Unmanned aerial vehicle energy supply method, system, device and readable storage medium

Abstract

This invention provides a method, system, device, and readable storage medium for recharging unmanned aerial vehicles (UAVs). The method includes: acquiring multi-dimensional state information of a target UAV; based on the multi-dimensional state information of the target UAV, indexing a preset membership function library and a fuzzy rule library, and outputting a quantified value of charging demand; when the quantified value of charging demand is greater than or equal to a preset charging threshold, determining the charging mode of the target UAV and the rendezvous coordinates of the target UAV and the mobile charging platform based on real-time environmental state information, and recharging the target UAV at the rendezvous coordinates. Compared with existing technologies, this invention weighted and fused information from three dimensions: energy state, task urgency, and arrival difficulty, and simulated the judgment method of human experts through a fuzzy logic controller, achieving on-demand intelligent charging. Simultaneously, this invention can ensure that the target area is always covered by UAVs, allowing for zero-interruption mission execution.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) swarm energy management and collaborative control technology, specifically relating to a method, system, device, and readable storage medium for UAV energy replenishment. Background Technology

[0002] Drone technology has been widely applied in fields such as inspection, surveying, logistics, and agricultural plant protection, and its mission execution capability is directly related to its flight endurance. However, limited by current battery energy density, the flight endurance of drones is generally insufficient, making it difficult to meet the needs of long-term continuous operation. To solve this problem, existing technologies mainly adopt solutions such as setting up fixed charging stations in the mission area or using dedicated charging drones to provide aerial or ground resupply for the working drones.

[0003] The former requires drones to interrupt their current mission and return to a fixed charging station when their battery is low, before returning to the mission area. This results in gaps in mission coverage and prevents continuous monitoring. Setting up multiple stations significantly increases costs and is unsuitable for dynamically changing environments. The latter focuses on precise docking technology. It is sensitive to environmental factors, and the docking success rate is greatly affected by disturbances. Furthermore, the lack of a coordinated scheduling mechanism between charging and working drones makes seamless coordination difficult in complex mission scenarios.

[0004] Therefore, in order to address the aforementioned technical problems, it is necessary to provide a method, system, device, and readable storage medium for replenishing energy for unmanned aerial vehicles (UAVs).

[0005] The information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a method, system, device, and readable storage medium for replenishing energy for unmanned aerial vehicles (UAVs), which can improve the adaptability of UAV charging.

[0007] To achieve the above objectives, the technical solution provided by the present invention is as follows:

[0008] In a first aspect, the present invention provides a method for replenishing the energy of a drone, comprising:

[0009] Acquire multi-dimensional state information of the target drone, including at least energy state information representing its energy level, task state information representing its mission urgency, and environmental state information representing the difficulty of its arrival at the charging platform.

[0010] Based on the multi-dimensional state information of the target UAV, the preset membership function library and fuzzy rule library are indexed to output a quantitative value of charging demand.

[0011] When the quantified value of the charging demand is greater than or equal to the preset charging threshold, the charging mode of the target drone and the rendezvous coordinates of the target drone and the mobile charging platform are determined based on the real-time environmental status information, and the target drone is replenished with energy at the rendezvous coordinates.

[0012] In one or more embodiments of the present invention, when the quantified value of the charging demand is greater than or equal to a preset charging threshold, the method further includes:

[0013] Determine the quantified charging requirements of other drones in the cluster besides the target drone;

[0014] Based on the inter-machine communication link, the context data corresponding to the unfinished task of the target drone is transmitted to the collaborating drones in the cluster; the collaborating drones are the other drones in the cluster with the smallest quantified charging demand value, and the quantified charging demand value of the collaborating drones is less than the charging threshold.

[0015] The collaborative drone generates an extended working area based on context data sent from the target drone, and plans a flight path within the extended working area based on the unfinished tasks of the target drone.

[0016] In one or more embodiments of the present invention, obtaining the multi-dimensional state information of the target UAV includes:

[0017] Based on the target drone's real-time battery level, battery health, and estimated remaining flight time, the energy state information is calculated using a first weighted fusion function.

[0018] Based on the amount of tasks already completed by the target UAV, the priority of the current task, and the criticality of the task area, the task status information is calculated using a second weighted fusion function.

