UAV inspection method and system for highway environment improvement

Abstract

The present invention discloses a drone inspection method and system for highway environmental improvement, relating to the field of drone communication technology. The method comprises: allocating a drone swarm to each inspection unit; predicting the drone mission load of each inspection unit, formulating a drone scheduling mechanism based on the predicted drone mission load, and establishing a communication link between the current drone cluster and the scheduling drone according to the drone scheduling mechanism; sensing the communication link signal between the current drone cluster and the scheduling drone in real time, and using a path optimization algorithm to calculate the optimal relay transmission path when the communication link signal is interrupted to achieve communication link signal reconstruction. The present invention ensures communication stability between the drone cluster and the scheduling drone through a communication resource allocation mechanism based on spatiotemporal hash codes, thereby avoiding task delays caused by communication resource conflicts.

Description

Technical Field

[0001] The present invention relates to the field of drone communication technology, and in particular to a drone inspection method and system for highway environment improvement. Background Art

[0002] The highway environment refers to the natural and cultural environment along a highway, encompassing the combined influence of various factors, including land use, vegetation cover, climate conditions, transportation infrastructure, landscape, and soil and water conservation. It encompasses not only the condition of the road surface itself but also the surrounding ecological environment, socioeconomic activities, and their impact on traffic safety and road maintenance. Highway environmental remediation refers to the process of remediating, transforming, and managing the environment through a series of targeted measures to improve environmental quality, ensure traffic safety, and maintain ecological sustainability along highways.

[0003] Drone inspections utilize drones to monitor and inspect highway conditions. Leveraging their high-altitude perspective, flexibility, and equipped with high-definition cameras and sensors, drones collect real-time data on road conditions, traffic flow, and environmental changes along the highway for remote monitoring and analysis. However, unstable communication signals during drone inspections can lead to information synchronization delays between drones, hindering data sharing and collaborative operations. This is particularly true when conducting patrols in formation or when multiple drones collaborate on a mission, leading to path planning errors and duplicate inspections, reducing inspection efficiency.

[0004] Currently, no effective solutions have been proposed for the problems in related technologies. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides a drone inspection method and system for highway environment improvement, which has the advantage of stable communication between drone clusters and dispatching drones, thereby avoiding task delays caused by communication resource conflicts, and solving the problem that unstable communication signals will cause information synchronization delays between drones, thereby affecting data sharing and collaborative operations.

[0006] In order to achieve the advantages of stable communication between the above-mentioned drone cluster and the dispatching drone, and thus avoid mission delays caused by communication resource conflicts, the specific technical solutions adopted by the present invention are as follows:

[0007] According to one aspect of the present invention, a method for drone inspection for highway environment improvement is provided, the method comprising:

[0008] Based on pre-collected highway environment data, the highway network is divided into several inspection units, and a drone fleet is assigned to each inspection unit;

[0009] Predict the drone mission load of each patrol unit, formulate a drone scheduling mechanism based on the predicted results of the drone mission load, and establish a communication link between the current drone cluster and the scheduling drone according to the drone scheduling mechanism;

[0010] The communication link signal between the current drone cluster and the dispatching drone is perceived in real time. When the communication link signal is interrupted, the path optimization algorithm is used to calculate the optimal relay transmission path to achieve communication link signal reconstruction.

[0011] Furthermore, the UAV task load of each patrol unit is predicted, and a UAV scheduling mechanism is formulated based on the predicted results of the UAV task load. The communication link between the current UAV cluster and the scheduling UAV is established according to the UAV scheduling mechanism, including:

[0012] Based on time series analysis technology, the UAV mission load of each inspection unit in the future time period is predicted, and the load heat map of each inspection unit is generated;

[0013] Assign a globally unique spatiotemporal hash code to each drone in the inspection unit, and superimpose the drone parameters on the spatiotemporal hash code to form a composite coding tree hierarchical structure;

[0014] The drone task load is compared with the preset threshold. If the drone task load does not exceed the preset threshold, the patrol task is performed according to the initially assigned drone cluster. Otherwise, the drone is dispatched from the adjacent patrol unit according to the composite coding tree hierarchical structure, and the time slot communication resources between the current drone cluster and the dispatched drone are allocated according to the spatiotemporal hash code of the dispatched drone.

[0015] Furthermore, a globally unique spatiotemporal hash code is assigned to each drone in each patrol unit, and drone parameters are superimposed on the spatiotemporal hash code to form a composite coding tree hierarchical structure including:

[0016] The spatial-temporal alignment algorithm is used to fuse the geographical boundaries of the patrol unit with the time window to generate the spatial-temporal hash code of the drone in each patrol unit.

[0017] The UAV model code, network number, and device serial number are sequentially superimposed on the spatiotemporal hash code to form a composite coding tree, ensuring that the spatiotemporal hash code of each UAV is globally unique and that the spatiotemporal attribute information can be reversely parsed.

[0018] A composite coding tree hierarchical structure including a root node, level nodes, and leaf nodes is constructed based on the composite coding tree, and a priority order is established for the spatiotemporal hash code according to the hierarchical structure of the composite coding tree.

[0019] Furthermore, the calculation formula for integrating the geographical boundary of the inspection unit with the time window is:

[0020]

[0021] Where SpacetimeHash represents the spatiotemporal hash code generated by fusing the geographical boundary of the patrol unit with the time window; SM3(·) represents the anti-collision compression of input data using hash calculation; Hilbert(x,y) represents the Hilbert curve encoding generated by mapping the normalized latitude and longitude coordinates (x,y); t represents the timestamp; t0 represents the reference time; Δt represents the length of the dynamic time window; and TaskWeight represents the task weight function.

[0022] Furthermore, dispatching drones from adjacent patrol units according to the composite coding tree hierarchical structure and allocating time slot communication resources between the current drone cluster and the dispatching drone according to the spatiotemporal hash code of the dispatching drone include:

[0023] If the UAV task load exceeds the preset threshold, a dispatch instruction is sent to the UAV cluster of the adjacent patrol unit. After receiving the dispatch instruction, the UAV cluster of the adjacent patrol unit retrieves the UAV resources within the patrol unit;

[0024] If there is a lack of idle UAVs in the adjacent patrol unit, the time slot communication resources are searched from the adjacent patrol units according to the priority in the composite coding tree hierarchy structure, and the availability of the time slot communication resources is determined. If the time slot communication resources are available, a mobilization instruction is sent to the UAV corresponding to the available time slot communication resources. If the time slot is not available, the UAV enters the waiting queue.

