Sublimation computing power-driven dynamic link joint optimization method and system, terminal and storage medium

Abstract

The invention discloses a mercuric chloride computing power driven dynamic link joint optimization method and system, a terminal and a storage medium. The method comprises the steps that a link prediction network predicts performance change of each link according to communication link performance data and historical data; the environment sensing network predicts the influence of the environment on the link performance according to the real-time state data of the unmanned aerial vehicle and the environment data; the flight control task resolving module obtains a flight control instruction priority according to the task type and the task requirement; and the reinforcement learning decision module dynamically selects an optimal communication mode according to the performance change of each link, the influence of the environment on the performance of the link and the priority of the flight control instruction. According to the method, the link state is predicted based on the mercuric chloride computing power and by fusing the priority instruction based on the flight control task demand, the communication mode is dynamically switched in real time according to the prediction result, high-reliability, low-delay and high-robustness communication guarantee is realized in a low-altitude economic scene, and core support is provided for users such as unmanned aerial vehicles and air traffic management.

Description

Technical Field

[0001] The present invention relates to the technical field of link optimization, and in particular to a dynamic link joint optimization method, system, terminal and computer-readable storage medium driven by Ascend computing power. Background Art

[0002] With the development of the Internet of Things and 5G technologies, the combination of multi-mode communication modules and edge computing has become a key path to address the large-scale and high-volume data processing requirements. In the low-altitude economy field (such as drone logistics, urban air traffic management), drones can use multi-mode communication protocols to achieve beyond-line-of-sight control and real-time data processing.

[0003] However, the high latency problem during multi-mode protocol switching in the prior art severely restricts application reliability. For example, when a drone flying at low altitude needs to switch from 5G to satellite communication due to building occlusion, the traditional static allocation method lacks an edge computing power dynamic coordination mechanism, and obvious latency is likely to occur during the migration of navigation data in the switching process, which cannot meet the real-time control requirements of the drone, making it difficult for the drone or drone cluster to seamlessly switch protocols and fall into risks such as flight path deviation and mission interruption. This problem greatly limits the large-scale application of the low-altitude economy.

[0004] Therefore, the prior art still needs to be improved and developed. Summary of the Invention

[0005] The main purpose of the present invention is to provide a dynamic link joint optimization method, system, terminal and computer-readable storage medium driven by Ascend computing power, aiming to solve the problem that the high latency problem during multi-mode protocol switching in the prior art severely restricts application reliability.

[0006] To achieve the above objective, the present invention provides a dynamic link joint optimization method driven by Ascend computing power, and the dynamic link joint optimization method driven by Ascend computing power includes the following steps: Obtain communication link performance data and historical data, and input the communication link performance data and the historical data into a link prediction network, and the link prediction network predicts the performance changes of each link according to the communication link performance data and the historical data; Obtain the real-time state data and environmental data of the drone, and input the real-time state data and the environmental data of the drone into an environmental perception network, and the environmental perception network predicts the impact of the environment on the link performance according to the real-time state data and the environmental data of the drone; Obtain the task type and task requirements, and input the task type and the task requirements into a flight control task resolution module, and the flight control task resolution module obtains the flight control instruction priority according to the task type and the task requirements; Input the performance changes of each link, the impact of the environment on the link performance, and the priority of the flight control instructions into the reinforcement learning decision module. The reinforcement learning decision module dynamically selects the optimal communication mode according to the performance changes of each link, the impact of the environment on the link performance, and the priority of the flight control instructions.

[0007] Optionally, in the dynamic link joint optimization method driven by Ascend computing power, where the reinforcement learning decision module dynamically selects the optimal communication mode according to the performance changes of each link, the impact of the environment on the link performance, and the priority of the flight control instructions, it further includes: Output a dynamic switching decision, and feedback the dynamic switching decision back to the link prediction network, the environment perception network, and the flight control task solution module to optimize the network performance.