[0019] The environmental state information is calculated using a third weighted fusion function based on the relative distance between the target drone and the intelligent mobile charging platform, the current wind speed, and the density of obstacles along the path.

[0020] In one or more embodiments of the present invention, the step of outputting a quantitative value of charging demand based on the multi-dimensional state information of the target UAV, indexing a preset membership function library and a fuzzy rule library, includes:

[0021] The precise values ​​of the energy state information, task state information, and environmental state information are mapped to corresponding fuzzy linguistic values ​​through their respective membership functions.

[0022] The fuzzy language values ​​are input into a preset IF-THEN fuzzy rule base for parallel reasoning to obtain a fuzzy output result of charging demand.

[0023] The fuzzy output result is defuzzified to obtain the quantified value of the target drone's charging requirement.

[0024] In one or more embodiments of the present invention, determining the charging mode of the target drone based on real-time environmental state information includes:

[0025] When the wind speed value in the environmental status information is less than or equal to the first preset threshold, the wired charging mode that transmits energy through physical contact is selected.

[0026] When the wind speed value in the environmental status information is greater than the first preset threshold, a wireless charging mode that transmits energy through electromagnetic induction or magnetic resonance is selected.

[0027] In one or more embodiments of the present invention, the method further includes:

[0028] Construct a neural network model and set relevant parameters for the neural network model, wherein the relevant parameters include at least the weight coefficients and biases of each network layer;

[0029] Obtain a first training sample set, which includes multi-dimensional state data, charging decision data, and task completion effect evaluation data stored in a historical database;

[0030] The neural network model is trained based on the training sample set, and the parameters of the neural network model are corrected until the loss function of the neural network model converges or reaches the preset number of training iterations.

[0031] In one or more embodiments of the present invention, the method further includes:

[0032] Record the current multi-dimensional status information, decision results, and mission execution effects of the drone;

[0033] When the judgment result of the charging demand based on the trained neural network model is inconsistent with the judgment result based on fuzzy logic, the judgment result based on the neural network model updates the charging threshold used by the fuzzy logic, the weight coefficient of the weighted fusion function, or the rules in the fuzzy rule base.

[0034] Alternatively, the relevant parameters of the neural network model can be updated based on the judgment result of the fuzzy logic.

[0035] Secondly, the present invention provides a drone energy replenishment system, wherein the drone energy replenishment method described above includes:

[0036] The acquisition module is used to acquire multi-dimensional state information of the target UAV. The multi-dimensional state information includes at least energy state information representing its energy level, task state information representing its mission execution urgency, and environmental state information representing the difficulty of its arrival at the charging platform.

[0037] The demand determination module is used to output a quantitative value of charging demand based on the multi-dimensional state information of the target UAV, indexing a preset membership function library and a fuzzy rule library.

[0038] The replenishment module is used to determine the charging mode of the target drone and the rendezvous coordinates of the target drone and the mobile charging platform based on real-time environmental status information when the quantified value of the charging demand is greater than or equal to a preset charging threshold, and to replenish the energy of the target drone at the rendezvous coordinates.

[0039] Thirdly, the present invention provides an electronic device comprising:

[0040] At least one processor; and a memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform the UAV energy replenishment method.

[0041] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the aforementioned UAV energy replenishment method.

[0042] Compared with existing technologies, the drone energy replenishment method provided by this invention weighted and fused information from three dimensions: energy status, mission urgency, and arrival difficulty. It also simulates the judgment of human experts through a fuzzy logic controller, achieving intelligent on-demand charging. Furthermore, this invention ensures continuous drone coverage of the target area through a dual-drone dynamic role switching and mission relay mechanism, enabling zero-interruption mission execution. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, 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 recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1This is a schematic diagram illustrating an implementation scenario of the UAV energy replenishment method in one embodiment of the present invention;

[0045] Figure 2 This is a flowchart of a drone energy replenishment method in one embodiment of the present invention;

[0046] Figure 3 This is a structural block diagram of an unmanned aerial vehicle (UAV) energy replenishment system according to another embodiment of the present invention;

[0047] Figure 4 This is a structural block diagram of an electronic device according to one embodiment of the present invention;

[0048] Figure 5 This is a schematic diagram of the operation logic of a fuzzy logic controller according to one embodiment of the present invention;

[0049] Figure 6 This is a flowchart illustrating the instantiation of task relay in one embodiment of the present invention. Detailed Implementation

[0050] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this invention.