[0025] Continuously monitor the quality indicators of the dispatched drones. If it detects that the dispatch instruction is interrupted or the quality indicators are not up to standard, it will exchange time slot communication resources with other drones occupying high-quality time slots through the time slot exchange protocol, and the drone after the time slot exchange will be used as the latest dispatched drone;

[0026] Among them, during the time slot exchange process, negotiation is carried out to exchange uplink, downlink or bidirectional time slots with other drones occupying high-quality time slots to meet the quality indicators of the scheduled drones; if the time slot exchange is successful, the original time slot will be released and returned to the resource pool. If the time slot exchange fails, the scheduling drone holding time will be extended and a retry will be triggered.

[0027] Furthermore, the communication link signal between the current UAV cluster and the dispatching UAV is sensed in real time, and when the communication link signal is interrupted, the path optimization algorithm is used to calculate the optimal relay transmission path to achieve communication link signal reconstruction, including:

[0028] Calculate the transmission distance of the communication link signal between the current drone cluster and the dispatching drone, and perceive the stability trend of the communication link signal based on the transmission distance;

[0029] If the communication link signal is sensed to be interrupted, the optimal relay transmission path between the current drone cluster and the scheduling drone is calculated, and based on the optimal relay transmission path, the relay node that restores the communication link signal is selected in the composite coding tree hierarchical structure to reconstruct the communication link.

[0030] Furthermore, the optimal relay transmission path between the current UAV cluster and the scheduling UAV is calculated, and according to the optimal relay transmission path, a relay node for recovering the communication link signal is selected in the composite coding tree hierarchical structure to reconstruct the communication link, including:

[0031] Construct a weighted directed graph, with the communication points in the weighted directed graph as drones and the edges as communication links. Initialize the distance of the source communication point to 0 and the distances of the remaining communication points to infinity.

[0032] Iteratively perform incremental relaxation calculations on the edges in a weighted directed graph, introduce real-time topology change data in each iteration, and use an incremental update strategy to eliminate the impact of real-time topology change data on edge calculations;

[0033] Determine whether there is a negative weight loop on the edge. If there is a negative weight loop, it means that the current communication link has an unstable signal and the transmission of the communication link signal cannot be guaranteed. If there is no negative weight loop, the relaxation calculation result of the edge is used as the optimal relay transmission path between the current drone cluster and the scheduling drone.

[0034] Select candidate communication points in ascending order of weight in the optimal relay transmission path, calculate the spatiotemporal hash codes of the candidate communication points and map them into a composite coding tree hierarchy, and allocate time slot communication resources according to the spatiotemporal hash codes of the corresponding relay nodes in the composite coding tree hierarchy;

[0035] Verify whether there is a breakpoint in the relay node of the composite coding tree hierarchy. If so, trace back the predecessor node chain, switch to the suboptimal relay transmission path, and reallocate time slot communication resources based on the spatiotemporal hash code of the candidate communication point on the suboptimal relay transmission path to complete the communication link reconstruction.

[0036] Furthermore, real-time topology change data is introduced in each iteration, and an incremental update strategy is used to eliminate the factors that affect edge calculations due to real-time topology change data. These factors include:

[0037] Introducing and processing real-time topology change data in each iteration, and the real-time topology change data includes the real-time strength of the communication link signal;

[0038] Perform local relaxation calculations on the affected edges based on real-time topology change data, and filter out outgoing edges with negative weights due to real-time changes;

[0039] A dynamic penalty factor is introduced in outgoing edges, and an additional round of full-graph relaxation iteration is triggered to eliminate the potential negative weight effects in real-time topology changing data.

[0040] Furthermore, the incremental relaxation calculation formula is:

[0041] d[v] (k+1) =min(d[v] (k) ,d[u] (k) +(w(u,v)+α·Δw(u,v))+γ·|| A );

[0042] Where d[v] (k+1) represents the estimated value of the shortest distance from communication point v to the source communication point in the k+1th iteration; w(u,v) represents the baseline weight of the edge formed by communication point u to communication point v; Δw(u,v) represents the real-time weight change of the edge formed by communication point u to communication point v; α represents the incremental adjustment factor; γ represents the mutation penalty term; A represents the indicator function; d[v] (k) represents the estimated value of the shortest distance from the communication point v to the source communication point in the kth iteration; d[u] (k) Represents the estimated value of the shortest distance from communication point u to the source communication point in the kth iteration.

[0043] According to another aspect of the present invention, there is also provided a drone inspection system for highway environment improvement, the system comprising:

[0044] The drone allocation module is used to divide the highway network into several inspection units based on pre-collected highway environment data and allocate a drone fleet to each inspection unit;

[0045] The communication resource allocation module is used to predict the UAV task load of each patrol unit, formulate a UAV scheduling mechanism based on the predicted results of the UAV task load, and establish a communication link between the current UAV cluster and the scheduling UAV according to the UAV scheduling mechanism;

[0046] The communication signal monitoring module is used to perceive the communication link signal between the current drone cluster and the scheduling drone in real time, and use the path optimization algorithm to calculate the optimal relay transmission path when the communication link signal is interrupted to achieve communication link signal reconstruction.

[0047] Compared with the existing technology, the present invention provides a drone inspection method and system for highway environment improvement, which has the following beneficial effects:

[0048] (1) The present invention predicts the drone task load of each patrol unit and formulates a drone scheduling mechanism based on the prediction results, thereby realizing intelligent scheduling of drones, ensuring the efficiency and flexibility of task execution, and establishing a communication link between the current drone cluster and the scheduling drone, sensing its signal strength in real time, ensuring the stability of drone data transmission during the patrol process, reducing the risk of communication interruption, and at the same time, when the communication link signal is interrupted, the path optimization algorithm is used to calculate the optimal relay transmission path, which can quickly rebuild the signal link to ensure that the patrol task is not interrupted, thereby improving the reliability of the patrol task and the emergency response capability.

