Unmanned aerial vehicle data transmission system based on information importance grading

Abstract

The invention discloses an unmanned aerial vehicle data transmission system based on information importance grading. In the invention, through dynamic priority division and intelligent resource scheduling, the execution reliability of the key task in a complex environment is obviously improved. The system preferentially guarantees stable transmission of first-level data (such as positions and emergency instructions) in a weak signal scene, and a standby machine is quickly switched to a relay node or a cluster topological structure is adjusted, so that task failure caused by communication interruption is avoided. Even if the signal strength of the sub-machine is suddenly reduced, the system can still ensure real-time return of key instructions, meanwhile, unnecessary data are automatically compressed or delayed, bandwidth occupation is reduced, the task success rate is effectively improved, and the overall stability and adaptability of a communication network are enhanced through predictive resource deployment and dynamic path optimization. The intelligent relay scheduling module combines real-time environment data and task requirements, deploys a standby machine to a high-risk area in advance, and reduces the risk of communication interruption caused by terrain shielding or weather interference.

Description

Technical Field

[0001] The present invention belongs to the technical field of UAV communication, and specifically relates to a UAV data transmission system based on information importance classification. Background Technique

[0002] UAV data transmission refers to the wireless data communication process carried out by UAVs during flight to achieve information interaction with ground control stations or other devices. It is the core technical support for UAV systems to achieve remote control, status monitoring, task execution, and intelligence acquisition. Through the equipped communication module, UAVs can upload flight status parameters (such as position, altitude, speed, attitude), data collected by sensors (such as video, images, environmental monitoring data), and mission payload information in real-time or near real-time; at the same time, it also receives downlink data from ground control stations or preset instructions, including flight instructions, mission planning parameters, navigation correction information, etc. The data transmission link usually adopts wireless communication technologies such as Wi-Fi, Bluetooth, 4G / 5G, satellite communication, or dedicated data transmission radios, and its performance directly affects the control distance of UAVs, the real-time performance of mission payloads, anti-interference ability, and overall combat effectiveness or operation efficiency.

[0003] However, in the prior art, UAV cluster communication systems generally lack dynamic adaptive adjustment of task priorities and real-time signal strengths, resulting in unstable transmission of critical data in weak signal environments and an inability to flexibly adjust transmission content according to the urgency of tasks; at the same time, traditional solutions rely on fixed rules in communication resource scheduling, making it difficult to intelligently predict environmental interference or deploy redundant links in advance, resulting in communication interruptions easily caused by factors such as terrain occlusion and weather changes in complex scenarios, with low task fault tolerance and resource utilization. Summary of the Invention

[0004] The purpose of the present invention is to provide a UAV data transmission system based on information importance classification to solve the above-mentioned problems.

[0005] The technical solution adopted by the present invention is as follows: A UAV data transmission system based on information importance classification, the system includes: a signal strength monitoring and classification module, a data importance classification module, a dynamic adaptive transmission control module, an intelligent communication scheduling module, and a data processing and feedback module;

[0006] The intelligent communication scheduling module is internally provided with a network status real-time monitoring sub-module, a relay path optimization decision sub-module, a node dynamic management sub-module, a cluster topology dynamic adjustment sub-module, and a predictive resource deployment sub-module;

[0007] The output end of the signal strength monitoring and classification module is connected to the signal input interface of the dynamic adaptive transmission control module;

[0008] The output end of the data importance grading module is connected to the data priority connection of the dynamic adaptive transmission control module;

[0009] The transmission strategy instruction output end of the dynamic adaptive transmission control module is connected to the path optimization interface of the intelligent communication scheduling module;

[0010] The network status feedback end of the intelligent communication scheduling module is connected to the real-time data interface of the data processing and feedback module;

[0011] The global task view output end of the data processing and feedback module is bidirectionally connected to the threshold calibration interface of the signal strength monitoring and grading module and the priority update interface of the data importance grading module respectively.

[0012] In a preferred embodiment, the signal strength monitoring and grading module consists of a signal acquisition unit, a threshold determination unit, and a dynamic calibration unit. The signal acquisition unit obtains the communication signal strength with the host in real time through the wireless sensors deployed on the slave machine and transmits the original signal strength to the threshold determination unit. The threshold determination unit grades the signals according to the preset three-level thresholds of weak signal, medium signal, and strong signal. The dynamic calibration unit combines the requirements of the task scenario to improve the adaptability of signal grading to the dynamic environment. The grading results are output to the network status heat map in real time to mark the positions of the weak signal slave machines and the signal attenuation trend.

[0013] In a preferred embodiment, the data importance grading module is provided with a task classification engine, a user-defined interface, and a priority mapping unit. The task classification engine automatically divides the data levels according to the task types. The user-defined interface allows the operator to adjust the grading criteria through the control background. The priority mapping unit binds the data levels to the transmission strategies.

