Unmanned aerial vehicle roaming scheduling method and device, electronic equipment and storage medium

Abstract

The application provides a method and device for unmanned aerial vehicle roaming scheduling, an electronic device and a storage medium. The method comprises: acquiring state features of a target unmanned aerial vehicle in real time; screening evaluation indexes required by a decision model based on the state features of the target unmanned aerial vehicle to obtain screened evaluation indexes; determining whether to adjust the hierarchical structure of the current decision model according to the screened evaluation indexes; adjusting the hierarchical structure of the current decision model based on the screened evaluation indexes to obtain an adjusted decision model in the case of determining that the hierarchical structure of the current decision model needs to be adjusted; determining the subjective weight of the screened evaluation indexes based on the adjusted decision model; determining the scores of each base station capable of communicating with the target unmanned aerial vehicle based on at least the subjective weight and the quantitative value of the screened evaluation indexes; and selecting a target base station for communicating with the target unmanned aerial vehicle based on the scores of the base stations. The application can reduce the frequency of base station switching, improve the stability of the link, and reduce the risk of reconnection.

Description

Technical Field

[0001] This application belongs to the field of unmanned aerial vehicle (UAV) technology, and more specifically relates to a UAV roaming scheduling method, device, electronic device and storage medium. Background Technology

[0002] Currently, drones and ground stations primarily rely on point-to-point communication, but this method suffers from limited communication range. Cellular-based drone roaming communication can address this range limitation. However, in cellular networks, drone roaming decisions often rely solely on a single metric such as received signal power or signal-to-noise ratio (SNR) to trigger handover. In complex environments, such as those caused by terrain and buildings leading to multipath interference, these interference signals can amplify at certain moments, forming so-called "false strong signals." When relying on a single RSSI metric for handover decisions, these interference signals may be mistakenly identified as strong signals. The drone receiver might then mistakenly identify these interference signals as the dominant signal, believing the current link quality is good, thus missing the opportunity to switch to a better base station. This can easily lead to link instability and frequent handovers. Especially during high-speed flight, this single-standard handover strategy cannot effectively cope with dynamic changes, resulting in higher handover latency and impacting communication quality.

[0003] Therefore, drone roaming communication based on cellular networks, which uses a single-index switching mechanism, suffers from poor link stability, frequent switching, resulting in high switching overhead and high reconnection risk. Summary of the Invention

[0004] One of the purposes of this application is to provide a method, apparatus, electronic device and storage medium for drone roaming scheduling, which at least solves the problem of frequent switching caused by using a single indicator switching mechanism, improves the stability of the link, and reduces switching overhead and reconnection risk.

[0005] A first aspect of this application provides a method for scheduling unmanned aerial vehicle (UAV) roaming, including:

[0006] Real-time acquisition of the target drone's status characteristics;

[0007] The evaluation indicators required for the selection decision model based on the state characteristics of the target UAV are obtained.

[0008] Determine whether to adjust the hierarchical structure of the current decision-making model based on the selected evaluation indicators;

[0009] If it is determined that the hierarchical structure of the current decision-making model needs to be adjusted, the hierarchical structure of the current decision-making model is adjusted based on the selected evaluation indicators to obtain the adjusted decision-making model.

[0010] The subjective weights of the evaluation indicators for screening are determined based on the adjusted decision-making model.

[0011] The scores of each base station capable of communicating with the target drone are determined based at least on the subjective weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators.

[0012] The target base station for communication with the target drone is selected based on the scores of each base station.

[0013] A second aspect of this application provides a drone roaming scheduling device, comprising:

[0014] The data acquisition module is used to acquire the status characteristics of the target UAV in real time.

[0015] The feature selection module is used to select the evaluation indicators that the decision model needs to use based on the state characteristics of the target UAV, and obtain the selected evaluation indicators.

[0016] The judgment module is used to determine whether to adjust the hierarchical structure of the current decision-making model based on the selected evaluation indicators;

[0017] The adjustment module is used to adjust the hierarchical structure of the current decision model based on selected evaluation indicators when it is determined that the hierarchical structure of the current decision model needs to be adjusted, so as to obtain the adjusted decision model.

[0018] The weight calculation module is used to determine the subjective weights of the selected evaluation indicators based on the adjusted decision model.

[0019] The scoring calculation module is used to determine the scores of each base station that can communicate with the target UAV based at least on the subjective weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators.

[0020] The base station selection module is used to select the target base station to communicate with the target drone based on the scores of each base station.

[0021] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the above-described drone roaming scheduling method.

[0022] In a fourth aspect of this application, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-described UAV roaming scheduling method.

[0023] The beneficial effects of the UAV roaming scheduling method, apparatus, electronic device, and storage medium provided in this application are as follows:

[0024] This application utilizes real-time acquired state characteristics of the target UAV to filter evaluation indicators for the decision-making model. Based on the filtered evaluation indicators, it determines whether to adjust the hierarchical structure of the current decision-making model. If adjustment is necessary, the hierarchical structure of the current decision-making model is adjusted according to the filtered evaluation indicators, resulting in an adjusted decision-making model. The adjusted decision-making model is then used to determine the subjective weights of the filtered evaluation indicators. At least based on the subjective weights and quantified values ​​of the filtered evaluation indicators, a score is determined for each base station capable of communicating with the target UAV. Based on this score, the target base station for communication with the target UAV is selected. Through this method, this application enables the decision-making model to flexibly adapt to changing flight scenarios, allowing it to quickly respond to real-time changes and improving the accuracy of base station selection and communication efficiency.