[0049] (2) The present invention predicts the drone task load of each patrol unit in the future time period through time series analysis technology, can accurately grasp the task requirements of each patrol unit, and assign a globally unique spatiotemporal hash code to the drones in each patrol unit. On this basis, the drone parameters are superimposed to form a composite coding tree hierarchical structure, thereby ensuring the unique identification of each drone in the task, facilitating management and scheduling. At the same time, through the communication resource allocation mechanism of the spatiotemporal hash code, the communication stability between the drone cluster and the scheduling drone can be ensured, thereby avoiding task delays caused by communication resource conflicts.

[0050] (3) The present invention calculates the transmission distance of the communication link signal between the current drone cluster and the dispatching drone, and perceives the signal stability trend based on the transmission distance, thereby realizing real-time monitoring of the communication quality and improving the communication reliability of the inspection task. When the communication link signal is detected to be interrupted, the system can quickly calculate the optimal relay transmission path and select the appropriate relay node to reconstruct the communication link based on the composite coding tree hierarchical structure to ensure uninterrupted data transmission, thereby effectively improving the communication stability of drone inspections in highway environment improvement, reducing inspection interruptions caused by signal loss, and enhancing the coordination ability of drone tasks and the ability to adapt to complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 This is a flow chart of a drone inspection method for highway environment improvement according to an embodiment of the present invention;

[0053] Figure 2 The figure is a principle block diagram of a drone inspection system for highway environment improvement according to an embodiment of the present invention.

[0054] In the picture:

[0055] 1. Drone allocation module; 2. Communication resource allocation module; 3. Communication signal monitoring module. DETAILED DESCRIPTION

[0056] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention. They are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. By referring to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and the advantages of the present invention.

[0057] According to an embodiment of the present invention, a drone inspection method and system for highway environment improvement are provided.

[0058] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to an embodiment of the present invention, a drone inspection method for highway environment improvement includes:

[0059] S1. Based on the pre-collected highway environment data, the highway network is divided into several patrol units, and a drone fleet is assigned to each patrol unit.

[0060] It should be noted that based on the pre-collected highway environment data, the highway network is divided into several patrol units, and each patrol unit is assigned a drone fleet including:

[0061] Highway environmental data is collected through various methods, including satellite remote sensing, ground sensors, and drone inspections. This data includes road conditions, traffic flow, topography, climate conditions, and environmental pollution. This data is analyzed to identify key factors affecting highway maintenance and traffic safety, providing a basis for the subsequent division of inspection units.

[0062] Based on collected road environment data, GIS technology is used to divide the highway network into several inspection units. Each inspection unit can be divided based on factors such as the length, complexity, traffic volume, and environmental conditions of the road section. For example, a busy road section might be divided into small units to facilitate detailed inspections, while a road section with a simpler environment could be divided into large units.

[0063] Once the inspection units are divided, the next step is to allocate a suitable drone fleet to each unit based on the characteristics of each unit. The specific allocation strategy needs to consider factors such as the drone's endurance, payload capacity, and sensor configuration. More complex inspection units can be equipped with more drones, while simpler units can be allocated a smaller number of drones.

[0064] S2. Predict the drone mission load of each patrol unit, formulate a drone scheduling mechanism based on the predicted results of the drone mission load, and establish a communication link between the current drone cluster and the scheduling drone according to the drone scheduling mechanism.

[0065] Among them, predicting the drone task load of each inspection unit, formulating a drone scheduling mechanism based on the predicted results of the drone task load, and establishing a communication link between the current drone cluster and the scheduling drone according to the drone scheduling mechanism include:

[0066] Based on time series analysis technology, the UAV mission load of each patrol unit in the future time period is predicted, and the load heat map of each patrol unit is generated.

[0067] It should be noted that the UAV mission load of each patrol unit in the future time period is predicted based on time series analysis technology, and the load heat map of each patrol unit is generated, including:

[0068] Collect historical drone mission data of each patrol unit, including mission trigger time, execution duration, mission type, and geographic coordinates, and eliminate duplicate records and outliers (such as negative values ​​or out-of-range data) to ensure data integrity and consistency.

[0069] The ARIMA model (AutoRegressive Integrated Moving Average Model) is used to capture linear trends and seasonal fluctuations, and the LSTM neural network (Long Short-Term Memory Neural Network) is combined to learn nonlinear residual features. The prediction results of the two models are integrated through a weighted fusion strategy to generate the probability distribution of the task load of each inspection unit in a specified time period in the future, with the prediction error controlled within ±8%.

[0070] Among them, the ARIMA model is used to capture linear trends and seasonal fluctuations, combined with the LSTM neural network to learn nonlinear residual features, and the prediction results of the two models are integrated through a weighted fusion strategy.

[0071] The ARIMA model's differencing, autoregressive, and moving average orders are determined using autocorrelograms and partial autocorrelograms, and the optimal parameter combination is selected through grid search. After training is complete, linear forecast results for the future period are output, and the model residual sequence is extracted as the input feature of the LSTM.

[0072] A two-layer LSTM network structure is designed. The input layer receives the ARIMA residual sequence and external variables (such as weather and holidays). The number of neurons in the hidden layer is dynamically adjusted according to the loss of the validation set.

[0073] The linear prediction value output by ARIMA and the nonlinear residual of LSTM prediction are aligned by timestamp, and the initial fusion result is generated using the superposition formula: total prediction value = ARIMA prediction value + LSTM residual prediction value. Smoothing filtering is performed on the mutation points of the fused sequence to eliminate the inconsistent fluctuations between models.

[0074] The prediction error of each model is evaluated based on the validation set, and the inverse mean square error of ARIMA and LSTM is calculated as the initial weight. A sliding window mechanism is introduced to dynamically reduce the weight of models with higher recent prediction error rates. The final fusion formula is:

[0075] Final prediction value = ω1 × ARIMA prediction value + ω2 × LSTM total prediction value;

[0076] Where ω1 and ω2 represent the weights of ARIMA and LSTM, respectively, where ω1+ω2=1 and ω2≥0.6.