[0008] Optionally, in the dynamic link joint optimization method driven by Ascend computing power, where the communication link performance data includes: signal strength, delay, channel application response time, single-message success rate, transmission rate, remaining battery power, and data security level of 4G / 5G, Beidou, and Tiantong; The historical data includes: the number and timing of communication link switches during flight, message response time, Tiantong transmission rate, and error code / packet loss rate; The link prediction network predicts the performance changes of each link according to the communication link performance data and the historical data, specifically including: Fuse historical data, hardware device status data, and environmental data to obtain multi-source data, and perform spatio-temporal coordinate alignment on the multi-source data through Ascend NPU; Adopt long short-term memory network and Transformer algorithm to fuse historical data and environmental features for time series data prediction. Among them, the long short-term memory network is used to capture the long-term strong correlation relationship between communication links through the gating mechanism, predict the delay trend within a preset future time, so as to realize the trend prediction of time series data, and the Transformer algorithm is used to analyze the association changes between multiple communication links; Perform communication link switching according to the prediction result.

[0009] Optionally, in the dynamic link joint optimization method driven by Ascend computing power, where the real-time state data of the UAV includes: UAV position, speed, attitude, and altitude; The environmental data includes: surrounding environmental map frequency, obstacles, and no-fly zone information; The environment perception network predicts the impact of the environment on the link performance according to the real-time state data of the UAV and the environmental data, specifically including: Based on the real-time state data of the UAV and the environmental data, analyze the environment where the UAV is located and its own flight state and trend; Predict the transmission conditions, attenuation, and enhancement trends of link signals in the environment space through the environmental perception network algorithm.

[0010] Optionally, in the dynamic link joint optimization method driven by Ascend computing power, the environmental perception network algorithm includes: Feature parsing layer: Locally scan the input environmental map frequency through a convolution kernel array. Each convolution kernel automatically extracts basic features, and the overall shape and size parameters of obstacles are resolved through feature fusion. Decision output layer: Map the spatial features to obstacle attribute parameters, output the classification results of position, size, and shape, and combine the spatial features with link status data to evaluate the transmission and occlusion of signals and data of the communication link in the environment space.

[0011] Optionally, in the dynamic link joint optimization method driven by Ascend computing power, the task types include: inspection, logistics distribution, and emergency rescue; The task requirements include: obstacle avoidance, cruising, and ascending and descending; The flight control task resolution module obtains the flight control instruction priority according to the task type and the task requirements, specifically including: Based on the task type and the task requirements, analyze the requirements for the bandwidth, delay, and data type of data transmission corresponding to the task, and output the required communication link type, priority, and redundancy strategy; Parse the task requirements through Ascend NPU and classify multi-modal data, map the task type to communication link metrics in parallel, dynamically screen the main link from available links and forcefully bind redundant links, and dynamically adjust protocol parameters.

[0012] Optionally, in the dynamic link joint optimization method driven by Ascend computing power, the reinforcement learning decision module dynamically selects the optimal communication mode according to the performance changes of each link, the impact of the environment on link performance, and the flight control instruction priority, specifically including: Based on the performance changes of each link, the impact of the environment on link performance, and the flight control instruction priority, select the optimal communication mode through the reinforcement learning algorithm and ensure redundant links; According to the link state prediction result and task requirements, set a reward function including bandwidth, delay, and handover penalty, use the deep Q-network algorithm to dynamically select the optimal communication mode, and dynamically allocate bandwidth, transmission interval, main link, and redundant link through the proximal policy optimization algorithm, and use Ascend NPU to complete the communication link allocation decision within the delay requirement.

[0013] In addition, to achieve the above object, the present invention further provides a dynamic link joint optimization system driven by Ascend computing power, wherein the dynamic link joint optimization system driven by Ascend computing power includes: A link prediction network for predicting the performance changes of each link according to communication link performance data and historical data; An environment perception network for predicting the impact of the environment on link performance according to the real-time state data of the drone and environmental data; A flight control task resolution module for obtaining the priority of flight control instructions according to the task type and task requirements; A reinforcement learning decision-making module for dynamically selecting the optimal communication mode according to the performance changes of each link, the impact of the environment on link performance, and the priority of flight control instructions.

[0014] In addition, to achieve the above object, the present invention further provides a terminal, wherein the terminal includes: a memory, a processor, and an Ascend computing power-driven dynamic link joint optimization program stored on the memory and executable on the processor. When the Ascend computing power-driven dynamic link joint optimization program is executed by the processor, the steps of the above-mentioned Ascend computing power-driven dynamic link joint optimization method are implemented.

