pixel region within the associated image frame, and package the anchoring result, associated image frame, and current three-dimensional position coordinates into a retest data packet; the task linkage closed-loop module is used to acquire network latency characteristic parameters. If the latency exceeds the preset threshold, the retest data packet is written to the onboard storage asynchronous cache. The upload is triggered when the network latency characteristic parameters fall back to within the preset latency threshold; if the preset latency threshold is met, it is synchronously distributed to the ground control terminal and the field terminal; in response to the rectification confirmation signal from the field terminal, the UAV is controlled to return to the interrupted position and resume gridded flight.

[0041] In this implementation plan, the high-altitude initial screening and scanning module serves as the hardware collaborative control and preliminary sensing front-end of the entire system, primarily undertaking the task of large-scale, high-efficiency acquisition of basic ground-level data. This module coordinates the UAV flight control unit and high-precision positioning components to ensure the flight platform can safely and smoothly conduct adaptive undulating cruise over complex terrain, and forcibly activates the highest sensitivity measurement channel of the radiation detector. Its core function is to provide the entire system with an initial radiation state slice with extremely high spatiotemporal consistency, continuously supplying high-quality, low-level raw data bound to weak radiation intensities for subsequent attenuation compensation and algorithm reconstruction, while ensuring the UAV does not risk collisions. The dynamic compensation and early warning module is the core algorithm analysis hub of the system, primarily responsible for the soft computation of eliminating physical environmental interference and accurate calibration and early warning. After receiving the initial data collected by the front-end, this module specifically performs nonlinear compensation mapping for the air distance attenuation caused by high-altitude detection and the flight altitude fluctuation errors caused by terrain undulations, accurately restoring the weak aerial signals distorted by distance to the true ground pollution source strength. Simultaneously, it performs rigorous exceedance comparisons based on the early warning boundary adaptively evolved from historical background data. Its core function lies in completely eliminating the interference of complex terrain and topography on detection accuracy at the algorithm level, achieving high-confidence initial screening and locking of potential excessive target areas, and effectively avoiding false alarms and missed detections caused by altitude changes. The visual-nuclear fusion retest module, as the system's microscopic close-range detection and multi-source heterogeneous data consolidation execution mechanism, is mainly responsible for visually and accurately identifying the source of abnormal target areas locked in the initial screening. After receiving instructions, this module directly takes over the UAV's altitude reduction and obstacle avoidance flight control and the detector's hardware range switching, effectively avoiding data saturation paralysis that may occur when approaching strong radiation sources at close range. By forcibly spatially mapping and fusing the detected invisible radiation extreme values ​​with the visible terrain images output by the airborne camera on the underlying absolute time axis, its core function is to directly concretize the abstract excessive values ​​into high-definition real-time evidence with physical pollution source boundaries, providing intuitive and conclusive evidence for subsequent manual risk mitigation. The task linkage closed-loop module is the security guarantee of the system's communication link and the overall scheduling center of the entire business process, mainly responsible for breaking down the information silos between aerial monitoring data and ground physical disposal. This module not only endows the system with adaptive asynchronous caching and loss-prevention resume capabilities when encountering weak network environments under harsh geological conditions, but also directly establishes dual links to target and push retest early warning evidence to front-line operation terminals. Its core function is to upgrade simple aerial radiation monitoring into an automated business closed loop that includes anomaly detection, multi-terminal distribution, guidance for on-site investigation, terminal feedback to clear alarms, and control of drones to automatically resume their original patrol routes, greatly improving the efficiency of emergency response to abnormal events and the safety of human-machine collaborative risk mitigation.