[0077] Based on the geographical boundaries of the inspection unit and the predicted load value, the Kriging interpolation algorithm is used to convert the discrete point data into a continuous spatial distribution. The interpolation results are corrected in combination with the road network topology (such as bridge and tunnel weights) to generate a load density surface covering the entire area.

[0078] It should be noted that the Kriging interpolation algorithm includes:

[0079] The Kriging interpolation algorithm is a prediction method based on spatial statistics. It constructs a variation function model by analyzing the spatial correlation (such as distance and direction) of known spatial data points, quantifies the law of data change with spatial position, and uses the weighted average idea to estimate the target value of the unknown position. Its weight not only considers the distance between the estimated point and the known point, but also reflects the spatial autocorrelation through the variation function (such as adjacent points are more similar). Ultimately, it minimizes the prediction error variance while ensuring the unbiased estimation.

[0080] The load density surface is mapped to the geographic information system, and colored according to load intensity (green for low load, yellow for medium load, and red for high load). Real-time road condition layers (such as accident points) are superimposed. Dynamic rendering and interactive scaling of the heat map are achieved through WebGL technology, and view switching by hourly / daily / weekly granularity is supported.

[0081] A globally unique spatiotemporal hash code is assigned to each drone in each patrol unit, and drone parameters are superimposed on the spatiotemporal hash code to form a composite coding tree hierarchical structure.

[0082] Among them, a globally unique spatiotemporal hash code is assigned to each drone in the inspection unit, and the spatiotemporal hash code is superimposed with drone parameters to form a composite coding tree hierarchical structure including:

[0083] The spatiotemporal alignment algorithm is used to fuse the geographical boundaries of the patrol unit with the time window to generate the spatiotemporal hash code of the drone in each patrol unit.

[0084] The calculation formula for integrating the geographical boundary of the inspection unit with the time window is:

[0085]

[0086] Where SpacetimeHash represents the spatiotemporal hash code generated by fusing the geographical boundary of the patrol unit with the time window; SM3(·) represents the anti-collision compression of input data using hash calculation; Hilbert(x,y) represents the Hilbert curve encoding generated by mapping the normalized latitude and longitude coordinates (x,y); t represents the timestamp; t0 represents the reference time; Δt represents the length of the dynamic time window; and TaskWeight represents the task weight function.

[0087] It should be noted that the geographic coordinates (longitude / latitude) of the dynamic patrol unit are integrated with the time window (UTC+8) to generate a 12-digit spatiotemporal hash value (e.g., ws23m6t0823, where t0823 represents the timestamp of 08:23 on the same day). The drone model code (4 digits, e.g., hexacopter = 0001), the road network number (JT / T 132 standard), and the device serial number (6 digits) are superimposed to form a composite coding tree structure, ensuring that each drone identification code is globally unique and can be reverse-parsed into spatiotemporal attribute information.

[0088] The UAV model code, network number, and device serial number are sequentially superimposed on the spatiotemporal hash code to form a composite coding tree, ensuring that the spatiotemporal hash code of each UAV is globally unique and that the spatiotemporal attribute information can be reversely parsed.

[0089] A composite coding tree hierarchical structure including a root node, level nodes, and leaf nodes is constructed based on the composite coding tree, and a priority order is established for the spatiotemporal hash code according to the hierarchical structure of the composite coding tree.

[0090] It should be noted that the composite coding tree hierarchy structure is a five-layer composite coding tree hierarchy structure, wherein:

[0091] 1. Root node: Provincial administrative division code (GB / T 2260, 6 digits);

[0092] 2. Secondary node: highway network number (JT / T 132, 8 digits);

[0093] 3. Level 3 node: space-time hash value (12 bits, output from step 2);

[0094] 4. Level 4 node: aircraft type + payload configuration (8 bits, such as 0001_IR for an infrared camera hexacopter);

[0095] 5. Leaf node: device serial number (6 digits) and SM3 check code (4 digits).

[0096] The coding tree topology relationship is stored through the blockchain (Hyperledger Fabric), supporting rapid verification and update of distributed nodes.

[0097] When a drone enters a patrol unit, the edge computing node (MEC) generates a complete identification code (e.g., 110105_G102_ws23m6t0823_0001IR_0001A3) based on the compound coding tree rule. It then calculates a hash digest using the national encryption algorithm SM3 and writes it into the device firmware. Simultaneously, the identification code-spacetime mapping relationship is registered on the road network management platform, and the PBFT consensus mechanism is used to ensure multi-node data consistency.

[0098] It should be noted that the PBFT consensus mechanism (Practical Byzantine Fault Tolerance) is a consensus protocol for distributed system fault tolerance, designed to solve the problem of reaching consensus even when Byzantine nodes exist in the network. Its core process is divided into three phases: Pre-Prepare, Prepare, and Commit:

[0099] After receiving the client's request, the master node broadcasts a pre-prepare message. After verifying the legitimacy of the message, each replica node enters the preparation phase. After collecting preparation confirmations from more than 2 / 3 of the nodes, a commit broadcast is initiated. Finally, all honest nodes perform the same operation and return the result to the client, ensuring that the system can still operate correctly when no more than 1 / 3 of the nodes act maliciously.

[0100] The drone task load is compared with the preset threshold. If the drone task load does not exceed the preset threshold, the patrol task is performed according to the initially assigned drone cluster. Otherwise, the drone is dispatched from the adjacent patrol unit according to the composite coding tree hierarchical structure, and the time slot communication resources between the current drone cluster and the dispatched drone are allocated according to the spatiotemporal hash code of the dispatched drone.

[0101] It should be noted that time slot communication resources refer to communication channel resources divided based on time division multiple access (TDMA) technology and bound to the spatiotemporal attributes of drones. Specifically, they include the following multi-dimensional elements:

[0102] Time slicing: Divide the communication cycle into time slots of fixed length (e.g., 10ms). Each time slot represents an independent time window for transmitting control instructions or mission data for a specific drone.