[0015] In addition, to achieve the above object, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores an Ascend computing power-driven dynamic link joint optimization program. When the Ascend computing power-driven dynamic link joint optimization program is executed by a processor, the steps of the above-mentioned Ascend computing power-driven dynamic link joint optimization method are implemented.

[0016] In the present invention, communication link performance data and historical data are obtained, and the communication link performance data and the historical data are input into a link prediction network. The link prediction network predicts changes in the performance of each link based on the communication link performance data and the historical data. Real-time state data and environmental data of the unmanned aerial vehicle (UAV) are obtained, and the real-time state data and the environmental data of the UAV are input into an environmental perception network. The environmental perception network predicts the impact of the environment on the link performance based on the real-time state data and the environmental data of the UAV. The task type and task requirements are obtained, and the task type and the task requirements are input into a flight control task solution module. The flight control task solution module obtains the priority of the flight control instructions based on the task type and the task requirements. The changes in the performance of each link, the impact of the environment on the link performance, and the priority of the flight control instructions are input into a reinforcement learning decision-making module. The reinforcement learning decision-making module dynamically selects the optimal communication mode based on the changes in the performance of each link, the impact of the environment on the link performance, and the priority of the flight control instructions. The present invention is based on Ascend computing power, integrates priority instructions based on flight control task requirements, predicts the link state, and dynamically switches the communication mode in real time according to the prediction results, realizing high-reliability, low-latency, and strong-robustness communication guarantee in the low-altitude economy scenario, and providing core support for users such as UAVs and air traffic management. Description of the Drawings

[0017] Figure 1 is a flowchart of a preferred embodiment of the dynamic link joint optimization method driven by Ascend computing power of the present invention; Figure 2 is a schematic diagram of a preferred embodiment of the dynamic link joint optimization system driven by Ascend computing power of the present invention. Detailed Embodiments

[0018] To make the objectives, technical solutions, and advantages of the present invention clearer and more explicit, the following further elaborates on the present invention by way of examples with reference to the accompanying drawings. 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.

[0019] The dynamic link joint optimization method driven by Ascend computing power according to a preferred embodiment of the present invention, as Figure 1 and Figure 2 shown, the dynamic link joint optimization method driven by Ascend computing power includes the following steps: Step S10: Obtain communication link performance data and historical data, and input the communication link performance data and the historical data into a link prediction network. The link prediction network predicts changes in the performance of each link based on the communication link performance data and the historical data.

[0020] Specifically, the communication link performance data includes: signal strength, latency, channel application response time, single-message success rate, transmission rate, remaining battery power, data security level, etc. of 4G / 5G, Beidou, and Tian Tong; the historical data includes: the number and timing of communication link switches during flight, message response time, Tian Tong transmission rate, and error code / packet loss rate, etc.; the link prediction network is based on the input of communication link performance data and historical data, and predicts the temporal variation law of the three-mode communication link state through the link prediction network algorithm. The temporal variation law of the communication link covers periodic fluctuations (such as the difference in base station load between day and night), trend attenuation (signal weakens with distance), and sudden interference (obstacle occlusion).

[0021] The link prediction network predicts the changes in the performance of each link according to the communication link performance data and the historical data, specifically including: (1) Multi-source data fusion and feature extraction: fuse historical data (historical latency, bandwidth, etc. of the three communication modes, and the three communication modes are Tian Tong satellite communication, Beidou-3 short message communication, and 4G / 5G communication), hardware device status data, and environmental data (satellite angle, real-time weather, surrounding environment, etc.) to obtain multi-source data, and perform spatio-temporal coordinate alignment on the multi-source data through Ascend NPU.