[0025] In addition, this application uses multiple evaluation indicators of the target UAV for dynamic screening to select the target base station, avoiding the problem of frequent base station switching caused by a single indicator switching mechanism, thereby improving the stability of the link and reducing the switching overhead and reconnection risk. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 A schematic diagram of the system architecture used in the drone roaming scheduling method provided in an embodiment of this application;

[0028] Figure 2 A flowchart illustrating a drone roaming scheduling method provided in an embodiment of this application;

[0029] Figure 3 This is a structural block diagram of a drone roaming scheduling device provided in an embodiment of this application;

[0030] Figure 4 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0031] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0032] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0033] Figure 1 A schematic diagram of the system architecture used in an embodiment of the UAV roaming scheduling method provided in this application. (Reference) Figure 1 In this embodiment, multiple drones and multiple base stations are provided. As shown in the figure, the multiple drones and multiple base stations are in the same mesh ad hoc network. Each drone can establish a communication connection with any one of the multiple base stations, or with any one of the multiple drones. One base station can connect to multiple drones simultaneously. In this mesh ad hoc network, all nodes are interconnected. Each base station and each drone can broadcast its own status characteristics in real time, and each drone can wirelessly receive status characteristics broadcast by each base station and other drones. Each drone can act as a dispatch station to select the optimal base station. This decentralized decision-making based on the mesh ad hoc network in this embodiment can also maintain communication performance as the number of drones increases, avoiding central bottlenecks. Maintaining communication performance means that various indicators measuring communication performance reach set thresholds, including but not limited to: packet loss rate ≤ 2% (communication reliability performance), target base station handover success rate ≥ 99.5% (handover performance), etc.

[0034] Please refer to Figure 2 , Figure 2 The flowchart is a method for scheduling unmanned aerial vehicles (UAVs) roaming according to an embodiment of this application. The method may include steps S101 to S106.

[0035] S101: Real-time acquisition of the target UAV's status characteristics.

[0036] In this embodiment, the target drone can be any drone in the mesh ad hoc network. The target drone's state characteristics include: service requirement characteristics, flight environment characteristics, and network state characteristics. There are typically multiple state characteristics for the drone. Among them, service requirement characteristics include at least one of the following: service stage, service type, service priority, and service scenario requirements. Service stage can include characteristics related to the stage of the drone's flight and service execution process, such as takeoff, mission execution, and landing. Service type includes video stream data, image data, and sensor data. Service priority indicates the priority of the service currently being executed by the target drone. Service scenario requirements include characteristics related to the scenario's requirements, such as low latency and high throughput. For example, the target drone's service requirement characteristics could be that, during mission execution, it needs to transmit data collected by sensors to ensure low latency. Alternatively, it could be that, during mission execution, the target drone needs to transmit collected video data to ensure the network maintains high throughput during transmission.

[0037] Flight environment characteristics refer to the environmental characteristics of the target UAV during flight, such as the terrain (mountainous or urban high-rise area) where the target UAV is located, the weather characteristics (wind speed, humidity and temperature) and GPS status data.

[0038] Network state characteristics include the network state characteristics between the target UAV and multiple base stations in the same mesh ad hoc network that can communicate with the target UAV, such as signal quality, link latency, and interference level. Furthermore, parameters characterizing signal quality may include at least one of the following: Received Signal Strength Indicator (RSSI), Signal-to-Noise Ratio (SNR), Link Load Rate, Packet Loss Rate, Bit Error Rate, and Number of Communication Interruptions; parameters characterizing link latency may include: Round-Trip Time (RTT); parameters characterizing interference level may include: Interference Level, i.e., Interference Detection Value.

[0039] In one implementation, a state feature set can be established, containing all state feature-related parameters that the target UAV needs to acquire. The target UAV acquires all state feature-related parameters in the state feature set in real time. It is understood that the parameters included in this state feature set can be added, deleted, or modified as needed.

[0040] In this embodiment, the status characteristics of the target UAV can be acquired in real time. These status characteristics include multiple features related to the UAV's service requirements, flight environment, and network status, which is beneficial for providing rich reference indicators for the roaming scheduling of the target UAV and improving the accuracy of the UAV in selecting target base stations.

[0041] S102: Evaluation indicators to be used in the decision-making model based on the state characteristics of the target UAV, and obtain the selected evaluation indicators.

[0042] In one embodiment, the state features acquired by the target UAV include all features in the state feature set. When determining the scores for each base station, it is not necessary to use all features; that is, the decision model does not need to use all state features to calculate weights. Therefore, this embodiment can select the state features with the greatest impact on decision-making as evaluation indicators based on the real-time acquired state features. The number of selected evaluation indicators can be at least one or more, and the specific number of selected evaluation indicators in this application can be appropriately adjusted according to actual needs.

[0043] In one embodiment, the evaluation metrics used in the target UAV state characteristic-based screening decision model may include any of the following:

[0044] 1) Utilize a pre-trained feature selection model to select the evaluation metrics required for the decision-making model based on the state characteristics of the target UAV;

[0045] In this embodiment, the state characteristics of the target UAV are input into a pre-trained feature selection model, and the feature selection model is used to select the state characteristics that have the greatest impact on decision-making as the evaluation indicators to be used by the decision-making model.

[0046] 2) Using preset rules, select the evaluation indicators to be used in the decision-making model based on the state characteristics of the target UAV.

[0047] In this embodiment, the preset rules can be custom rules. For example, a weather indicator can be added under specified weather conditions (such as wind reaching a specified level); a task urgency indicator can be added when the task urgency increases; a load balancing indicator can be added when network congestion occurs; an interference level indicator can be added when interference levels increase; and a "terrain occlusion impact" evaluation indicator can be added when the target UAV enters a terrain-complex area (such as mountainous areas or urban high-rise areas). Including the interference level indicator in the evaluation indicators used by the decision model can prevent the selection of base stations corresponding to high-interference links as target base stations. It is understood that the corresponding indicators can be deleted in the opposite situations described in the above examples.

[0048] This embodiment can utilize a pre-trained feature selection model or preset rules to select the evaluation indicators to be used by the decision model based on the state characteristics of the target UAV. This process does not require human intervention and can automatically select the optimal combination of evaluation indicators for the decision model. This enables the decision model to calculate a more accurate score for each base station based on the optimal combination of evaluation indicators, thereby improving the accuracy of target base station selection and reducing handover overhead and reconnection risk.

[0049] S103: Determine whether to adjust the hierarchical structure of the current decision-making model based on the selected evaluation indicators.

[0050] In one embodiment, the method for determining whether to adjust the hierarchical structure of the current decision-making model based on the selected evaluation indicators includes:

[0051] Determine whether the difference between the evaluation indicators used in the current decision-making model and the selected evaluation indicators is greater than a specified difference threshold. If it is greater than the specified difference threshold, then it is determined that the hierarchical structure of the current decision-making model needs to be adjusted.