[0014] In a preferred embodiment, the dynamic adaptive transmission control module includes: a signal-data matching unit, a bandwidth allocation unit, and a standby machine trigger logic. The signal-data matching unit selects the transmission content according to the real-time signal strength level. The bandwidth allocation unit adopts a probability feedback mechanism. Under weak signals, it calculates the task completion probability through a Bayesian network. If the probability is lower than 10%, it triggers the intervention of the standby machine. Otherwise, it compresses the redundant data according to the priority. The standby machine trigger logic integrates a reinforcement learning model to analyze the position, power, and link quality of the standby machine, and preferentially calls the node that is the closest and has a power higher than 60% as a relay. The module reduces the communication load by 20% by adaptively discarding the three-level non-critical data.

[0015] In a preferred embodiment, the network status real-time monitoring sub-module generates a dynamic network quality assessment report by collecting in real time the communication status data of all sub-aircraft in the UAV cluster (including signal strength, packet loss rate, latency, position coordinates, and task priority), and combining the global information received by the host. This module uses multi-dimensional data fusion technology to associate discrete communication metrics with task requirements. After the monitoring data is normalized, it outputs a network quality report containing the real-time signal attenuation trend, the health status of the sub-aircraft, and the risk of task interruption, providing a decision-making basis for subsequent relay scheduling and topology adjustment.

[0016] The comprehensive communication score of the sub-aircraft is obtained by multiplying the normalized value of the real-time signal strength by the signal weight coefficient, adding the product of the task priority coefficient and the task weight, and then subtracting the product of the packet loss rate and the penalty factor. The calculation formula of the signal strength weighted scoring model is:

[0017] Q = S·w s +P·w p -L·α;

[0018] Where S represents the normalized value of the real-time signal strength, with a range of [0,1][0,1], which is linearly mapped from the signal strength measurement value (such as RSSI).

[0019] w s represents the signal weight coefficient, which is dynamically adjusted according to the task scenario (ws = 0.6 for disaster rescue, ws = 0.4 for regular tasks).

[0020] P represents the task priority coefficient, which is defined by the data classification module (P = 1.0 for first-level tasks, P = 0.7 for second-level tasks, P = 0.3 for third-level tasks).

[0021] w p represents the task weight, which is fixed at 0.3 to balance the impact of task priority on the score.

[0022] L represents the packet loss rate (percentage), with a value range of [0,100][0,100].

[0023] α represents the packet loss penalty factor, with α = 0.05 corresponding to every 1% packet loss rate, and the total penalty upper limit is Qmax·0.3.

[0024] In a preferred embodiment, the relay path optimization decision sub-module dynamically analyzes the network status and task requirements of the UAV cluster through a reinforcement learning algorithm, and calculates the optimal communication path in real time. This module takes the network quality report, task importance grading data, and the available status of standby machines as inputs, and defines a decision trigger mechanism based on the signal strength threshold and task priority: when it is detected that a single sub-machine is in a weak signal but undertakes a critical task, the standby machine is preferentially enabled to switch to a relay node to establish a redundant communication link; if multiple sub-machines are simultaneously in a weak signal area, the cluster topology adjustment algorithm is called to drive the host to move towards the weak signal area to shorten the communication distance. During the decision-making process, the module quantifies the balance relationship among network stability, task completion rate, and resource consumption through the reward function in the reinforcement learning model, and combines real-time environmental data to predict the future signal attenuation trend, autonomously selects the relay enabling or topology adjustment strategy, and dynamically optimizes the decision weight to ensure the transmission continuity of first-level critical data is preferentially guaranteed when communication resources are limited;

[0025] The calculation formula for the dynamic path reward function is as follows:

[0026] R = w s ·S + w p ·P - α·L;

[0027] Where:

[0028] S represents the normalized value of the real-time signal strength (range 0 ≤ S ≤ 1), which is calculated by linearly mapping the RSSI values of the sub-machine and the host.

[0029] w s represents the signal strength weight (default ws = 0.6, increased to 0.8 in the disaster rescue scenario).

[0030] P represents the task priority coefficient (P = 1.0 for first-level tasks, P = 0.7 for second-level tasks, P = 0.3 for third-level tasks).

[0031] w p represents the task importance weight (fixed wp = 0.4).

[0032] L represents the packet loss rate (percentage, 1000 ≤ L ≤ 100).

[0033] α represents the packet loss penalty factor (α = 0.05 corresponding to each 1% packet loss rate)

[0034] The calculation formula for the predictive resource allocation model is:

[0035]

[0036] β represents the historical signal attenuation coefficient.

[0037] γ represents the environmental interference factor (range 1.0 ≤ γ ≤ 2.0), integrating terrain complexity (DEM elevation) and weather data (such as rainfall intensity).

[0038] τ represents the task time urgency factor (τ = remaining task time / total task duration, for urgent tasks τ > 0.8).

[0039] T represents the remaining task duration (unit: minutes).