[0022] (2) Spatio-temporal model prediction and NPU accelerated inference: adopt the long short-term memory network (LSTM, Long Short-Term Memory, which is a special recurrent neural network specifically designed to solve the problem of gradient disappearance / explosion of traditional RNNs when dealing with long sequence data and can effectively capture long-term dependencies) and Transformer algorithms to fuse historical data and environmental features for temporal data prediction. Among them, the long short-term memory network is used to capture the long-term strong correlation relationship between communication links through the gating mechanism and predict the latency trend within a preset future time (such as within the next 10 seconds) (such as the direct causal relationship that the increase in base station load leads to an increase in latency) to achieve the trend prediction of temporal data. The Transformer algorithm is used to analyze the associated changes between multiple communication links (such as the attenuation of 5G signals affecting Tian Tong load).

[0023] (3) Dynamic decision-making: perform communication link switching according to the prediction results. Assume that the predicted 5G latency exceeds the threshold after 5 seconds, and switch the main link to Tian Tong satellite 200 ms in advance (active switching), and bind Beidou as redundant to avoid the risk of signal interruption.

[0024] Step S20: Obtain the real-time state data and environmental data of the UAV, and input the real-time state data and environmental data of the UAV into the environmental perception network. The environmental perception network predicts the impact of the environment on the link performance according to the real-time state data and environmental data of the UAV.

[0025] Specifically, the real-time status data of the UAV includes: UAV position, speed, attitude, and altitude; the environmental data includes: surrounding environmental maps, obstacles, no-fly zone information, etc.; based on the input of the UAV real-time status data and environmental data, the environmental perception network analyzes the environment where the UAV is located, its own flight status and trends, and predicts the transmission situation, attenuation, and enhancement trends of the link signal in the environmental space through the environmental perception network algorithm.

[0026] Environmental perception network algorithm: For a large number of input environmental maps, a convolutional neural network (CNN) is used to extract spatial features in the environmental data, such as the position, size, and shape of obstacles.

[0027] Feature analysis layer: Locally scan the input environmental map through a convolutional kernel array. Each convolutional kernel automatically extracts basic features such as the edges of obstacles, adapts to the position offset of obstacles while retaining significant features; analyzes the overall shape and size parameters of obstacles through feature fusion.

[0028] Decision output layer: Map the spatial features to obstacle attribute parameters, and output the classification results of position, size, and shape. Combine the spatial features with the link status data to evaluate the transmission and occlusion of signals and data of the communication link in the environmental space.

[0029] Step S30: Obtain the task type and task requirements, and input the task type and the task requirements into the flight control task solution module. The flight control task solution module obtains the flight control instruction priority according to the task type and the task requirements.

[0030] Specifically, the task types include: inspection, logistics distribution, and emergency rescue; the task requirements include specific scenario requirements such as obstacle avoidance, cruising, and ascending / descending; based on the task type and the task requirements, the flight control task solution module analyzes the requirements for the bandwidth, delay, and data type of the data transmission corresponding to the task, and outputs the required communication link type, priority, and redundancy strategy (i.e., sorting the priorities from high to low in the order of 4G / 5G communication, Beidou-3 short message communication, and Tiantong satellite communication), as shown in Table 1: Table 1: Allocation relationship between the main link and the redundant link

[0031] The flight control task resolution module analyzes task requirements (such as obstacle avoidance, inspection, logistics, emergency, etc.) through the Ascend NPU and classifies multi-modal data (meteorology, environmental images and frequencies collected by UAV sensors, its own state, etc.). It maps the task types in parallel to communication link metrics, such as latency, bandwidth, reliability, etc., and then dynamically filters the main link from the available links and forcibly binds redundant links (i.e., forced dual-channel communication, where one channel is used as a redundant backup to prevent complete communication interruption), and dynamically adjusts protocol parameters.

[0032] Step S40: Input the performance changes of each link, the impact of the environment on the link performance, and the priority of the flight control instructions into the reinforcement learning decision module. The reinforcement learning decision module dynamically selects the optimal communication mode according to the performance changes of each link, the impact of the environment on the link performance, and the priority of the flight control instructions.