[0052] Understandably, the difference threshold can be set as needed. For example, the difference threshold can be set to 0, meaning that if the evaluation index used by the current decision model is different from the selected evaluation index, then the hierarchical structure of the current decision model needs to be adjusted. Alternatively, the difference threshold can be set to any value greater than 0. By controlling the size of this difference threshold, the frequency of adjustment of the hierarchical structure of the decision model can be controlled.

[0053] S104: If it is determined that the hierarchical structure of the current decision-making model needs to be adjusted, the hierarchical structure of the current decision-making model is adjusted based on the selected evaluation indicators to obtain the adjusted decision-making model.

[0054] This embodiment overcomes the problem that decision models using static hierarchical structures are difficult to adapt to changes in the dynamic environment, business requirements, network status, etc., during UAV flight. In this application, by automatically adjusting the hierarchical structure of the decision model, adaptive changes in the decision model can be achieved.

[0055] In this embodiment, the hierarchical structure of the decision-making model is adjusted based on the selected evaluation indicators, which can prevent decision-making errors caused by sudden data changes. This method of adjusting the hierarchical structure of the decision-making model based on real-time changing evaluation indicators enables the decision-making model to respond quickly to real-time changes, improves the accuracy of base station selection and communication efficiency, is suitable for dynamic scenarios such as UAV roaming scheduling, and can also better ensure communication quality and resource utilization efficiency.

[0056] S105: Determine the subjective weights of the selected evaluation indicators based on the adjusted decision-making model.

[0057] The adjusted decision model is a decision model with a determined hierarchical structure. In one implementation, the decision model is an AHP decision model. The method for determining the subjective weights of the selected evaluation indicators based on the decision model with a determined hierarchical structure, i.e., the method for determining the subjective weights of evaluation indicators using the AHP decision model, includes the following steps: First, constructing the AHP hierarchical structure; then, constructing the judgment matrix of the evaluation indicators; and finally, implementing hierarchical single ranking and consistency verification. Therefore, step S105 involves constructing the judgment matrix of the selected evaluation indicators after determining the hierarchical structure of the AHP decision model, and further performing hierarchical single ranking and consistency verification operations to determine the subjective weights of the selected evaluation indicators. It can be understood that if the AHP hierarchical structure includes a decision layer, it can further include the steps of hierarchical overall ranking and consistency verification, finally obtaining the subjective weights of the selected evaluation indicators.

[0058] S106: Determine the scores of each base station that can communicate with the target UAV based at least on the subjective weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators.

[0059] In this embodiment, the score is used to characterize the comprehensive performance index of each base station, i.e., whether it is suitable for establishing a communication connection with the target drone. Each base station includes both the base station currently communicating with the target drone and multiple base stations in the same mesh ad hoc network that have not yet established communication with the target drone. The higher the score, the more suitable the corresponding base station. The scores of each base station capable of communicating with the target drone are determined based on the adjusted decision model, making the selection of the optimal base station more informed.

[0060] In one embodiment, determining the scores of each base station capable of communicating with the target drone, based at least on the subjective weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators, includes:

[0061] Based on the subjective weights and quantitative values ​​of the selected evaluation indicators, a weighted calculation is used to determine the scores of each base station that can communicate with the target drone.

[0062] In one embodiment, when the decision model is an AHP decision model, fuzzy numbers (such as triangular fuzzy numbers) can be added to the AHP scoring process to accommodate the uncertainty in expert scoring and improve the robustness of the weights. The AHP scoring process is as follows: based on the AHP decision model, the subjective weights of the selected evaluation indicators are determined; then, based on the subjective weights and quantified values ​​of the selected evaluation indicators, a weighted calculation is performed to determine the score of each base station capable of communicating with the target UAV. Here, the subjective weights of the selected evaluation indicators refer to the normalized subjective weights of the selected evaluation indicators. The quantified values ​​of the selected evaluation indicators can be obtained by normalizing the original values ​​of the selected evaluation indicators.

[0063] The scores for each base station are calculated using the subjective weights and quantitative values ​​of the selected evaluation indicators. The formula is as follows:

[0064] , i=1,2,…,m; j=1,2,…,n.

[0065] In the above formula, For the score of the j-th base station, Let be the quantified value of the ith evaluation indicator selected for the j-th base station. Let m be the normalized subjective weight of the i-th selected evaluation indicator, m represent the number of selected evaluation indicators, i represent the i-th selected evaluation indicator, n represent the number of base stations, and j represent the j-th base station.

[0066] S107: Select the target base station to communicate with the target drone based on the scores of each base station.

[0067] In one embodiment, if the score of the base station currently communicating with the target drone meets the requirements, the current base station can be used as the target base station without base station handover. If the score of the base station currently communicating with the target drone does not meet the requirements, a target base station can be selected based on the scores of each base station. The target drone continuously updates the scores of each base station during flight and determines whether to trigger a handover operation based on the scores. The handover triggering mechanism of the target base station in this embodiment differs from the traditional mechanism that relies solely on a single RSSI trigger. This embodiment comprehensively considers multiple factors such as the target drone's service requirements, flight environment characteristics, and network status characteristics, which can avoid unnecessary handovers, reduce the number of handovers, and improve communication continuity.

[0068] As can be concluded from the above, firstly, this embodiment of the application judges the communication signal quality based on multiple state characteristics of the target UAV acquired in real time, such as service demand characteristics, flight environment characteristics, and network state characteristics. This can largely avoid misjudgments caused by a single evaluation index, reduce interruptions, reduce the frequency of base station handover, and thus enable the target UAV to select the optimal communication link, ensuring the stability of the communication link. Secondly, this embodiment of the application can adjust the hierarchical structure of the decision model based on the current state characteristics of the target UAV, making it adaptable to the current operating environment, thereby achieving more accurate base station selection and reducing the risk of communication quality degradation caused by selecting suboptimal connections. Finally, this embodiment of the application can comprehensively consider multiple state characteristics of the target UAV acquired in real time, effectively responding to dynamic changes, thereby reducing handover latency, reducing handover overhead, and improving communication quality.

[0069] In one embodiment, the decision model is the AHP decision model. The AHP decision model decomposes complex problems into a multi-level structure, such as the goal layer, criterion layer, and indicator layer, with different evaluation indicators at each level. The AHP decision model constructs decision tree nodes according to the hierarchy. For example, the root node can be an evaluation indicator in the goal layer, the higher-level nodes can be evaluation indicators in the criterion layer, and the lower-level nodes can be evaluation indicators in the indicator layer. Each node has a corresponding weight.