[0040] In a preferred embodiment, the node dynamic management sub-module consists of a status monitoring unit, a mode switching unit, and a link maintenance unit. The status monitoring unit collects the battery power, communication load, and task priority data of the standby machine in real time, and identifies abnormal nodes through a threshold determination mechanism. The mode switching unit dynamically adjusts the role of the standby machine according to the network status heat map and task urgency. In the initial state, the standby machine executes regular tasks. When it detects that the main link signal strength continuously drops below the preset threshold, it immediately switches to the relay mode to build a redundant link to preferentially transmit first-level critical data. The link maintenance unit continuously monitors the communication quality between the slave machine and the host machine in the relay mode. If the packet loss rate exceeds 10% or the relay node's battery power is insufficient, it automatically triggers the seamless switching process of the standby machine.

[0041] In a preferred embodiment, the cluster topology dynamic adjustment sub-module minimizes the total signal attenuation of the communication link by optimizing the host position and the cluster formation shape. This module takes the real-time position of the slave machine, the host movement ability limit, and the environmental obstacle data as inputs, calculates the host movement direction based on the gradient descent algorithm, and the objective function is the weighted sum of the signal attenuation coefficient and the square of the distance. The gradient direction is generated by the partial derivative of the objective function with respect to the host coordinates, and a momentum acceleration mechanism is introduced: if the gradient directions are the same for three consecutive times, the host movement speed is increased by 20% to quickly approach the weak signal area. The obstacle avoidance logic detects obstacles through LiDAR or an environmental map, and superimposes a dynamic repulsive vector on the gradient direction. When the distance to the obstacle is less than 5 meters, the repulsive weight is increased to 3 times to forcefully avoid the collision risk. After the adjustment is completed, the module outputs the host movement instruction and the cluster formation parameters (such as honeycomb or star topology) to ensure the shortest backbone link and maintain the global communication stability;

[0042] The calculation formula of the momentum acceleration communication loss optimization model is as follows:

[0043]

[0044] Where v represents the actual movement speed of the host (unit: m / s).

[0045] m represents the momentum acceleration factor, with an initial value of 1, and it increases to 1.2 if the gradient directions are the same for three consecutive times.

[0046] v baseRepresents the basic moving speed defined by the UAV hardware.

[0047] η represents the learning rate, which is fixed at 0.1 and controls the accuracy of the moving step size.

[0048] Represents the partial derivative of the communication loss function with respect to the host coordinates, reflecting the rate of change of signal attenuation with position.

[0049] The calculation formula of the dynamic repulsive vector obstacle avoidance model is:

[0050]

[0051] Where F represents the obstacle avoidance repulsive force (unit: N), and the direction is away from the obstacle.

[0052] d represents the real-time distance between the obstacle and the host (unit: m).

[0053] w terrain Represents the terrain complexity weight, which is 1.5 for mountainous terrain and 1.0 for plain terrain.

[0054] k repel Represents the dynamic repulsion coefficient, which is increased to 3 when the obstacle distance is less than 5 meters, otherwise it is 1.

[0055] In a preferred embodiment, the predictive resource deployment sub-module is provided with a historical data analysis unit, an environment modeling unit, and a resource pre-allocation engine. The historical data analysis unit extracts the signal attenuation law from the communication log, the environment modeling unit fuses the real-time terrain elevation, meteorological data, and obstacle distribution to generate a multi-dimensional interference assessment map, and the resource pre-allocation engine dynamically deploys standby machines based on the prediction results. If the predicted value of the signal attenuation rate exceeds 30% and the remaining task time is less than 10 minutes, 2 standby machines are scheduled in advance to high-risk areas, and 20% of the bandwidth is reserved specifically for the first-level data transmission.

[0056] In a preferred embodiment, the data processing and feedback module consists of a data fusion engine, a decision generation unit, and an exception response unit. The data fusion engine integrates the information transmitted back by multiple sub-machines to generate a global task view. The decision generation unit sends instructions to the control background based on the task importance grading results. The exception response unit has a built-in threshold warning mechanism.

[0057] In summary, due to the adoption of the above technical solutions, the beneficial effects of the present invention are:

[0058] 1. In the present invention, through dynamic priority division and intelligent resource scheduling, the execution reliability of critical tasks in complex environments is significantly improved. The system preferentially guarantees the stable transmission of first-level data (such as location and emergency instructions) in weak signal scenarios. By quickly switching the standby machine to a relay node or adjusting the cluster topology structure, it avoids task failures caused by communication interruptions. For example, in highly dynamic scenarios such as disaster relief, even if the signal strength of the slave machine drops suddenly, the system can still ensure the real-time feedback of critical instructions, while automatically compressing or delaying unnecessary data to reduce bandwidth occupancy, effectively improving the task success rate.