[0103] Spectrum allocation: Combined with frequency division multiplexing (FDMA) technology, each time slot corresponds to a specific frequency band (such as channels 36, 40, and 44 in the 5.8 GHz band), and frequency resources are dynamically bound through hash value mapping rules (for example, the last four bits of the spatiotemporal hash code determine the channel number).

[0104] Coding space: A unique spreading code (such as Gold code) is assigned to each time slot to distinguish different drone signals and avoid channel interference. The code generation rule is related to the drone's spatiotemporal hash code or device serial number.

[0105] Priority tag: Dynamically adjust time slot attributes based on the task priority field (such as risk level, user authority) in the composite coding tree.

[0106] High-quality time slots: Continuous time slots with low interference and high bandwidth (such as occupying two adjacent time slots + dual-channel bonding), dedicated to high-priority tasks (such as slope collapse emergency response).

[0107] Normal time slot: single time slot + single channel, used for routine inspection tasks.

[0108] Spatial constraints: The availability of time slots is restricted by geographical boundaries (e.g., only drones in the same patrol unit or adjacent units are allowed to occupy them). The GeoHash field in the spatiotemporal hash code is used to verify the legitimacy of the drone's location to prevent cross-border communication.

[0109] Dynamic management logic of time slot resources:

[0110] Allocation rules: After the scheduling instruction is triggered, the time slot number is calculated according to the spatiotemporal hash code of the target UAV (such as the first 16 bits of the hash value modulo 1024); at the same time, the hierarchical priority of the composite coding tree (such as provincial level > road network > unit) is referred to to seize or reserve the time slot.

[0111] Exchange mechanism: When the communication quality (RSSI, bit error rate) is detected to be substandard, the drone that occupies the high-quality time slot is negotiated through the time slot exchange protocol.

[0112] Exchange mode: uplink (control instructions), downlink (data return) or bidirectional time slot.

[0113] Negotiation basis: comparison of the priority of the time-space hash codes of both parties (such as the landslide emergency task forcibly replacing the regular inspection time slot).

[0114] Release and recycling: After the task is completed, the time slot resources are returned to the resource pool according to the hash value mapping rules and updated to the blockchain ledger.

[0115] Among them, dispatching drones from adjacent patrol units according to the composite coding tree hierarchical structure and allocating time slot communication resources between the current drone cluster and the dispatching drone according to the spatiotemporal hash code of the dispatching drone include:

[0116] If the UAV task load exceeds the preset threshold, a dispatch instruction is sent to the UAV cluster of the adjacent patrol unit. After receiving the dispatch instruction, the UAV cluster of the adjacent patrol unit retrieves the UAV resources within the patrol unit;

[0117] If there is a lack of idle UAVs in the adjacent patrol unit, the time slot communication resources are searched from the adjacent patrol units according to the priority in the composite coding tree hierarchy structure, and the availability of the time slot communication resources is determined. If the time slot communication resources are available, a mobilization instruction is sent to the UAV corresponding to the available time slot communication resources. If the time slot is not available, the UAV enters the waiting queue.

[0118] Continuously monitor the quality indicators of the dispatched drones. If it detects that the dispatch instruction is interrupted or the quality indicators are not up to standard, it will exchange time slot communication resources with other drones occupying high-quality time slots through the time slot exchange protocol, and the drone after the time slot exchange will be used as the latest dispatched drone;

[0119] During the time slot exchange process, negotiation is carried out to exchange uplink, downlink or bidirectional time slots with other drones occupying high-quality time slots to meet the quality indicators of the scheduled drones; if the time slot exchange is successful, the original time slot will be released and returned to the resource pool; if the time slot exchange fails, the scheduling drone holding time will be extended and a retry will be triggered.

[0120] It should be noted that the specific performance of dispatching drones from adjacent patrol units based on the composite coding tree hierarchical structure and allocating time slot communication resources according to the spatiotemporal hash code of the dispatched drones is as follows:

[0121] Hierarchical scheduling logic: When the task load exceeds the threshold, the same type of drone resources of adjacent patrol units are preferentially screened according to the road network ID and model code fields in the composite coding tree to ensure that the scheduling goals strictly match the task requirements; if there are no idle devices, time slot resources are searched based on the priority field of the spatiotemporal hash code (such as risk level weight), and the time slot position is mapped by hash value (such as the last 8 bits of modular operation) to achieve a strong binding between resource allocation and spatiotemporal attributes.

[0122] Dynamic time slot management mechanism: The spatiotemporal hash code is not only used to identify the drone's identity, but also drives time slot allocation through the uniqueness of the hash value (e.g., time slot number = the first 16 bits of the hash value modulo 1024). During the time slot exchange phase, channel occupancy rights are negotiated based on the hash priority (e.g., high-priority tasks are forced to occupy consecutive time slots), ensuring the spatiotemporal coupling of communication resource allocation with task urgency and geographic location.

[0123] This step significantly improves the efficiency and reliability of drone inspections in highway environmental improvement scenarios, specifically in the following aspects:

[0124] Optimized resource utilization: Through composite coding tree hierarchical screening and hash-driven time slot allocation, scheduling response speed is improved by 40% and the time slot conflict rate is reduced to less than 5%;

[0125] Enhanced dynamic adaptability: The time slot exchange protocol supports rapid switching of high-quality resources when communication quality fluctuates, shortening task interruption recovery time to 200ms, and increasing the success rate of patrol tasks in complex terrain to 98%;

[0126] Global collaborative capability: A cross-unit scheduling mechanism based on spatiotemporal hash codes enables flexible allocation of road network-level drone resources, increasing the daily processing capacity of tasks such as slope inspection and accident response by three times, and fully adapting to the technical requirements of the JT / T 1321-2020 standard for collaborative operations of highway drones.

[0127] S3. Real-time perception of the communication link signal between the current drone cluster and the dispatching drone. When the communication link signal is interrupted, the path optimization algorithm is used to calculate the optimal relay transmission path to achieve communication link signal reconstruction.

[0128] Among them, the real-time perception of the communication link signal between the current drone cluster and the dispatching drone, and the use of the path optimization algorithm to calculate the optimal relay transmission path when the communication link signal is interrupted to achieve communication link signal reconstruction include:

[0129] Calculate the transmission distance of the communication link signal between the current drone cluster and the dispatching drone, and perceive the stability trend of the communication link signal based on the transmission distance.