[0070] In one embodiment of this application, step 104, adjusting the hierarchical structure of the current decision model based on the selected evaluation indicators, includes at least one of the following:

[0071] Add or remove hidden layers from the current decision model;

[0072] Modify the inter-layer connection method of the current decision model;

[0073] Switch the current decision model's network architecture to a preset alternative network architecture.

[0074] Adding or deleting hidden layers in the current decision-making model is to adapt the hierarchical structure of the decision-making model to the state characteristics of the target UAV, thereby optimizing the selection of the target base station. In one embodiment of this application, the basis for triggering the addition or deletion of hidden layers in the current decision-making model includes at least one of the following: network performance, overfitting risk, and resource constraints. The values ​​of the above items can be determined based on selected evaluation indicators (i.e., determining the quantitative values ​​of network performance, overfitting risk, resource constraints, and differences in evaluation indicators). For example, if the current network performance (based on the output of the decision-making model) cannot meet the communication needs of the UAV (latency exceeds a set latency threshold, signal quality does not meet set requirements, or throughput is lower than a set throughput threshold), then adding hidden layers to the current decision-making model can be triggered to improve the expressive power of the decision-making model; if the current decision-making model is complex, causing computational resource consumption to exceed a set resource consumption threshold or response time to exceed a set duration threshold, then deleting hidden layers from the current decision-making model can be triggered to simplify the decision-making model and reduce the computational burden.

[0075] In one embodiment of this application, the connection methods between the layers of the decision model include residual connections and attention connections, and may also include fully connected connections, convolutional connections, recurrent connections, and sparse connections. In another embodiment of this application, the basis for triggering modification of the inter-layer connection methods of the current decision model includes at least one of the following: task complexity and environmental changes. The values ​​of the above items (i.e., the quantitative values ​​of task complexity and environmental changes) can be determined based on selected evaluation metrics. For example, when the quantitative values ​​of task complexity or environmental changes change, the connection between the layers of the decision model can be adjusted from fully connected connections to residual connections.

[0076] In one embodiment of this application, the network architecture of the decision model includes a Transformer model, which is suitable for dynamic, multimodal data processing. The network architecture may also include a Convolutional Neural Network (CNN) and a Recurrent Neural Network (RNN). In one embodiment of this application, the basis for triggering the switching of the network architecture of the decision model includes at least one of the following: environmental changes, task changes, and performance bottlenecks.

[0077] In this embodiment, by adjusting the hierarchical structure of the current decision-making model in any of the above ways, the decision-making model can flexibly adapt to changing flight scenarios and avoid the limitations brought about by static structures.

[0078] In one embodiment of this application, step 104, adjusting the hierarchical structure of the current decision model based on the selected evaluation indicators, includes: selecting the required hierarchical structure from a set of preset hierarchical structures using a structure search algorithm based on the selected evaluation indicators. That is, multiple hierarchical structures of the decision model are pre-set, and the structure search algorithm directly selects the most suitable hierarchical structure from the multiple hierarchical structures based on the selected evaluation indicators, which serves as the adjusted hierarchical structure of the current decision model. This eliminates the need to perform operations such as adding or deleting hidden layers of the current decision model, modifying the inter-layer connections of the current decision model, or switching the network architecture of the current decision model to other preset network architectures, as described above. This allows for a faster determination of the hierarchical structure of the current decision model.

[0079] In one embodiment of this application, to ensure the smoothness and stability of the subjective weight adjustment of the selected evaluation indicators, a sliding window technique (smoothing weight changes based on historical data) and a feedback optimization mechanism (correcting weights according to decision results) can be introduced to prevent decision-making errors caused by sudden environmental changes.

[0080] To facilitate understanding of this application, a specific example is provided below to illustrate the drone roaming scheduling method of this application. Assume there are h (h greater than or equal to 1) drones and n (n greater than 1) base stations in the same mesh ad hoc network. Drone number 1 is selected as the target drone, and the base station currently connected to this target drone is base station number 9. The evaluation metrics used when selecting base station number 9 as the target base station include: service stage is task execution, service type is image data, service scenario requirement is low latency, received signal strength (RSSI) is -50dBm, link load rate is 20%, and packet loss rate is 1%. For ease of description, this set of evaluation metrics is referred to as the first set of evaluation metrics. The roaming scheduling method provided for this target drone includes:

[0081] The state characteristics of the target UAV are acquired in real time, and the acquired state characteristics include all features in the state characteristic set.

[0082] The acquired state features are input into a pre-trained feature selection model to obtain the evaluation metrics required by the decision model. In other words, the evaluation metrics are selected from the acquired state features to be used by the decision model in this instance. Assume the selected evaluation metrics include: the business stage is task execution, the business type is image data, the business scenario requirements are low latency, the received signal strength RSSI is -60dBm (signal strength weakened), the link load rate is 40% (load increased), and the packet loss rate is 2% (packet loss rate increased). For ease of description, this set of evaluation metrics is referred to as the second set of evaluation metrics.

[0083] Calculate the difference between the second set of evaluation indicators and the first set of evaluation indicators, and compare the difference with the specified difference threshold. Assuming the calculated difference is 0.25 and the specified difference threshold is 0.15, it can be determined that the difference is greater than the specified difference threshold. At this time, the hierarchical structure of the current decision model needs to be adjusted.

[0084] Based on the second set of evaluation indicators, a structure search algorithm is used to select the required hierarchical structure from a range of preset hierarchical structures. The decision model of this hierarchical structure is then used to determine the subjective weights of the second set of evaluation indicators. Based on the quantified values ​​of the second set of evaluation indicators and their determined subjective weights, scores are calculated for N base stations. The base station with the highest score is selected as the target base station. For example, if the base station with the highest score is number 3, the target drone will switch its target base station from base station number 9 to base station number 3.

[0085] In one embodiment of this application, selecting a target base station to communicate with the target drone based on the scores of each base station includes:

[0086] Determine whether the score of the current base station communicating with the target drone meets the specified requirements;

[0087] If the current base station's score does not meet the specified requirements, the base station with the highest score will be selected as the target base station for communication with the target drone.