[0059] 2. In the present invention, through predictive resource deployment and dynamic path optimization, the overall stability and adaptability of the communication network are enhanced. The intelligent relay scheduling module combines real-time environmental data with task requirements to pre-deploy standby machines to high-risk areas in advance, reducing the risk of communication interruptions caused by terrain occlusion or weather interference. At the same time, the module supports the parallel execution of multiple schemes, and the activation of the relay node is synchronized with the adjustment of the host position to ensure the redundancy and continuity of the critical data feedback link. This design enables the UAV cluster to cooperate efficiently in scenarios such as complex terrain exploration and large-scale inspection, reducing the frequency of manual intervention and extending the effective operation time of critical equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is the overall system block diagram of the present invention;

[0061] Figure 2 is the system block diagram of the intelligent communication scheduling module in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0063] Embodiment:

[0064] Refer to Figure 1-2 , a UAV data transmission system based on information importance classification, the system includes: a signal strength monitoring and classification module, a data importance classification module, a dynamic adaptive transmission control module, an intelligent communication scheduling module, and a data processing and feedback module;

[0065] The intelligent communication scheduling module internally sets a real-time network status monitoring sub-module, a relay path optimization decision sub-module, a node dynamic management sub-module, a cluster topology dynamic adjustment sub-module, and a predictive resource deployment sub-module;

[0066] The output end of the signal strength monitoring and classification module is connected to the signal input interface of the dynamic adaptive transmission control module;

[0067] The output end of the data importance grading module is connected to the data priority connection of the dynamic adaptive transmission control module;

[0068] The transmission strategy instruction output end of the dynamic adaptive transmission control module is connected to the path optimization interface of the intelligent communication scheduling module;

[0069] The network status feedback end of the intelligent communication scheduling module is connected to the real-time data interface of the data processing and feedback module;

[0070] The global task view output end of the data processing and feedback module is bidirectionally connected to the threshold calibration interface of the signal strength monitoring and grading module and the priority update interface of the data importance grading module respectively.

[0071] The signal strength monitoring and grading module consists of a signal acquisition unit, a threshold determination unit, and a dynamic calibration unit. The signal acquisition unit obtains the communication signal strength with the host in real time through wireless sensors deployed on the slave machine, such as RSSI values or signal-to-noise ratio data, and transmits the original signal strength to the threshold determination unit. The threshold determination unit classifies the signals according to the preset three-level thresholds of weak signal, medium signal, and strong signal. For example, a weak signal is defined as a strength below 20%, a medium signal is 20% - 80%, and a strong signal is above 80%. The dynamic calibration unit combines the task scenario requirements. For example, during disaster rescue, it automatically reduces the weak signal determination threshold to 15% to improve the adaptability of signal classification to the dynamic environment. The classification result is output to the network status heat map in real time, marking the positions of the weak signal slave machines and the signal attenuation trend.

[0072] The data importance grading module is provided with a task classification engine, a user-defined interface, and a priority mapping unit. The task classification engine automatically divides the data levels according to the task types. For example, level 1 data includes ID, location, and emergency control instructions, level 2 data covers speed, heading, and battery status, and level 3 data includes sensor logs and attitude parameters. The user-defined interface allows operators to adjust the grading criteria through the control background. For example, in exploration tasks, IMU data is promoted to level 2. The priority mapping unit binds the data levels to the transmission strategies. For example, level 1 data is forced to be encrypted and transmitted preferentially, and level 3 data is allowed to be delayed or compressed. The module supports dynamic task adaptation. For example, when a sudden task is triggered, the newly added instructions are automatically classified as level 1 data.

[0073] The dynamic adaptive transmission control module includes: a signal-data matching unit, a bandwidth allocation unit, and a standby machine trigger logic. The signal-data matching unit selects the transmission content according to the real-time signal strength level. For example, only the first-level data is sent in the case of weak signals, and the second-level data is allowed to be transmitted at a 50% bandwidth occupancy rate in the case of medium signals. The bandwidth allocation unit adopts a probability feedback mechanism. Under weak signals, the probability of task completion is calculated through a Bayesian network. If the probability is less than 10%, the standby machine is triggered to intervene; otherwise, the redundant data is compressed according to the priority. The standby machine trigger logic integrates a reinforcement learning model, analyzes the position, power, and link quality of the standby machine, and preferentially calls the node that is the closest and has a power higher than 60% as a relay. The module reduces the communication load by 20% by adaptively discarding the third-level non-critical data.

[0074] The real-time network status monitoring sub-module generates a dynamic network quality assessment report by collecting the communication status data of all sub-machines in the UAV cluster in real time (including signal strength, packet loss rate, delay, position coordinates, and task priority), and combining the global information received by the host machine. This module uses multi-dimensional data fusion technology to associate discrete communication metrics with task requirements. For example, weak connection sub-machines are automatically marked based on the signal strength threshold, and the distribution of weak signal areas is visualized through a heat map. After the monitoring data is normalized, a network quality report containing the real-time signal attenuation trend, sub-machine health status, and task interruption risk is output, providing a decision-making basis for subsequent relay scheduling and topology adjustment.