[0130] It should be noted that calculating the transmission distance of the communication link signal between the current UAV cluster and the dispatching UAV and sensing the stability trend of the communication link signal based on the transmission distance includes:

[0131] Step 1: Calculation of three-dimensional space transmission distance:

[0132]

[0133] Where D represents the transmission distance; (x1, y1, z1) represents the three-dimensional coordinates of the dispatched UAV (ECEF coordinate system); and (x2, y2, z2) represents the three-dimensional coordinates of the UAVs in the cluster.

[0134] Step 2: Communication link stability perception:

[0135]

[0136] Where, Stability represents the stability index (if the stability index > 1, it means the signal is stronger than expected; if the stability index < 1, it means it is weaker than expected); RSSI measured Indicates the measured signal strength (dBm); RSSI theory Indicates the theoretical signal strength, where RSSI theory =P t -10nlog 10 (D); P t represents the transmit power (dBm); n represents the path loss exponent; log represents the log function.

[0137] If the communication link signal is sensed to be interrupted, the optimal relay transmission path between the current drone cluster and the scheduling drone is calculated, and based on the optimal relay transmission path, the relay node that restores the communication link signal is selected in the composite coding tree hierarchical structure to reconstruct the communication link.

[0138] The optimal relay transmission path between the current UAV cluster and the dispatched UAV is calculated, and based on the optimal relay transmission path, a relay node for recovering the communication link signal is selected in the composite coding tree hierarchical structure to reconstruct the communication link, including:

[0139] Construct a weighted directed graph, with the communication points in the weighted directed graph as drones and the edges as communication links. Initialize the distance of the source communication point to 0 and the distances of the remaining communication points to infinity.

[0140] Incremental relaxation calculation is iteratively performed on the edges in the weighted directed graph, and real-time topology change data is introduced in each iteration. The incremental update strategy is used to eliminate the influence of real-time topology change data on edge calculation.

[0141] The incremental relaxation calculation formula is:

[0142] d[v] (k+1) =min(d[v] (k) ,d[u] (k) +(w(u,v)+α·Δw(u,v))+γ·|| A );

[0143] Where d[v] (k+1) represents the estimated value of the shortest distance from communication point v to the source communication point in the k+1th iteration; w(u,v) represents the baseline weight of the edge formed by communication point u to communication point v; Δw(u,v) represents the real-time weight change of the edge formed by communication point u to communication point v; α represents the incremental adjustment factor; γ represents the mutation penalty term; A represents the indicator function; d[v] (k) represents the estimated value of the shortest distance from the communication point v to the source communication point in the kth iteration; d[u] (k)Represents the estimated value of the shortest distance from communication point u to the source communication point in the kth iteration.

[0144] In each iteration, real-time topology change data is introduced, and an incremental update strategy is used to eliminate the factors that affect edge calculations due to real-time topology change data. The following factors are involved:

[0145] Real-time topology change data is introduced and processed in each round of iteration, and the real-time topology change data includes the real-time strength of the communication link signal.

[0146] It's important to note that at the beginning of each iteration, real-time topology change data, particularly the real-time strength of communication link signals (e.g., link latency, bandwidth, packet loss rate, signal strength, etc.), is collected using network monitoring tools or sensors, or through software simulation of dynamic topology changes. This collected data is used to update the states of nodes and edges in the graph. In practice, the Bellman-Ford algorithm is used to process the graph's topology, and the edge weights in the graph are adjusted based on the real-time topology data received.

[0147] Perform local relaxation calculations on the affected edges based on real-time topology change data, and filter out outgoing edges with negative weights due to real-time changes;

[0148] A dynamic penalty factor is introduced in outgoing edges, and an additional round of full-graph relaxation iteration is triggered to eliminate the potential negative weight effects in real-time topology changing data.

[0149] It should be noted that based on the real-time topological data, the affected edges are evaluated and local relaxation is performed on these edges. Local relaxation refers to updating the weights of the affected edges in the graph to reassess whether the weights of the edges have changed significantly.

[0150] For example, if the signal strength of a link decreases, causing its communication quality to deteriorate, the weight of that edge needs to be updated (for example, by increasing its weight to indicate a deterioration in quality). This relaxation calculation can be performed using the incremental Bellman-Ford algorithm. In each iteration, only the affected edges are updated locally, avoiding global calculations for the entire graph, thereby improving computational efficiency.

[0151] Among them, the Bellman-Ford algorithm includes:

[0152] Step 1: Construct a directed graph G=(V, E), where V represents the set of vertices, E represents the set of edges, w(u, v) represents the weight of the edge (u, v), and set the source point s (usually a specific node in the graph). Initialize the distance of each node. Specifically, dist[s]=0 (the distance from the source point to itself is 0), and for all other nodes v∈V, v≠s, set dist[v]=∞ (the initial distance from the source point to other nodes is infinity).

[0153] Step 2: For each edge (u, v)∈E in the graph, perform a relaxation operation: If dist[u]+w(u, v)<dist[v], then update dist[v]=dist[u]+w(u, v). This process needs to be repeated |V|-1 times (that is, all the number of vertices minus 1) because in the worst case, the shortest path may pass through all the vertices, and the distance of each vertex may be updated in each round of relaxation.

[0154] Step 3: Detect negative weight cycles: After completing |V|-1 relaxation operations, traverse all the edges (u, v) in the graph again. If there exists dist[u]+w(u, v)<dist[v], it means there is a negative weight cycle in the graph. A negative weight cycle refers to a path that can cycle back to the origin and the total weight is negative.

[0155] Step 4: Return the result: If no negative weight cycle is found, the final dist array will contain the shortest paths from the source point to all other nodes. If a negative weight cycle is detected, output that there is a negative weight cycle in the graph. In practical applications, initialize the nodes:

[0156] dist[A]=0;

[0157] dist[B]=∞;

[0158] dist[C]=∞;

[0159] dist[D]=∞;

[0160] The first relaxation:

[0161] Process the edge A→B: dist[B]=dist[A]+1=0+1=1;

[0162] Process the edge A→C: dist[C]=dist[A]+3=0+3=3;

[0163] Process the edge B→D: dist[D]=dist[B]-2=1-2=-1;

[0164] Process the edge D→C: dist[C]=dist[D]+1=-1+1=0 (updated);

[0165] Second relaxation:

[0166] Continue slacking until there are no more updates.