[0088] In this embodiment, if the score of the current base station communicating with the target drone meets the specified requirements—for example, if the score of the current base station communicating with the target drone is greater than a set score threshold—it indicates that the current base station can meet the normal communication requirements of the target drone, and there is no need to select another base station. If the score of the current base station communicating with the target drone is less than or equal to the set score threshold, then the optimal base station (i.e., the target base station) needs to be selected for handover. The selection of the optimal base station is based on the scores of each base station, and the base station with the highest score is the optimal base station.

[0089] Understandably, the score threshold can be set as needed, and the frequency of base station handover can be controlled by adjusting the value of the score threshold.

[0090] In this embodiment, if the score of the current base station communicating with the target drone meets the specified requirements, even if there are other base stations with higher scores, the base station handover operation can be avoided. This can effectively avoid frequent base station handover, reduce base station handover overhead and reconnection risk, and ensure that the current base station can meet the service requirements.

[0091] In this embodiment, if the score of the current base station communicating with the target drone does not meet the specified requirements, the base station with the highest score can be selected as the target base station for communication with the target drone. This is because the decision model uses multiple evaluation indicators, and the score calculated based on these indicators is more realistic and accurate.

[0092] In another embodiment of this application, selecting the target base station for communication with the target drone based on the scores of each base station includes selecting the base station with the highest score as the target base station for communication with the target drone. This embodiment can ensure that the target drone always establishes communication with the optimal base station, which can improve the communication quality of the drone, improve the reliability of the air-to-ground link and the efficiency of service data transmission, thereby meeting the needs of high-capacity services.

[0093] In another embodiment of this application, after selecting the evaluation indicators to be used in the decision model based on the state characteristics of the target UAV and obtaining the selected evaluation indicators, the UAV roaming scheduling method further includes:

[0094] The objective weights of the selected evaluation indicators are calculated using the entropy weight method.

[0095] The subjective weights and objective weights are combined to obtain the combined weights.

[0096] In this embodiment, step S106, determining the scores of each base station capable of communicating with the target UAV based at least on the subjective weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators, includes:

[0097] The scores of each base station capable of communicating with the target UAV are determined based on the fusion weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators.

[0098] In this context, the subjective weight of the selected evaluation indicators refers to the normalized subjective weight, while the objective weight refers to the normalized objective weight. The quantitative value of the selected evaluation indicators can be obtained by normalizing the original values ​​of the selected evaluation indicators.

[0099] The fusion coefficient, which combines subjective and objective weights, can be used as follows: The fusion weights are calculated using the following formula. .

[0100] , i=1,2,…,m.

[0101] In the above formula, It is the fusion weight of the evaluation index of the i-th selection. It is the normalized subjective weight of the evaluation index selected for the i-th screening. is the normalized objective weight of the i-th selected evaluation indicator, m represents the number of selected evaluation indicators, and i represents the i-th selected evaluation indicator.

[0102] The scores of each base station are calculated using the fusion weights and the quantified values ​​of the selected evaluation indicators. , The base station with the largest value is the target base station, i.e., the optimal base station.

[0103] , i=1,2,…,m; j=1,2,…,n.

[0104] In the above formula, Let this be the score for the j-th base station; this score is a comprehensive score. Let be the quantified value of the ith evaluation indicator selected for the j-th base station. Let m be the fusion weight of the i-th selected evaluation indicator, m represent the number of selected evaluation indicators, i represent the i-th selected evaluation indicator, n represent the number of base stations, and j represent the j-th base station.

[0105] This embodiment combines subjective and objective weight values ​​to obtain a fused weight, and then calculates the score of each base station based on the fused weight. The target base station for communication with the target UAV is selected based on the score. This effectively combines subjective and objective weights, which can improve the accuracy of the score calculation of each base station, thereby improving the accuracy of target base station selection, further ensuring link stability, and reducing handover overhead and reconnection risk.

[0106] In one embodiment, the subjective weight and the objective weight are fused to obtain the fused weight, which includes:

[0107] By dynamically adjusting the fusion coefficient of subjective and objective weights using historical decision feedback and sliding window data, the adjusted fusion coefficient is obtained. Based on the adjusted fusion coefficient, the subjective and objective weights are then fused to obtain the fused weight.

[0108] This embodiment utilizes historical decision feedback to achieve a performance feedback closed loop, thereby dynamically optimizing the fusion coefficient. This allows for a reduction in the proportion of subjective weights when objective performance fluctuates significantly, and vice versa.

[0109] In one embodiment, when integrating subjective and objective weights, TOPSIS-GRA (a comprehensive evaluation method that combines the ranking method for approximating ideal solutions and grey relational analysis) can be introduced to enhance the ability to identify extreme solutions.

[0110] As described above, in one embodiment of this application, determining the scores of each base station capable of communicating with the target drone includes two methods: Method 1 determines the scores of each base station capable of communicating with the target drone based on the subjective weights and quantitative values ​​of the selected evaluation indicators. This method involves low computational cost, is flexible and easily adjustable, and allows for rapid decision-making to select the target base station. Method 2 determines the scores of each base station capable of communicating with the target drone based on the fusion weights and quantitative values ​​of the selected evaluation indicators. This method reduces subjective bias and uncertainty, enhances the robustness and credibility of the scores for each base station, and ensures that the selection of the target base station can balance fairness and performance.

[0111] Users can choose one of these two methods actively as needed, or the flight control system can intelligently select one of them.

[0112] In one embodiment of this application, intelligent selection of two methods can be achieved by setting intelligent selection rules. For example, the factors considered in setting intelligent selection include at least one of the following: historical data of the base station and the service's demand for network status. The corresponding rule can be: if historical data of each base station is available, and the current UAV's service demand for network status is higher than a set threshold, method two can be selected to calculate the base station's score. The service's demand for network status can be determined by the quantified values ​​of multiple evaluation indicators included in the network status characteristics.

[0113] In one embodiment of this application, the drone roaming scheduling method further includes: if there are multiple base stations with the highest score (or comprehensive score), then selecting the target base station to communicate with the target drone using any of the following methods:

[0114] a) Based on the value of an evaluation indicator that is not involved in the scoring, select the target base station that communicates with the target UAV from the base stations with the highest scores; the evaluation indicator that is not involved in the scoring is an indicator included in the state characteristics but not involved in the scoring (or comprehensive scoring) calculation; that is, select the base station with the better value of the evaluation indicator that is not involved in the scoring as the target base station. For example, if the network outage count indicator is not involved in the scoring calculation, select the base station with the fewest network outages from the base stations with the highest scores as the target base station.