[0075] The comprehensive communication score of the sub-machine is obtained by multiplying the normalized value of the real-time signal strength by the signal weight coefficient, adding the product of the task priority coefficient and the task weight, and then subtracting the product of the packet loss rate and the penalty factor. The calculation formula of the signal strength weighted scoring model is:

[0076] Q = S·w s +P·w p -L·α;

[0077] Where S represents the normalized value of the real-time signal strength, with a range of [0,1][0,1], which is linearly mapped from the signal strength measurement value (such as RSSI).

[0078] w s represents the signal weight coefficient, which is dynamically adjusted according to the task scenario (ws = 0.6 for disaster rescue, ws = 0.4 for regular tasks).

[0079] P represents the task priority coefficient, which is defined by the data classification module (P = 1.0 for first-level tasks, P = 0.7 for second-level tasks, P = 0.3 for third-level tasks).

[0080] w p represents the task weight, which is fixed at 0.3 to balance the impact of task priority on the score.

[0081] Let \(L\) represent the packet loss rate (percentage), and its value range is \([0, 100]\).

[0082] Let \(\alpha\) represent the packet loss penalty factor. For every 1% packet loss rate, \(\alpha = 0.05\), and the total penalty upper limit is \(Q_{max}\cdot0.3\).

[0083] The relay path optimization decision sub-module dynamically analyzes the network status and task requirements of the UAV cluster through the reinforcement learning algorithm, and calculates the optimal communication path in real-time. This module takes the network quality report, task importance grading data, and the available status of standby machines as inputs, and defines the decision trigger mechanism according to the signal strength threshold and task priority: when it is detected that a single sub-machine is in a weak signal but undertakes a key task, the standby machine is preferentially enabled to switch to a relay node to establish a redundant communication link; if multiple sub-machines are simultaneously in a weak signal area, the cluster topology adjustment algorithm is called to drive the host to move towards the weak signal area to shorten the communication distance. During the decision-making process, the module quantifies the balance relationship among network stability, task completion rate, and resource consumption through the reward function in the reinforcement learning model, and at the same time combines real-time environmental data to predict the future signal attenuation trend, autonomously selects the relay enabling or topology adjustment strategy, and dynamically optimizes the decision weights to ensure the transmission continuity of first-level key data is preferentially guaranteed when communication resources are limited;

[0084] Among them, the calculation formula of the dynamic path reward function is:

[0085] \(R = w\) s \(\cdot S+w\) p \(\cdot P-\alpha\cdot L\);

[0086] Among them:

[0087] Let \(S\) represent the normalized value of the real-time signal strength (range \(0\leq S\leq 1\)), which is calculated by the linear mapping of the RSSI values of the sub-machine and the host.

[0088] \(w\) s represents the signal strength weight (default \(w_s = 0.6\), increased to \(0.8\) in the disaster rescue scenario).

[0089] Let \(P\) represent the task priority coefficient (first-level task \(P = 1.0\), second-level \(P = 0.7\), third-level \(P = 0.3\)).

[0090] \(w\) p represents the task importance weight (fixed \(w_p = 0.4\)).

[0091] Let \(L\) represent the packet loss rate (percentage, \(0\leq L\leq 100\)).

[0092] Let \(\alpha\) represent the packet loss penalty factor (for every 1% packet loss rate, \(\alpha = 0.05\))

[0093] The calculation formula of the predictive resource allocation model is:

[0094]

[0095] β represents the historical signal attenuation coefficient (extracted based on similar scenario logs, for example, β = 1.5 in mountainous areas and β = 1.0 in plains).

[0096] γ represents the environmental interference factor (range 1.0 ≤ γ ≤ 2.0), integrating terrain complexity (DEM elevation) and weather data (such as rainfall intensity).

[0097] τ represents the task time urgency factor (τ = remaining task time / total task duration, for urgent tasks τ > 0.8).

[0098] T represents the remaining task duration (unit: minutes).

[0099] The node dynamic management sub-module consists of a status monitoring unit, a mode switching unit, and a link maintenance unit. The status monitoring unit collects real-time data on the battery power, communication load, and task priority of the standby machine, and identifies abnormal nodes through a threshold judgment mechanism. For example, when the battery power is below 20% or the packet loss rate exceeds 15%, it is marked as a high-risk node. The mode switching unit dynamically adjusts the role of the standby machine according to the network status heat map and task urgency. In the initial state, the standby machine executes regular tasks. When it detects that the signal strength of the main link continuously drops below the preset threshold, it immediately switches it to the relay mode to construct a redundant link to preferentially transmit first-level critical data. The link maintenance unit continuously monitors the communication quality between the sub-machine and the host in the relay mode. If the packet loss rate exceeds 10% or the relay node's battery power is insufficient, it automatically triggers the seamless switching process of the standby machine, preferentially selects the node with the best signal coverage and remaining battery power above 50% to take over the task, and simultaneously updates the network topology status to ensure the communication redundancy and stability of high-priority areas.