[0042] In summary, this application has at least the following effects: A method and system for rapid monitoring of radiation risks using unmanned aerial vehicles (UAVs) effectively overcomes detection distortion and missed detection caused by air attenuation and altitude fluctuations in complex terrain by controlling the UAV to perform gridded high-altitude initial screening in terrain-following mode and dynamically reconstructing the equivalent source strength data of the ground surface by combining distance attenuation rules and altitude fluctuation characteristics. After triggering the target area retest command, the system coordinates the UAV to descend and hover and adaptively switch to a high-range detection channel. By using visual kernel feature correlation analysis, the abstract nuclear radiation extreme value is accurately spatially anchored to the visible target abnormal pixel area, realizing accurate visual locking of physical pollution sources and solidification of multi-source heterogeneous data. At the same time, relying on the network adaptive asynchronous caching and multi-terminal targeted communication distribution mechanism based on state machine logic, it not only effectively ensures the integrity and data loss prevention of early warning and evidence collection in complex weak network environments, but also thoroughly opens up the closed-loop management of the entire automated business process from high-altitude wide-area initial screening, anomaly close-range source identification, on-site linkage and risk mitigation to automatic recovery of UAV cruise. This significantly improves the reliability of underlying data, the accuracy of multi-source fusion detection, and the efficiency of human-machine collaborative emergency response in nuclear radiation environmental monitoring operations.

[0043] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for rapid monitoring of radiation risk from unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: S1. Control the UAV that has entered the terrain following mode to perform grid-like flight at the first preset altitude, and collect three-dimensional position coordinates and the first radiation dose rate data with timestamp through the low-range channel of the dual-range GM detector. S2. Extract the elevation features of the three-dimensional position coordinates, calculate the elevation fluctuation variation coefficient, and dynamically update the spatial compensation weights in combination with the inverse square law of distance attenuation. Compensate and map the first radiation dose rate data to reconstruct the equivalent source strength data of the Earth's surface. If the reconstructed surface equivalent source strength data exceeds the dynamic early warning threshold constructed by fusing historical background and preset standard deviation, the corresponding coordinates will be marked as an abnormal node and a target area retest instruction will be generated. S3. Responding to the retest command, control the UAV to descend to the second preset altitude and hover at the abnormal node, switch the detector to the high-range channel to obtain the second radiation dose rate data sequence; extract the timestamps representing radiation extreme values ​​in the sequence and extract the associated image frames, perform visual kernel feature association analysis, anchor the radiation extreme value space to the target abnormal pixel region within the associated image frame, and package the anchoring result, associated image frame and current three-dimensional position coordinates into a retest data package; S4. Obtain network latency characteristic parameters. If the latency exceeds the preset threshold, write the retest data packet to the onboard storage asynchronous cache. Once the network latency characteristic parameters fall back to within the preset latency threshold, trigger the upload. If the preset latency threshold is met, distribute the data synchronously to the ground control terminal and the field terminal. Respond to the rectification confirmation signal from the field terminal and control the UAV to return to the interrupted position and resume gridded flight.

2. The method for rapid monitoring of radiation risk from unmanned aerial vehicles according to claim 1, characterized in that: The specific process of controlling the UAV to enter terrain-following mode to perform gridded flight at a first preset altitude, and to collect three-dimensional position coordinates and the first radiation dose rate data with timestamps through the low-range channel of the dual-range GM detector is as follows: Acquire three-dimensional terrain benchmark profile data of the target monitoring area, and generate a terrain-following guidance trajectory by combining the real-time centimeter-level positioning signal output by the airborne RTK. The airborne flight control unit drives the UAV to follow the terrain and maintain the first preset altitude to perform grid-based cruise. Configure a dual-range GM detector to lock the low-range detection channel and initialize a fixed sampling period. Within each fixed sampling period, extract the absolute timestamp of the airborne time reference. Spatiotemporally bind the absolute timestamp, the first radiation dose rate data synchronously output by the low-range detection channel, and the three-dimensional position coordinates synchronously output by the airborne RTK, and encapsulate them into a low-level raw data packet.