[0115] (b) Based on the historical data of the highest-rated base station, select the target base station for communication with the target drone from among the highest-rated base stations. Historical data includes the base station's call drop rate, average connection duration, or average throughput. When there is more than one base station with the highest rating, the dispatch station selects a base station with a more stable past performance as the target base station. For example, the base station with the lowest historical call drop rate or the longest average connection duration can be selected as the target base station for communication with the target drone.

[0116] Understandably, the selection can also be based on user preferences, or on preset rules (in ascending order of number) to select the target base station, or randomly select the target base station from the base stations with the highest scores.

[0117] As can be seen from the above, this embodiment takes into account the occurrence of two or more base stations with the highest scores, and can introduce multiple evaluation methods to select the optimal base station, which can maximize the improvement of user experience.

[0118] Current communication between drones and base stations uses static spectrum allocation and fixed configuration, meaning each user exclusively occupies a specific frequency band, which is not dynamically adjusted according to the flight environment. Static spectrum refers to a fixed frequency point; for example, if a drone is currently operating at 1360MHz, it cannot avoid interference from other drones or base station equipment operating on the same frequency. Under this architecture, the drone communication system has poor adaptability to multipath interference and environmental changes, and its resource utilization efficiency is also low. For example, when communication demand suddenly increases during flight, the statically allocated spectrum cannot be reused in time, resulting in wasted frequency bands. Simultaneously, multiple drones sharing a frequency band are prone to mutual interference, reducing system throughput.

[0119] In one embodiment of this application, after selecting a target base station to communicate with the target drone, the resources of other base stations in that frequency band can be simultaneously disabled to reduce interference. Alternatively, interference suppression can be achieved without disabling the resources of other base stations in that frequency band. "Disabling the resources of other base stations in that frequency band" means that after a base station establishes a connection with the target drone, other nearby base stations are prevented from continuing to transmit data or broadcast on the same frequency band, thereby reducing interference and lowering the transmission power to suppress co-channel interference and improve channel quality.

[0120] In one embodiment of this application, after selecting a target base station to communicate with the target UAV, a method for achieving interference suppression without completely shutting down the resources of other base stations includes at least one of the following methods, which can be used to suppress communication interference between the target UAV and the target base station:

[0121] First, link-layer interference suppression

[0122] The method involves acquiring periodically measured power spectral density (PSD) of adjacent and co-channel interference, calculating the interference-to-noise ratio (INR) based on the PSD, and adjusting the beam direction to point at the target UAV according to the INR. Essentially, the INR serves as a reference for adaptive adjustment of the UAV array antenna ports, allowing for real-time beam direction adjustment to minimize interference (this can be implemented for both air-to-ground and air-to-air links). This method enables the target base station to "point" its signal to the UAV it is communicating with, reducing signal leakage to other UAVs and improving spectrum reuse efficiency.

[0123] Secondly, MAC layer interference suppression

[0124] Dynamically adjusting the priority of the target UAV and other members in the same mesh ad hoc network within the time slot based on the interference level achieves MAC layer interference suppression through adaptive time slot allocation. This dynamic adjustment of the priority of members in the same mesh ad hoc network as the target UAV within the time slot can reduce concurrent transmission in high-interference environments.

[0125] In addition, based on the interference level and in conjunction with collision detection (Carrier Sense) and failure retransmission calculation, the transmit power, packet buffering, and / or retransmission parameters of the target drone and other members in the same mesh ad hoc network as the target drone can be dynamically adjusted to reduce the packet loss rate caused by interference. Interference level indicators include node mobility fluctuation, network latency jitter, signal strength, and packet loss rate. Collision detection and failure retransmission calculation involves identifying each drone or device in the same mesh ad hoc network as the target drone and statistically analyzing its collisions and failures during time slot contention or communication attempts.

[0126] Third, cross-layer cooperative interference control

[0127] A cross-layer interference suppression module is designed to input interference parameters (such as interference level) into the decision model during network selection, so as to avoid selecting the base station corresponding to the high interference link as the target base station.

[0128] The embodiments of this application can not only select the optimal base station based on the real-time acquired state characteristics of the target UAV, but also add an interference suppression method after selecting the optimal base station, thereby achieving a dual improvement in communication quality and resource utilization.

[0129] Corresponding to the drone roaming scheduling method in the above embodiment, Figure 3 This is a structural block diagram of a drone roaming scheduling device according to an embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 3The drone roaming scheduling device 20 includes: a data acquisition module 21, a feature filtering module 22, a judgment module 23, an adjustment module 24, a weight calculation module 25, a score calculation module 26, and a base station selection module 27.

[0130] Among them, the data acquisition module 21 is used to acquire the status characteristics of the target UAV in real time. The status characteristics include: business requirement characteristics, flight environment characteristics and network status characteristics.

[0131] The feature selection module 22 is used to select the evaluation indicators to be used in the decision model based on the state characteristics of the target UAV, and obtain the selection evaluation indicators.

[0132] The judgment module 23 is used to determine whether to adjust the hierarchical structure of the current decision-making model based on the selected evaluation indicators.

[0133] The adjustment module 24 is used to adjust the hierarchical structure of the current decision model based on the selected evaluation indicators when it is determined that the hierarchical structure of the current decision model needs to be adjusted, so as to obtain the adjusted decision model.

[0134] The weight calculation module 25 is used to determine the subjective weights of the selected evaluation indicators based on the adjusted decision model.

[0135] The scoring calculation module 26 is used to determine the scores of each base station that can communicate with the target UAV based at least on the subjective weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators.

[0136] The base station selection module 27 is used to select the target base station for communication with the target drone based on the scores of each base station.

[0137] In one embodiment of this application, when the feature selection module 22 is used to select the evaluation indicators to be used in the decision model based on the state characteristics of the target UAV, it is specifically used for:

[0138] Using a pre-trained feature selection model, the evaluation metrics required for the decision-making model based on the state characteristics of the target UAV are selected; or

[0139] Using preset rules, the evaluation indicators to be used in the decision-making model are selected based on the state characteristics of the target UAV.