[0100] The cluster topology dynamic adjustment sub-module minimizes the total signal attenuation of the communication link by optimizing the host position and cluster formation. This module takes the real-time position of the sub-machine, the host's movement ability limit, and environmental obstacle data as inputs, calculates the host's movement direction based on the gradient descent algorithm, and the objective function is the weighted sum of the signal attenuation coefficient and the square of the distance. The gradient direction is generated by the partial derivative of the objective function with respect to the host coordinates, and a momentum acceleration mechanism is introduced: if the gradient directions are the same three times in a row, the host's movement speed increases by 20% to quickly approach the weak signal area. The obstacle avoidance logic detects obstacles through LiDAR or environmental maps, and superimposes a dynamic repulsive vector on the gradient direction. When the distance to the obstacle is less than 5 meters, the repulsive weight is increased to 3 times to forcefully avoid the collision risk. After the adjustment is completed, the module outputs the host movement instruction and cluster formation parameters (such as honeycomb or star topology) to ensure the shortest backbone link while maintaining global communication stability;

[0101] Among them, the calculation formula of the momentum acceleration communication loss optimization model is as follows:

[0102]

[0103] Among them, v represents the actual moving speed of the host (unit: m / s).

[0104] m represents the momentum acceleration factor, with an initial value of 1. If the gradient directions are the same for three consecutive times, it will increase to 1.2.

[0105] v base represents the basic moving speed limited by the UAV hardware (for example, 5 m / s).

[0106] η represents the learning rate, which is fixed at 0.1 and controls the accuracy of the moving step size.

[0107] represents the partial derivative of the communication loss function with respect to the host coordinates, reflecting the rate of change of signal attenuation with position.

[0108] The calculation formula of the dynamic repulsive vector obstacle avoidance model is as follows:

[0109]

[0110] Among them, F represents the obstacle avoidance repulsive force (unit: N), and the direction is away from the obstacle.

[0111] d represents the real-time distance between the obstacle and the host (unit: m).

[0112] w terrain represents the terrain complexity weight, which is 1.5 for mountainous terrain and 1.0 for plain terrain.

[0113] k repel represents the dynamic repulsion coefficient, which is increased to 3 when the distance to the obstacle is less than 5 meters, otherwise it is 1.

[0114] The predictive resource deployment sub-module is provided with a historical data analysis unit, an environment modeling unit, and a resource pre-allocation engine. The historical data analysis unit extracts the signal attenuation law from the communication log. The environment modeling unit fuses the real-time terrain elevation, meteorological data, and obstacle distribution to generate a multi-dimensional interference assessment map. For example, when the rainfall intensity is higher than 5 mm / h, the environmental interference factor is increased to 1.8. The resource pre-allocation engine dynamically deploys standby machines based on the prediction results. If the predicted signal attenuation rate exceeds 30% and the remaining task time is less than 10 minutes, 2 standby machines are scheduled in advance to high-risk areas, and 20% of the bandwidth is reserved specifically for the first-level data transmission back. The module supports the adaptive configuration of task types. For example, in the disaster rescue task, the attenuation rate threshold is relaxed to 40% to ensure the task fault tolerance in extreme environments.

[0115] The data processing and feedback module consists of a data fusion engine, a decision-making generation unit, and an exception response unit. The data fusion engine integrates the information transmitted back by multiple sub-machines to generate a global task view. For example, it draws a heat map of cluster distribution through location data clustering analysis, and combines wind speed and heading to predict collision risks. The decision-making generation unit sends instructions to the control background based on the task importance grading results. For example, when a first-level data anomaly is detected, it immediately triggers path replanning or emergency shutdown. The exception response unit is built with a threshold warning mechanism. For example, when the position deviation exceeds 5 meters or the battery power is lower than 15%, a red alarm is activated, and safety instructions are broadcast to all sub-machines through an independent channel. The output results of the module are synchronized to the visual control interface, supporting real-time intervention and strategy backtracking by operators.

[0116] It can be known from the above that:

[0117] In the present invention, through dynamic priority division and intelligent resource scheduling, the execution reliability of critical tasks in complex environments is significantly improved. The system preferentially ensures the stable transmission of first-level data (such as location, emergency instructions) in weak signal scenarios, and avoids task failures caused by communication interruptions by quickly switching standby machines to relay nodes or adjusting the cluster topology. For example, in high-dynamic scenarios such as disaster relief, even when the signal strength of the sub-machines drops suddenly, the system can still ensure the real-time feedback of critical instructions, and at the same time automatically compresses or delays unnecessary data to reduce bandwidth occupancy, effectively improving the task success rate.

[0118] In the present invention, through predictive resource deployment and dynamic path optimization, the overall stability and adaptability of the communication network are strengthened. The intelligent relay scheduling module combines real-time environmental data and task requirements to deploy standby machines to high-risk areas in advance, reducing the risk of communication interruption caused by terrain occlusion or weather interference. At the same time, the module supports parallel execution of multiple solutions. For example, the activation of relay nodes is synchronized with the adjustment of the host position to ensure the redundancy and continuity of the critical data feedback link. This design enables the UAV cluster to cooperate efficiently in scenarios such as complex terrain exploration and large-scale inspection, reducing the frequency of manual intervention, and at the same time extending the effective operation time of critical equipment.