[0033] Specifically, the reinforcement learning decision module is based on the performance changes of each link, the impact of the environment on the link performance, and the priority of the flight control instructions (i.e., the changes in the performance metrics of each link output by the link prediction network, such as latency, etc.; the feature results of environmental obstacles, occlusions, etc. output by the environment perception network and the judgment conclusion on the impact on the link signal, and the most required link channel for the current task output by the flight control task resolution module. For example, if the current task requires real-time image transmission, then it is necessary to ensure that 4G / 5G has a signal as much as possible. At the flight control level, it is necessary to avoid obstacles and occlusions as soon as possible to restore 4G / 5G). Through the reinforcement learning algorithm, it selects the optimal communication mode and ensures redundant links. According to the link state prediction results and task requirements, a reward function including bandwidth, latency, and switching penalty is set (in the reinforcement learning algorithm, a penalty term related to action or state switching is introduced into the standard reward to constrain the behavior stability of the agent. The main reward functions include real-time reward (reward for low latency), switching frequency reward (reward for low frequency), and signal interruption reward (reward for no interruption)). The deep Q-network algorithm is used to dynamically select the optimal communication mode, and the proximal policy optimization algorithm is used to dynamically allocate bandwidth, transmission interval, main link, and redundant link. The Ascend NPU is used to complete the communication link allocation decision within the latency requirement.

[0034] Further, as Figure 2 shown, output the dynamic switching decision and feedback the dynamic switching decision back to the link prediction network, the environment perception network, and the flight control task resolution module to optimize the network performance.

[0035] Based on the Ascend computing power, the present invention integrates priority instructions based on flight control task requirements, predicts the link state, and dynamically switches three-mode communication in real time according to the prediction results, realizing "high reliability, low latency, and strong robustness" communication guarantee in the low-altitude economy scenario, and providing core support for users such as UAVs and air traffic management.

[0036] Further, as Figure 2 shown, based on the above-mentioned dynamic link joint optimization method driven by Ascend computing power, the present invention also correspondingly provides a dynamic link joint optimization system driven by Ascend computing power, wherein the dynamic link joint optimization system driven by Ascend computing power includes: A link prediction network for predicting the performance changes of each link according to communication link performance data and historical data; An environment perception network for predicting the impact of the environment on link performance according to the real-time state data of the unmanned aerial vehicle and environmental data; A flight control task solution module for obtaining the priority of flight control instructions according to the task type and task requirements; A reinforcement learning decision-making module for dynamically selecting the optimal communication mode according to the performance changes of each link, the impact of the environment on link performance, and the priority of flight control instructions.

[0037] Further, based on the above-mentioned dynamic link joint optimization method and system driven by Ascend computing power, the present invention also correspondingly provides a terminal, and the terminal includes a processor, a memory, and a display.

[0038] In some embodiments, the memory may be an internal storage unit of the terminal, such as the hard disk or memory of the terminal. In other embodiments, the memory may also be an external storage device of the terminal, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal. Further, the memory may also include both the internal storage unit and the external storage device of the terminal. The memory is used to store application software installed on the terminal and various types of data, such as the program code installed on the terminal. The memory may also be used to temporarily store data that has been output or will be output. In one embodiment, a dynamic link joint optimization program driven by Ascend computing power is stored on the memory, and the dynamic link joint optimization program driven by Ascend computing power can be executed by the processor, thereby implementing the dynamic link joint optimization method driven by Ascend computing power in this application.

[0039] In some embodiments, the processor may be a central processing unit (CPU), a microprocessor, or other data processing chips, for running the program code stored in the memory or processing data, such as executing the dynamic link joint optimization method driven by Ascend computing power, etc.

[0040] In some embodiments, the display may be an LED display, a liquid crystal display, a touch liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display is used to display information on the terminal and to display a visual user interface. The processor, memory, and display of the terminal communicate with each other via a system bus.

[0041] In one embodiment, when the processor executes the dynamic link joint optimization program driven by Ascend computing power in the memory, the steps of the following Ascend computing power-driven dynamic link joint optimization method are implemented.

[0042] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a dynamic link joint optimization program driven by Ascend computing power, and when the dynamic link joint optimization program driven by Ascend computing power is executed by a processor, the steps of the above-mentioned Ascend computing power-driven dynamic link joint optimization method are implemented.