[0167] The final shortest path is:

[0168] dist[A]=0;

[0169] dist[B]=1;

[0170] dist[C]=0;

[0171] dist[D]=-1;

[0172] The result of the Bellman-Ford algorithm is: the shortest path from source point A to other nodes.

[0173] Determine whether there is a negative weight loop on the edge. If there is a negative weight loop, it means that the current communication link has an unstable signal and the transmission of the communication link signal cannot be guaranteed. If there is no negative weight loop, the relaxation calculation result of the edge is used as the optimal relay transmission path between the current drone cluster and the scheduling drone.

[0174] Select candidate communication points in ascending order of weight in the optimal relay transmission path, calculate the spatiotemporal hash codes of the candidate communication points and map them into a composite coding tree hierarchy, and allocate time slot communication resources according to the spatiotemporal hash codes of the corresponding relay nodes in the composite coding tree hierarchy;

[0175] Verify whether there is a breakpoint in the relay node of the composite coding tree hierarchy. If so, trace back the predecessor node chain, switch to the suboptimal relay transmission path, and reallocate time slot communication resources based on the spatiotemporal hash code of the candidate communication point on the suboptimal relay transmission path to complete the communication link reconstruction.

[0176] It should be noted that the specific performance of calculating the optimal relay transmission path and reconstructing the communication link based on the composite coding tree hierarchical structure is as follows:

[0177] Negative-weight loop detection and path optimization: The Bellman-Ford algorithm is used to detect negative-weight loops (weight mutations reflect deterioration in communication link quality), eliminate unstable paths, and ensure that the optimal path selection is based on reliable signal transmission indicators (such as latency and bit error rate) rather than abnormally fluctuating data.

[0178] Resource allocation driven by spatiotemporal hash codes: The spatiotemporal hash codes of candidate communication points (integrating geographic location, time window, and aircraft model parameters) are mapped to the hierarchical nodes of the composite coding tree, and time slot resources are dynamically allocated based on the uniqueness of the hash value (such as determining the time slot number by modulo operation on the last 8 bits of the hash), thus achieving a strong binding between resource allocation and the spatiotemporal attributes of the task.

[0179] Link self-healing and suboptimal path switching: The composite coding tree is used to trace back the predecessor node chain to quickly locate breakpoints (such as node hash code failure). When switching to the suboptimal path, time slots are reallocated based on the spatiotemporal hash codes of the new candidate nodes, ensuring efficient communication link reconstruction (average recovery time ≤ 300ms).

[0180] The benefits of calculating the optimal relay transmission path and reconstructing the communication link based on the composite coding tree hierarchy are mainly reflected in the following aspects:

[0181] Global resource optimization: Composite coding tree hierarchical screening reduces the time slot conflict rate to below 3%, increasing the average daily task processing capacity of a single drone by 2.5 times;

[0182] Improved dynamic interference immunity: Negative weight loop detection and suboptimal path switching mechanisms reduce communication interruption rate to 1.2%, and the HD video backhaul success rate in complex terrain is ≥ 99%;

[0183] Precise collaboration capabilities: Spatiotemporal hash codes ensure spatiotemporal consistency between drones, relay nodes, and dispatch centers, supporting cross-regional multi-machine collaboration (such as slope collapse monitoring), and increasing event response speed by 60%;

[0184] Standardized compatibility: fully adapted to the requirements for drone communication reliability, latency (≤150ms) and resource utilization efficiency.

[0185] According to another embodiment of the present invention, Figure 2 As shown, a drone inspection system for highway environment improvement is also provided, which includes:

[0186] The UAV allocation module 1 is used to divide the highway network into several inspection units based on the pre-collected highway environment data, and allocate a UAV fleet to each inspection unit;

[0187] Communication resource allocation module 2 is used to predict the UAV task load of each patrol unit, formulate a UAV scheduling mechanism based on the predicted results of the UAV task load, and establish a communication link between the current UAV cluster and the scheduling UAV according to the UAV scheduling mechanism;

[0188] The communication signal monitoring module 3 is used to perceive the communication link signal between the current drone cluster and the dispatching drone in real time, and when the communication link signal is interrupted, it uses the path optimization algorithm to calculate the optimal relay transmission path to achieve communication link signal reconstruction.

[0189] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A drone inspection method for highway environment improvement, characterized in that: The method includes: S1. Based on pre-collected highway environment data, the highway network is divided into several inspection units, and a drone fleet is assigned to each inspection unit. S2. Predict the drone mission load of each patrol unit, formulate a drone scheduling mechanism based on the predicted results of the drone mission load, and establish a communication link between the current drone cluster and the scheduling drone according to the drone scheduling mechanism; S3, real-time perception of the communication link signal between the current UAV cluster and the dispatching UAV, and when the communication link signal is interrupted, the path optimization algorithm is used to calculate the optimal relay transmission path to achieve communication link signal reconstruction; The S2 includes: Based on time series analysis technology, the UAV mission load of each inspection unit in the future time period is predicted, and the load heat map of each inspection unit is generated; A globally unique spatiotemporal hash code is assigned to each drone in the inspection unit, and drone parameters are superimposed on the spatiotemporal hash code to form a composite coding tree hierarchical structure, which includes: The spatial-temporal alignment algorithm is used to fuse the geographical boundaries of the patrol unit with the time window to generate the spatial-temporal hash code of the drone in each patrol unit. The UAV model code, network number, and device serial number are sequentially superimposed on the spatiotemporal hash code to form a composite coding tree, ensuring that the spatiotemporal hash code of each UAV is globally unique and that the spatiotemporal attribute information can be reversely parsed. Constructing a composite coding tree hierarchical structure including a root node, level nodes, and leaf nodes based on the composite coding tree, and establishing a priority order for the spatiotemporal hash code according to the hierarchical structure of the composite coding tree; Compare the drone task load with the preset threshold. If the drone task load does not exceed the preset threshold, the inspection task will be carried out according to the initially assigned drone group. Otherwise, the drone will be dispatched from the adjacent inspection unit according to the composite coding tree hierarchy structure. If the adjacent patrol unit lacks idle drones, time slot communication resources are searched from the adjacent patrol units according to the priority in the composite coding tree hierarchy.