3. The method for rapid monitoring of radiation risk from unmanned aerial vehicles according to claim 2, characterized in that: The specific process of extracting elevation features from three-dimensional location coordinates, calculating the elevation fluctuation variation coefficient, dynamically updating spatial compensation weights based on the inverse square law of distance attenuation, and compensating for mapping the first radiation dose rate data to reconstruct the equivalent source intensity data of the Earth's surface is as follows: The elevation features of three-dimensional position coordinates in multiple consecutive underlying raw data packets within a preset time window are extracted to construct a dynamic elevation floating sequence. The deviation of the dynamic elevation floating sequence from the first preset altitude is extracted, and the preset flight attitude steady-state parameters of the UAV are fused to determine the elevation fluctuation variation coefficient. Based on the instantaneous ground height contained in the dynamic elevation floating sequence, a preset distance attenuation squared ratio rule library is mapped to extract the corresponding benchmark attenuation factor. The nonlinear feature correction of the benchmark attenuation factor is performed using the elevation fluctuation variation coefficient, and the spatial compensation weight is output. The spatial compensation weights are fused with the first radiation dose rate data at the data level, and the mapping is reversed to reconstruct the surface equivalent source strength data.

4. The method for rapid monitoring of radiation risk from unmanned aerial vehicles according to claim 3, characterized in that: If the reconstructed equivalent source strength data of the land surface exceeds the dynamic early warning threshold constructed by fusing historical background and preset standard deviation, the specific process of marking the corresponding coordinates as anomaly nodes and generating target area retest instructions is as follows: Historical background radiation baseline data of the target monitoring area under normal conditions is retrieved, the dispersion characteristics of the historical background radiation baseline data are extracted to generate a standard deviation base, a preset multiplier parameter is introduced to scale the standard deviation base, and the scaled standard deviation base is fused and reconstructed with the historical background radiation baseline data to output a dynamic early warning threshold. Establish a data comparison logic determiner, input the surface equivalent source strength data into the data comparison logic determiner and perform exceedance comparison detection with the dynamic early warning threshold; When the data comparison logic determiner outputs an over-determination signal, the three-dimensional position coordinates bound to the ground equivalent source strength data are marked as an abnormal node. The coordinates of the abnormal node are written into the preset flight control command message format, and a target area retest command containing a second preset altitude reduction command and a detector range switching trigger code is generated.

5. The method for rapid monitoring of radiation risk from unmanned aerial vehicles according to claim 1, characterized in that: In response to the retest command, the drone is controlled to descend to the second preset altitude and hover at the abnormal node. The specific process of switching the detector to the high-range channel to obtain the second radiation dose rate data sequence is as follows: The target area retest command is analyzed to extract the coordinates of abnormal nodes and the action code for altitude reduction triggering. The three-dimensional coordinate breakpoint features of the current high-altitude gridded flight interruption position of the UAV are recorded, and the airborne multi-directional visual obstacle avoidance sensor array is activated simultaneously. By combining the coordinates of abnormal nodes with the spatial obstacle distribution characteristics output by the airborne multi-directional visual obstacle avoidance sensor array, an obstacle avoidance vertical descent trajectory is generated, driving the UAV to descend to the second preset height along the obstacle avoidance vertical descent trajectory and lock the hovering attitude. A hardware channel switching control signal is sent to the dual-range GM detector to cut off the data stream of the low-range measurement channel and activate the high-range measurement channel, and initialize the high-frequency continuous sampling period to extract the second radiation dose rate data sequence.

6. The method for rapid monitoring of radiation risk from unmanned aerial vehicles according to claim 5, characterized in that: The specific process of extracting timestamps representing radiation extrema from the sequence and extracting associated image frames, performing kernel feature association analysis, anchoring the radiation extrema space to the target abnormal pixel region within the associated image frame, and packaging the anchoring results, associated image frames, and current 3D position coordinates into a retest data package is as follows: The second radiation dose rate data sequence is analyzed, radiation extrema with local maximum characteristics are extracted, the absolute timestamp bound to the radiation extrema in the hardware clock register is retrieved, the absolute timestamp is used as the retrieval key value and input into the video stream buffer queue of the airborne camera component, and video frames with time consistency characteristics are extracted to construct associated image frames. Perform terrain anomaly visual recognition analysis on the associated image frames, output the target anomaly pixel region, perform visual kernel feature association analysis, and perform two-dimensional spatial mapping superposition on the radiation extrema and the target anomaly pixel region to generate spatial anchoring results; The spatial anchoring results, associated image frames, radiation extrema, and the current three-dimensional position coordinates output by the airborne RTK at the second preset altitude are fused and encapsulated to generate a retest data packet.