[0043] In summary, the present invention provides a method, system, terminal, and computer-readable storage medium for dynamic link joint optimization driven by Ascend computing power. The method includes: obtaining communication link performance data and historical data, inputting the communication link performance data and the historical data into a link prediction network, and the link prediction network predicting changes in the performance of each link according to the communication link performance data and the historical data; obtaining real-time state data and environmental data of a drone, inputting the real-time state data and the environmental data of the drone into an environmental perception network, and the environmental perception network predicting the impact of the environment on link performance according to the real-time state data and the environmental data of the drone; obtaining a task type and task requirements, inputting the task type and the task requirements into a flight control task solution module, and the flight control task solution module obtaining the priority of flight control instructions according to the task type and the task requirements; inputting the changes in the performance of each link, the impact of the environment on link performance, and the priority of flight control instructions into a reinforcement learning decision-making module, and the reinforcement learning decision-making module dynamically selecting an optimal communication mode according to the changes in the performance of each link, the impact of the environment on link performance, and the priority of flight control instructions. Based on Ascend computing power, the present invention integrates priority instructions based on flight control task requirements, predicts link status, and dynamically switches the communication mode in real time according to the prediction results, achieving high-reliability, low-latency, and strong-robustness communication guarantee in the low-altitude economy scenario, and providing core support for users such as drones and air traffic management.

[0044] It should be noted that in this article, the term "including", "comprising", or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or terminal including a series of elements not only includes those elements but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article, or terminal. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, method, article, or terminal including such element.

[0045] Of course, those of ordinary skill in the art can understand that all or part of the processes of implementing the above-described embodiments of the method can be completed by instructing related hardware (such as a processor, a controller, etc.) through a computer program. The program can be stored in a computer-readable storage medium readable by a computer. When the program is executed, it can include the processes of the above-described method embodiments. The computer-readable storage medium can be a memory, a magnetic disk, an optical disk, etc.

[0046] It should be understood that the application of the present invention is not limited to the above examples. For those of ordinary skill in the art, improvements or transformations can be made according to the above description. All such improvements and transformations should fall within the protection scope of the appended claims of the present invention.

Claims

1. A dynamic link joint optimization method driven by Ascend computing power, characterized in that The Ascend computing power-driven dynamic link joint optimization method includes: Obtain communication link performance data and historical data, and input the communication link performance data and the historical data into a link prediction network. The link prediction network predicts the performance changes of each link according to the communication link performance data and the historical data; Obtain the real-time state data and environmental data of the unmanned aerial vehicle (UAV), and input the real-time state data and the environmental data of the UAV into an environmental perception network. The environmental perception network predicts the impact of the environment on the link performance according to the real-time state data and the environmental data of the UAV; Obtain the task type and task requirements, and input the task type and the task requirements into a flight control task solution module. The flight control task solution module obtains the flight control instruction priority according to the task type and the task requirements; Input the performance changes of each link, the impact of the environment on the link performance, and the flight control instruction priority into a reinforcement learning decision-making module. The reinforcement learning decision-making module dynamically selects the optimal communication mode according to the performance changes of each link, the impact of the environment on the link performance, and the flight control instruction priority.

2. The dynamic link joint optimization method driven by Ascend computing power according to claim 1, wherein After the reinforcement learning decision-making module dynamically selects the optimal communication mode according to the performance changes of each link, the impact of the environment on the link performance, and the flight control instruction priority, it further includes: Output a dynamic switching decision, and feedback the dynamic switching decision back to the link prediction network, the environmental perception network, and the flight control task solution module to optimize the network performance.

3. The dynamic link joint optimization method driven by Ascend computing power according to claim 1, wherein The communication link performance data includes: signal strength, delay, channel application response time, single-message success rate, transmission rate, remaining battery power, and data security level of 4G / 5G, Beidou, and Tiantong; The historical data includes: the number and timing of communication link switches during flight, message response time, Tiantong transmission rate, and error code / packet loss rate; When the link prediction network predicts the performance changes of each link according to the communication link performance data and the historical data, it specifically includes: Fuse historical data, hardware device status data, and environmental data to obtain multi-source data, and perform spatio-temporal coordinate alignment on the multi-source data through the Ascend NPU; Adopt a long short-term memory network and a Transformer algorithm to fuse historical data and environmental features for time-series data prediction. Among them, the long short-term memory network is used to capture the long-term strong correlation relationship between communication links through a gating mechanism, predict the delay trend within a preset future time, so as to realize the trend prediction of time-series data, and the Transformer algorithm is used to analyze the correlation changes between multiple communication links; Perform communication link switching according to the prediction results.