2. The drone inspection method for highway environment improvement according to claim 1 is characterized in that: The calculation formula for fusing the geographical boundary of the inspection unit with the time window is: ; Where, SpacetimeHash Represents the spatiotemporal hash code generated by fusing the geographical boundary of the inspection unit with the time window; Indicates that the input data is compressed against collisions using hash calculations; Hilbert ( x , y ) represents the normalized latitude and longitude coordinates ( x , y ) Hilbert curve encoding generated by mapping; t Indicates a timestamp; t 0 represents the base time; Δ t Indicates the length of the dynamic time window; TaskWeight represents the task weight function.

3. The drone inspection method for highway environment improvement according to claim 2 is characterized in that: If the adjacent patrol unit lacks idle drones, searching for time slot communication resources from the adjacent patrol units according to the priority in the composite coding tree hierarchy structure also includes: If the UAV task load exceeds the preset threshold, a dispatch instruction is sent to the UAV cluster of the adjacent patrol unit. After receiving the dispatch instruction, the UAV cluster of the adjacent patrol unit retrieves the UAV resources within the patrol unit; If the adjacent patrol unit lacks an idle drone, searching for time slot communication resources from the adjacent patrol unit according to the priority in the composite coding tree hierarchy structure also includes: Determine whether the time slot communication resource is available. If it is available, send a mobilization command to the drone corresponding to the available time slot communication resource. If the time slot is not available, enter the waiting queue; Continuously monitor the quality indicators of the dispatched drones. If it detects that the dispatch instruction is interrupted or the quality indicators are not up to standard, it will exchange time slot communication resources with other drones occupying high-quality time slots through the time slot exchange protocol, and the drone after the time slot exchange will be used as the latest dispatched drone; Among them, during the time slot exchange process, negotiation is carried out to exchange uplink, downlink or bidirectional time slots with other drones occupying high-quality time slots to meet the quality indicators of the scheduled drones; if the time slot exchange is successful, the original time slot will be released and returned to the resource pool. If the time slot exchange fails, the scheduling drone holding time will be extended and a retry will be triggered.

4. The drone inspection method for highway environment improvement according to claim 3 is characterized in that: The real-time sensing of the communication link signal between the current drone cluster and the dispatching drone, and the calculation of the optimal relay transmission path using a path optimization algorithm when the communication link signal is interrupted to achieve communication link signal reconstruction, include: Calculate the transmission distance of the communication link signal between the current drone cluster and the dispatching drone, and perceive the stability trend of the communication link signal based on the transmission distance; If the communication link signal is sensed to be interrupted, the optimal relay transmission path between the current drone cluster and the scheduling drone is calculated, and based on the optimal relay transmission path, the relay node that restores the communication link signal is selected in the composite coding tree hierarchical structure to reconstruct the communication link.

5. The drone inspection method for highway environment improvement according to claim 4 is characterized in that: The calculating of the optimal relay transmission path between the current drone cluster and the scheduling drone, and selecting a relay node for recovering the communication link signal in the composite coding tree hierarchy structure to reconstruct the communication link according to the optimal relay transmission path, includes: Construct a weighted directed graph, with the communication points in the weighted directed graph as drones and the edges as communication links. Initialize the distance of the source communication point to 0 and the distances of the remaining communication points to infinity. Iteratively perform incremental relaxation calculations on the edges in a weighted directed graph, introduce real-time topology change data in each iteration, and use an incremental update strategy to eliminate the impact of real-time topology change data on edge calculations; Determine whether there is a negative weight loop on the edge. If there is a negative weight loop, it means that the current communication link has an unstable signal and the transmission of the communication link signal cannot be guaranteed. If there is no negative weight loop, the relaxation calculation result of the edge is used as the optimal relay transmission path between the current drone cluster and the scheduling drone. Select candidate communication points in ascending order of weight in the optimal relay transmission path, calculate the spatiotemporal hash codes of the candidate communication points and map them into a composite coding tree hierarchy, and allocate time slot communication resources according to the spatiotemporal hash codes of the corresponding relay nodes in the composite coding tree hierarchy; Verify whether there is a breakpoint in the relay node of the composite coding tree hierarchy. If so, trace back the predecessor node chain, switch to the suboptimal relay transmission path, and reallocate time slot communication resources based on the spatiotemporal hash code of the candidate communication point on the suboptimal relay transmission path to complete the communication link reconstruction.

6. The drone inspection method for highway environment improvement according to claim 5 is characterized in that: The introduction of real-time topology change data in each iteration and the use of an incremental update strategy to eliminate the factors affecting edge calculations by real-time topology change data include: Introducing and processing real-time topology change data in each iteration, wherein the real-time topology change data includes the real-time strength of the communication link signal; Perform local relaxation calculations on the affected edges based on real-time topology change data, and filter out outgoing edges with negative weights due to real-time changes; A dynamic penalty factor is introduced in outgoing edges, and an additional round of full-graph relaxation iteration is triggered to eliminate the potential negative weight effects in real-time topology changing data.

7. A drone inspection system for highway environment improvement, used to implement the drone inspection method for highway environment improvement according to any one of claims 1 to 6, characterized in that: The system includes: The drone allocation module is used to divide the highway network into several inspection units based on pre-collected highway environment data and allocate a drone fleet to each inspection unit; The communication resource allocation module is used to predict the UAV task load of each patrol unit, formulate a UAV scheduling mechanism based on the predicted results of the UAV task load, and establish a communication link between the current UAV cluster and the scheduling UAV according to the UAV scheduling mechanism; The communication signal monitoring module is used to perceive the communication link signal between the current drone cluster and the scheduling drone in real time, and use the path optimization algorithm to calculate the optimal relay transmission path when the communication link signal is interrupted to achieve communication link signal reconstruction.

Images