7. The method for rapid monitoring of radiation risk from unmanned aerial vehicles according to claim 1, characterized in that: The process of acquiring network latency characteristic parameters and writing retest data packets to onboard storage asynchronous cache if they exceed a preset latency threshold, and triggering upload once the network latency characteristic parameters fall back to within the preset latency threshold, is as follows: The airborne communication unit continuously sends heartbeat detection messages to the ground control terminal. The round-trip response time of the heartbeat detection messages is extracted and reconstructed into network delay characteristic parameters. State machine logic judgment rules are introduced, and the network delay characteristic parameters are input into the state machine logic judgment rules to perform interval comparison. When the state machine logic judgment rule outputs a network disconnection status signal that exceeds the preset delay threshold, the local file system write permission of the onboard storage unit is activated, and the retest data packet is appended to the tail of the preset asynchronous cache queue. A background continuous monitoring mechanism for network latency characteristic parameters is maintained. When the network latency characteristic parameters are detected to be within a preset latency threshold range within a continuous preset time window, historical retest data packets are sequentially extracted from the head of the asynchronous cache queue to trigger uploading.

8. The method for rapid monitoring of radiation risk from unmanned aerial vehicles according to claim 7, characterized in that: If the preset delay threshold is met, the data is simultaneously distributed to the ground control terminal and the field terminal; in response to the rectification confirmation signal from the field terminal, the specific process of controlling the UAV to return to the interrupted position and resume gridded flight is as follows: When the state machine logic judgment rule outputs a normal state signal that meets the preset delay threshold, a dual-link communication channel is established, and the retest data packet is synchronously pushed to the ground control terminal database and the construction site terminal interaction interface through the dual-link communication channel. Maintain the drone in a hovering monitoring state at the second preset altitude until the airborne communication unit resolves the rectification confirmation signal triggered and reported by the construction site terminal based on physical manual interaction. Extract the three-dimensional coordinate features of the stored gridded flight interruption location to generate a return-to-home recovery trajectory, drive the UAV to climb along the return-to-home recovery trajectory to the spatial location of the interruption location, and control the UAV to return to the interruption location to resume gridded flight.

9. A system for rapid monitoring of radiation risk from unmanned aerial vehicles (UAVs), used to execute the method for rapid monitoring of radiation risk from UAVs as described in any one of claims 1-8, characterized in that, include: The high-altitude initial screening and scanning module is used to control the UAV that enters the terrain following mode to perform gridded flight at the first preset altitude, and to collect three-dimensional position coordinates and the first radiation dose rate data with timestamps through the low-range channel of the dual-range GM detector. The dynamic compensation and early warning module is used to extract the elevation features of the three-dimensional position coordinates, calculate the elevation fluctuation variation coefficient, and dynamically update the spatial compensation weight in combination with the distance attenuation square inverse rule to compensate and map the first radiation dose rate data to reconstruct the surface equivalent source strength data. If the reconstructed surface equivalent source strength data exceeds the dynamic early warning threshold constructed by fusing historical background and preset standard deviation, the corresponding coordinates will be marked as an abnormal node and a target area retest instruction will be generated. The visual nucleus fusion retest module is used to respond to retest commands, control the UAV to descend to the second preset altitude and hover at the abnormal node, switch the detector to the high-range channel to obtain the second radiation dose rate data sequence; extract the timestamps representing radiation extreme values ​​in the sequence and extract the associated image frames, perform visual nucleus feature association analysis, anchor the radiation extreme value space to the target abnormal pixel region within the associated image frame, and package the anchoring result, associated image frame and current three-dimensional position coordinates into a retest data package; The task linkage closed-loop module is used to obtain network latency characteristic parameters. If the latency exceeds the preset threshold, the retest data packet is written to the onboard storage asynchronous cache. The upload is triggered when the network latency characteristic parameters fall back to within the preset latency threshold. If the preset latency threshold is met, it is synchronously distributed to the ground control terminal and the field terminal. In response to the rectification confirmation signal from the field terminal, the UAV is controlled to return to the interrupted position and resume gridded flight.

Images