4. The dynamic link joint optimization method driven by Ascend computing power according to claim 1, characterized in that The real-time state data of the UAV includes: UAV position, speed, attitude, and altitude; The environmental data includes: surrounding environmental map frequency, obstacles, and no-fly zone information; When the environmental perception network predicts the impact of the environment on the link performance according to the real-time state data and the environmental data of the UAV, it specifically includes: Based on the real-time state data and the environmental data of the UAV, analyze the environment where the UAV is located, its own flight state and trend; Predict the transmission, attenuation, and enhancement trends of link signals in the environment space through the environmental perception network algorithm.

5. The dynamic link joint optimization method driven by Ascend computing power according to claim 4, wherein The environmental perception network algorithm includes: Feature analysis layer: Locally scan the input environmental map frequency through a convolutional kernel array. Each convolutional kernel automatically extracts basic features, and resolves the overall shape and size parameters of obstacles through feature fusion. Decision output layer: Map the spatial features to obstacle attribute parameters, output the classification results of position, size, and shape, and combine the spatial features with the link status data to evaluate the transmission and occlusion of signals and data of the communication link in the environmental space.

6. The dynamic link joint optimization method driven by Ascend computing power according to claim 1, wherein The task types include: inspection, logistics distribution, and emergency rescue; The task requirements include: obstacle avoidance, cruising, and ascending / descending; The flight control task solution module obtains the flight control instruction priorities according to the task types and the task requirements, specifically including: Based on the task types and the task requirements, analyze the requirements for the bandwidth, delay, and data type of the data transmission corresponding to the task, and output the required communication link type, priority, and redundancy strategy; Parse the task requirements through the Ascend NPU and classify the multi-modal data, map the task types to the communication link metrics in parallel, dynamically screen the main link from the available links and forcefully bind the redundant link, and dynamically adjust the protocol parameters.

7. The dynamic link joint optimization method driven by Ascend computing power according to claim 1, characterized in that The reinforcement learning decision module dynamically selects the optimal communication mode according to the performance changes of each link, the impact of the environment on the link performance, and the flight control instruction priorities, specifically including: Based on the performance changes of each link, the impact of the environment on the link performance, and the flight control instruction priorities, select the optimal communication mode through the reinforcement learning algorithm and ensure the redundant link; According to the link state prediction result and the task requirements, set a reward function including bandwidth, delay, and handover penalty, use the deep Q-network algorithm to dynamically select the optimal communication mode, and dynamically allocate the bandwidth, transmission interval, main link, and redundant link through the proximal policy optimization algorithm, and use the Ascend NPU to complete the communication link allocation decision within the delay requirement.

8. An Ascend computing power-driven dynamic link joint optimization system, characterized in that The Ascend computing power-driven dynamic link joint optimization system includes: Link prediction network, used to predict the performance changes of each link according to the communication link performance data and historical data; Environmental perception network, used to predict the impact of the environment on the link performance according to the real-time state data of the unmanned aerial vehicle and the environmental data; Flight control task solution module, used to obtain the flight control instruction priorities according to the task types and the task requirements; Reinforcement learning decision module, used to dynamically select the optimal communication mode according to the performance changes of each link, the impact of the environment on the link performance, and the flight control instruction priorities.

9. A terminal, characterized in that, The terminal includes: a memory, a processor, and an Ascend computing power-driven dynamic link joint optimization program stored on the memory and executable on the processor. When the Ascend computing power-driven dynamic link joint optimization program is executed by the processor, it implements the steps of the Ascend computing power-driven dynamic link joint optimization method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a dynamic link joint optimization program driven by Ascend computing power. When the dynamic link joint optimization program driven by Ascend computing power is executed by a processor, the steps of the dynamic link joint optimization method driven by Ascend computing power according to any one of claims 1-7 are implemented.

Images