[0119] It should be noted that, in this document, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or device. Without further limitation, an element defined by the phrase "comprising an..." does not exclude the presence of additional identical elements in the process, method, article, or device comprising the said element.

[0120] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A UAV data transmission system based on information importance grading, characterized in that: The system includes: a signal strength monitoring and grading module, a data importance grading module, a dynamic adaptive transmission control module, an intelligent communication scheduling module, and a data processing and feedback module; The internal of the intelligent communication scheduling module is provided with a real-time network status monitoring sub-module, a relay path optimization decision sub-module, a node dynamic management sub-module, a cluster topology dynamic adjustment sub-module, and a predictive resource deployment sub-module; The output end of the signal strength monitoring and grading module is connected to the signal input interface of the dynamic adaptive transmission control module; The output end of the data importance grading module is connected to the data priority interface of the dynamic adaptive transmission control module; The transmission strategy instruction output end of the dynamic adaptive transmission control module is connected to the path optimization interface of the intelligent communication scheduling module; The network status feedback end of the intelligent communication scheduling module is connected to the real-time data interface of the data processing and feedback module; The global task view output end of the data processing and feedback module is bidirectionally connected to the threshold calibration interface of the signal strength monitoring and grading module and the priority update interface of the data importance grading module respectively.

2. The drone data transmission system based on information importance grading according to claim 1, characterized in that: The signal strength monitoring and grading module consists of a signal acquisition unit, a threshold determination unit, and a dynamic calibration unit; the signal acquisition unit obtains the communication signal strength with the host in real time through wireless sensors deployed on the slave machines and transmits the original signal strength to the threshold determination unit; the threshold determination unit classifies the signals according to the preset three-level thresholds of weak signals, medium signals, and strong signals, and the dynamic calibration unit combines the task scenario requirements to improve the adaptability of signal classification to the dynamic environment; the classification results are output to the network status heat map in real time to mark the positions of weak signal slave machines and the signal attenuation trend.

3. The drone data transmission system based on information importance classification according to claim 1, characterized in that: The data importance grading module is provided with a task classification engine, a user-defined interface, and a priority mapping unit; the task classification engine automatically divides the data level according to the task type, the user-defined interface allows operators to adjust the grading standard through the control background, and the priority mapping unit binds the data level to the transmission strategy.

4. The drone data transmission system based on information importance classification according to claim 1, characterized in that: The dynamic adaptive transmission control module includes: a signal-data matching unit, a bandwidth allocation unit, and a standby machine trigger logic; the signal-data matching unit selects the transmission content according to the real-time signal strength level; the bandwidth allocation unit adopts a probability feedback mechanism, calculates the task completion probability through a Bayesian network under weak signals, and if the probability is lower than 10%, it triggers the intervention of the standby machine, otherwise it compresses redundant data according to the priority; the standby machine trigger logic integrates a reinforcement learning model to analyze the position, power, and link quality of the standby machine, and preferentially calls the node closest in distance and with a power higher than 60% as a relay.

5. The drone data transmission system based on information importance grading according to claim 1, wherein: The real-time network status monitoring sub-module generates a dynamic network quality assessment report by collecting the communication status data of all slave machines in the UAV cluster in real time and combining the global information received by the host; this module uses multi-dimensional data fusion technology to associate discrete communication metrics with task requirements, and after the monitoring data is normalized, it outputs a network quality report including the real-time signal attenuation trend, the health status of slave machines, and the task interruption risk, providing a decision basis for subsequent relay scheduling and topology adjustment; The comprehensive communication score of the slave device is obtained by multiplying the normalized value of the real-time signal strength by the signal weight coefficient, adding the product of the task priority coefficient and the task weight, and then subtracting the product of the packet loss rate and the penalty factor. The calculation formula of the signal strength weighted scoring model is as follows: Q = S·w s + P·w p - L·α; Where S represents the normalized value of the real-time signal strength, with a range of [0,1][0,1], which is linearly mapped from the signal strength measurement value; w s Represents the signal weight coefficient, which is dynamically adjusted according to the task scenario; P represents the task priority coefficient, which is defined by the data classification module; w p Indicates the task weight, fixed at 0.3, to balance the impact of task priority on scoring; L represents the packet loss rate, with a value range of [0,100][0,100]; α represents the packet loss penalty factor, where α = 0.05 for every 1% packet loss rate, and the total penalty upper limit is Qmax·0.

3.