[0140] In one embodiment of this application, when the judgment module 23 is used to determine whether to adjust the hierarchical structure of the current decision model based on the selected evaluation indicators, it is specifically used for:

[0141] Determine whether the difference between the current evaluation index used in the current decision-making model and the selected evaluation index is greater than a specified difference threshold. If it is greater than the specified difference threshold, then it is determined that the hierarchical structure of the current decision-making model needs to be adjusted.

[0142] In one embodiment of this application, when adjusting the hierarchical structure of the current decision model based on the selected evaluation indicators, the adjustment module 24 is specifically used for:

[0143] Add or remove hidden layers from the current decision model;

[0144] Modify the inter-layer connection method of the current decision model;

[0145] Switch the current decision model's network architecture to a preset alternative network architecture.

[0146] In one embodiment of this application, when adjusting the hierarchical structure of the current decision model based on the selected evaluation indicators, the adjustment module 24 is specifically used for:

[0147] Based on the evaluation indicators used in the screening, a structure search algorithm is used to select an adjusted hierarchical structure from a variety of preset hierarchical structures.

[0148] In one embodiment of this application, the weight calculation module 25 is specifically used for:

[0149] The objective weights of the selected evaluation indicators are calculated using the entropy weight method.

[0150] The subjective weights and objective weights are combined to obtain the combined weights.

[0151] In this embodiment, the scoring calculation module 26, when determining the scores of each base station capable of communicating with the target UAV based at least on the subjective weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators, specifically performs the following:

[0152] The scores of each base station capable of communicating with the target UAV are determined based on the fusion weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators.

[0153] In one embodiment of this application, the weight calculation module 25 is specifically used for:

[0154] The fusion coefficient is obtained by dynamically adjusting the subjective and objective weights using historical decision feedback and sliding window data.

[0155] The subjective weights and objective weights are merged based on the adjusted fusion coefficients to obtain the fusion weights.

[0156] In one embodiment of this application, when the base station selection module 27 is used to select a target base station for communication with the target drone based on the scores of each base station, it is specifically used for:

[0157] Determine whether the score of the current base station communicating with the target drone meets the specified requirements;

[0158] If the current base station's score does not meet the specified requirements, the base station with the highest score will be selected as the target base station for communication with the target drone.

[0159] In one embodiment of this application, if there are multiple base stations with the highest scores, the base station selection module 27 is further configured to:

[0160] Based on the value of an evaluation metric that was not included in the scoring, the target base station for communication with the target UAV is selected from the base stations with the highest scores; the evaluation metric that was not included in the scoring is an metric included in the state characteristics but not involved in the scoring calculation; or

[0161] Based on the historical data of the highest-rated base station, a target base station for communication with the target drone is selected from the base stations with the highest ratings.

[0162] In one embodiment of this application, the drone roaming scheduling device 20 further includes an interference suppression module (not shown in the figure);

[0163] The interference suppression module is used to acquire periodically measured power spectral density (PSD) of adjacent and co-channel interference, calculate the interference-to-noise ratio (INR) based on the PSD, and adjust the beam direction to point at the target UAV according to the INR; or,

[0164] The priority of the target UAV and other members in the same mesh ad hoc network as the target UAV in the time slot is dynamically adjusted based on the interference level; or...

[0165] Based on the interference level and in conjunction with collision detection and failure retransmission calculations, the transmit power, packet buffering, and / or retransmission parameters of the target UAV and each member in the same mesh ad hoc network as the target UAV are dynamically adjusted.

[0166] In summary, the UAV roaming scheduling method and apparatus of this application have at least the following beneficial effects:

[0167] (1) Improved decision-making accuracy: The hierarchical structure of the decision-making model is dynamically adjusted based on the real-time acquired state characteristics of the target UAV, so that the decision-making model matches the current operating environment, thereby achieving more accurate target base station selection and reducing the risk of communication quality degradation caused by suboptimal connection.

[0168] (2) Enhanced adaptability: By determining the score of each base station through the real-time acquisition of the target UAV's status characteristics, the roaming scheduling method and device can quickly respond to environmental interference or changes in task priority, reduce communication interruptions, and maintain the best communication link.

[0169] (3) Reduction of handover frequency: Comprehensively evaluate multiple factors such as service demand characteristics, flight environment characteristics and network status characteristics to avoid unnecessary manual handover in the traditional single RSSI method, reduce network handover overhead and communication interruption risk.

[0170] (4) Optimization of resource utilization: Dynamically adjusting the hierarchical structure of the decision-making model can effectively avoid base station overload, especially in scenarios with multiple drones or high traffic, thereby improving network resource efficiency.

[0171] (5) Improved communication reliability: By dynamically adjusting the hierarchical structure of the decision-making model, stability can be prioritized during the mission execution phase, ensuring reliable communication and guaranteeing the safety of the UAV and the success of the mission.

[0172] (6) System scalability: Based on decentralized decision-making of mesh network, it supports maintaining communication performance when the number of drones increases, avoiding central bottleneck.

[0173] (7) Energy efficiency optimization: The hierarchical structure of the decision-making model can be dynamically adjusted, and the energy efficiency weight can be adjusted first in resource-constrained scenarios to extend the running time of the UAV, which is suitable for long-term missions.

[0174] See Figure 4 , Figure 4 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 4 The electronic device 300 in this embodiment may include: a memory 304, a processor 301, and a computer program stored in the memory 304 and running on the processor 301. When the processor 301 executes the computer program, it implements the steps of the method in the above-described method embodiments. Specifically, the electronic device 300 may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memory 304 stores the computer program, and when the processor 301 executes the computer program, it implements the steps of the method in the above-described method embodiments. The computer program includes program instructions, and the processor 301 executes the program instructions stored in the memory 304. The processor 301 is configured to invoke the program instructions to perform the functions of each module / unit in the above-described device embodiments, for example... Figure 3 The functions of the data acquisition module 21, feature filtering module 22, judgment module 23, adjustment module 24, weight calculation module 25, score calculation module 26, and base station selection module 27 are shown.

[0175] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), but it may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0176] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0177] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also contain historical data for each base station, the number of base stations, etc.

[0178] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation method described in the UAV roaming scheduling method provided in the embodiments of this application, or they can execute the implementation method of the electronic device described in the embodiments of this application, which will not be repeated here.