6. The unmanned aerial vehicle data transmission system based on information importance grading according to claim 1, wherein: The relay path optimization decision sub-module dynamically analyzes the network state and task requirements of the UAV cluster through the reinforcement learning algorithm, and calculates the optimal communication path in real time; this module takes the network quality report, task importance classification data, and the available status of the standby device as inputs, and defines the decision trigger mechanism according to the signal strength threshold and task priority: when it is detected that a single slave device is in a weak signal but undertakes a key task, the standby device is preferentially enabled to switch to a relay node to establish a redundant communication link; if multiple slave devices are simultaneously in a weak signal area, the cluster topology adjustment algorithm is called to drive the host to move towards the weak signal area to shorten the communication distance; during the decision-making process, the module quantifies the balance relationship between network stability, task completion rate, and resource consumption through the reward function in the reinforcement learning model, and at the same time combines the real-time environmental data to predict the future signal attenuation trend, autonomously selects the relay enabling or topology adjustment strategy, and dynamically optimizes the decision weight to ensure the transmission continuity of the first-level key data is preferentially guaranteed when communication resources are limited; The calculation formula of the dynamic path reward function is as follows: R = w s ·S + w p ·P - α·L; Where: S represents the normalized value of the real-time signal strength, which is linearly mapped and calculated from the RSSI values of the slave device and the host; w s Indicates the signal strength weight; P represents the task priority coefficient; w p Indicates the task importance weight; L represents the packet loss rate; α represents the packet loss penalty factor The calculation formula of the predictive resource allocation model is as follows: β represents the historical signal attenuation coefficient; γ represents the environmental interference factor, which integrates terrain complexity and weather data; τ represents the task time urgency factor; T represents the remaining task duration.

7. The drone data transmission system based on importance grading as claimed in claim 1, wherein: The node dynamic management sub-module consists of a status monitoring unit, a mode switching unit, and a link maintenance unit; the status monitoring unit real-time collects the battery power, communication load, and task priority data of the standby device, and identifies abnormal nodes through the threshold determination mechanism. The mode switching unit dynamically adjusts the role of the standby device according to the network state heat map and task urgency. In the initial state, the standby device executes regular tasks. When it is detected that the signal strength of the main link continuously drops below the preset threshold, it is immediately switched to the relay mode to construct a redundant link to preferentially transmit the first-level key data.

8. The drone data transmission system based on information importance classification according to claim 1, characterized in that: The cluster topology dynamic adjustment sub-module minimizes the total signal attenuation of the communication link by optimizing the host position and the cluster formation pattern; this module takes the real-time position of the slave machines, the host movement ability limit, and the environmental obstacle data as inputs, calculates the host movement direction based on the gradient descent algorithm, and the objective function is the weighted sum of the signal attenuation coefficient and the square of the distance; the gradient direction is generated by the partial derivative of the objective function with respect to the host coordinates, and a momentum acceleration mechanism is introduced: if the gradient directions are the same three times in a row, the host movement speed is increased by 20% to quickly approach the weak signal area; the obstacle avoidance logic detects obstacles through LiDAR or an environmental map, and a dynamic repulsive vector is superimposed on the gradient direction. When the distance to the obstacle is less than 5 meters, the repulsive weight is increased to 3 times to forcefully avoid the collision risk; After the adjustment is completed, the module outputs the host movement instruction and the cluster formation parameters to ensure the shortest backbone link while maintaining the global communication stability; The calculation formula of the momentum acceleration communication loss optimization model is: where v represents the actual movement speed of the host; m represents the momentum acceleration factor, with an initial value of 1, which increases to 1.2 if the gradient directions are the same three times in a row; v base represents the basic movement speed defined by the drone hardware; η represents the learning rate, which is fixed at 0.1 to control the movement step accuracy; Denotes the partial derivative of the communication loss function with respect to the host coordinates, reflecting the rate of change of signal attenuation with position; The calculation formula of the dynamic repulsive vector obstacle avoidance model is: where F represents the obstacle avoidance repulsive force, and the direction is away from the obstacle; d represents the real-time distance between the obstacle and the host; w terrain Indicates the terrain complexity weight, with a value of 1.5 for mountainous terrain and 1.0 for plains; k repel Indicates the dynamic repulsion coefficient, which is increased to 3 when the obstacle distance is less than 5 meters, otherwise it is 1.

9. The drone data transmission system based on information importance classification according to claim 1, characterized in that: The predictive resource deployment sub-module is provided with a historical data analysis unit, an environmental modeling unit, and a resource pre-allocation engine; the historical data analysis unit extracts the signal attenuation law from the communication logs, the environmental modeling unit fuses the real-time terrain elevation, meteorological data, and obstacle distribution to generate a multi-dimensional interference assessment map, and the resource pre-allocation engine dynamically deploys standby machines based on the prediction results.

10. A drone data transmission system based on information importance classification as described in claim 1, characterized in that: The data processing and feedback module consists of a data fusion engine, a decision generation unit, and an exception response unit; the data fusion engine integrates the information transmitted back by multiple slave machines to generate a global task view; The decision generation unit sends instructions to the control background based on the task importance grading results; The exception response unit has a built-in threshold warning mechanism.

Images