[0179] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the methods described in the above embodiments. Specifically, the computer program includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described in the above embodiments. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0180] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0181] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0182] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0183] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the couplings or direct couplings or communication connections shown or discussed may be indirect couplings or communication connections through some interfaces or units, or they may be electrical, mechanical, or other forms of connection.

[0184] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0185] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0186] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for scheduling unmanned aerial vehicle (UAV) roaming, characterized in that, include: Real-time acquisition of the target drone's status characteristics; Based on the state characteristics of the target UAV, the evaluation indicators to be used in the decision-making model are selected, and the selection evaluation indicators are obtained. Based on the selected evaluation indicators, determine whether to adjust the hierarchical structure of the current decision-making model; The step of determining whether to adjust the hierarchical structure of the current decision-making model based on the selected evaluation indicators includes: Determine whether the difference between the current evaluation index used by the current decision model and the selected evaluation index is greater than a specified difference threshold. If it is greater than the specified difference threshold, then it is determined that the hierarchical structure of the current decision model needs to be adjusted. If it is determined that the hierarchical structure of the current decision-making model needs to be adjusted, the hierarchical structure of the current decision-making model is adjusted based on the selected evaluation indicators to obtain the adjusted decision-making model. The subjective weights of the selected evaluation indicators are determined based on the adjusted decision model. The scores of each base station capable of communicating with the target UAV are determined based at least on the subjective weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators. The target base station for communication with the target drone is selected based on the scores of each base station.

2. The method as described in claim 1, characterized in that, The evaluation metrics required for the decision-making model based on the state characteristics of the target UAV include any one of the following: Using a pre-trained feature selection model, the evaluation metrics required for the decision-making model based on the state characteristics of the target UAV are selected. Using preset rules, the evaluation indicators to be used in the decision-making model are selected based on the state characteristics of the target UAV.

3. The method as described in claim 1, characterized in that, The adjustment of the hierarchical structure of the current decision-making model based on the selected evaluation indicators includes at least one of the following: Add or remove hidden layers from the current decision model; Modify the inter-layer connection method of the current decision model; Switch the network architecture of the current decision model to a preset network architecture.

4. The method as described in claim 1, characterized in that, The adjustment of the hierarchical structure of the current decision-making model based on the selected evaluation indicators includes: Based on the evaluation indicators selected, a structure search algorithm is used to select the desired hierarchical structure from a variety of preset hierarchical structures.

5. The method according to any one of claims 1 to 4, characterized in that, After obtaining the evaluation indicators to be used in the decision-making model based on the state characteristics of the target UAV, the method further includes: The objective weights of the selected evaluation indicators are calculated using the entropy weight method. The subjective weights and the objective weights are then combined to obtain the combined weights. The determination of the scores for each base station capable of communicating with the target UAV, based at least on the subjective weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators, includes: The scores of each base station capable of communicating with the target UAV are determined based on the fusion weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators.

6. The method as described in claim 5, characterized in that, The process of fusing the subjective weights and the objective weights to obtain the fused weights includes: The fusion coefficient of the subjective weight and the objective weight is dynamically adjusted by using historical decision feedback and sliding window data to obtain the adjusted fusion coefficient. The subjective weight and the objective weight are fused based on the adjusted fusion coefficient to obtain the fused weight.

7. The method according to any one of claims 1 to 4, characterized in that, The target base stations selected for communication with the target UAV based on the scores of each base station include: Determine whether the score of the current base station communicating with the target drone meets the specified requirements; If the score of the current base station does not meet the specified requirements, the base station with the highest score shall be selected as the target base station for communication with the target drone.

8. The method according to any one of claims 1 to 4, characterized in that, The method further includes: if there are multiple base stations with the highest scores, then selecting the target base station to communicate with the target drone using any of the following methods: Based on the value of an evaluation indicator that was not involved in the scoring, a target base station for communicating with the target UAV is selected from among the multiple base stations with the highest scores; the evaluation indicator that was not involved in the scoring is an indicator included in the state features but not involved in the scoring calculation. Based on historical data from the top-rated base stations, a target base station for communication with the target drone is selected from the top-rated base stations.

9. The method according to any one of claims 1 to 4, characterized in that, After selecting the target base station for communication with the target UAV based on the scores of each base station, the method further includes: suppressing communication interference between the target UAV and the target base station using at least one of the following methods: The power spectral density (PSD) of adjacent and co-channel interference is obtained through periodic measurements. The interference-to-noise ratio (INR) is calculated based on the PSD, and the beam direction is adjusted to point towards the target UAV according to the INR. The priority of the target UAV and each member in the same mesh ad hoc network as the target UAV in the time slot is dynamically adjusted according to the interference level. Based on the interference level and in conjunction with collision detection and failure retransmission calculations, the transmit power, packet buffering, and / or retransmission parameters of the target UAV and each member in the same mesh ad hoc network as the target UAV are dynamically adjusted.

10. A drone roaming scheduling device, characterized in that, include: The data acquisition module is used to acquire the status characteristics of the target UAV in real time. The feature selection module is used to select the evaluation indicators to be used by the decision model based on the state characteristics of the target UAV, and obtain the selected evaluation indicators. The judgment module is used to determine whether to adjust the hierarchical structure of the current decision-making model based on the selected evaluation indicators; The judgment module is used to determine whether to adjust the hierarchical structure of the current decision-making model based on the selected evaluation indicators. Specifically, it is used for: Determine whether the difference between the current evaluation index used in the current decision-making model and the selected evaluation index is greater than a specified difference threshold. If it is greater than the specified difference threshold, then it is determined that the hierarchical structure of the current decision-making model needs to be adjusted. The adjustment module is used to adjust the hierarchical structure of the current decision model based on the selected evaluation indicators when it is determined that the hierarchical structure of the current decision model needs to be adjusted, so as to obtain the adjusted decision model. The weight calculation module is used to determine the subjective weights of the selected evaluation indicators based on the adjusted decision model. The scoring calculation module is used to determine the scores of each base station that can communicate with the target UAV based at least on the subjective weights of the selected evaluation indicators and the quantitative values ​​of the selected evaluation indicators. The base station selection module is used to select a target base station to communicate with the target UAV based on the scores of each base station.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 9.

12. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 9.

Images