[0051] Unless otherwise expressly stated, throughout the specification and claims, the term "comprising" or its variations such as "including" or "comprises" shall be understood to include the stated elements or components without excluding other elements or other components.

[0052] Please refer to Figure 1 The diagram shows an application scenario of the UAV energy replenishment method provided by the present invention under one embodiment. The scenario specifically includes: a perception layer 101, a decision layer 102, and an execution layer 103.

[0053] It should be noted that communication connections are established between the perception layer 101, the decision-making layer 102, and the execution layer 103, forming a bidirectional data flow channel. The communication network upon which these connections rely can include various connection types, including but not limited to wired connections, wireless connections, or fiber optic cable connections. Furthermore, this communication network can be a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), or any combination of these three types to ensure low-latency, highly reliable information exchange between multiple nodes.

[0054] Specifically, the perception layer 101 can consist of a drone swarm and an intelligent mobile charging platform. Both the drones and the intelligent mobile charging platform are equipped with perception and communication modules. The perception modules can include, but are not limited to, visible light cameras, infrared sensors, and millimeter-wave radar. The perception layer is used to achieve self-state perception and surrounding environment perception. The core state information collected includes at least the energy state representing remaining available energy, the task state representing task progress, and the environmental and positional state representing environmental interference and spatial distance. The perception layer exchanges the above core state information in real time among the three entities through the communication module, especially establishing a high-speed inter-drone data link between the target drone and the cooperating drones.

[0055] The decision-making layer 102 can consist of a centralized scheduling controller deployed on a ground station server or a cloud platform. The centralized scheduling controller is responsible for receiving multi-dimensional status information uploaded from the perception layer, running a fuzzy logic controller to make real-time judgments on charging demand, and optimizing decision parameters offline or online through reinforcement learning and fuzzy neural network models. Based on the judgment results, the decision-making layer generates scheduling instructions and sends these instructions to the execution layer.

[0056] The execution layer 103 is the physical action implementation unit of the system. The physical actions may include, but are not limited to: flight control and mission execution of the UAV, autonomous movement control of the intelligent mobile charging platform, mechanical docking and energy transfer of charging equipment, and automatic switching of multi-mode charging. After receiving instructions from the decision layer, the execution layer drives the coordinated actions of each physical entity to complete the complete closed-loop process from mission continuation to energy replenishment.

[0057] Please refer to Figure 2 The diagram shown illustrates a process for recharging a drone's energy according to an embodiment of the present invention. The drone energy recharging method specifically includes the following steps:

[0058] S201: Acquire multi-dimensional status information of the target UAV;

[0059] In an exemplary embodiment, the multi-dimensional state information includes at least energy state information characterizing its energy level, task state information characterizing its task execution urgency, and environmental state information characterizing its difficulty in reaching the charging platform.

[0060] Specifically, in one embodiment, the energy state can be denoted as Its quantitative indicator is energy level. ,Depend on Decide, yes , , The function. Among them Represents real-time battery level. This indicates the battery's health status. This represents the estimated remaining battery life. , , All values ​​need to be normalized, and the range is 0 to 1. It is worth noting that... , , All of these can be provided by the battery management software equipped on the drone.

[0061] The specific function form can be expressed as:

[0062]

[0063] In the formula, and Let be the weighting coefficient, satisfying The initial stage can be set. and To emphasize the dominant role of currently available power, this weight can be optimized and adjusted in subsequent reinforcement learning phases. This is the effective real-time battery level. (Based on battery health status) This directly affects the actual capacity of energy; therefore, the first step is to use... Real-time power consumption Corrections are made to obtain the effective real-time power level. ,Right now .

[0064] Furthermore, the task status can be recorded as... Its quantitative indicator is the urgency of the task. ,Depend on Decide, yes The function, where Represents the amount of tasks executed. Represents task priority. This indicates the criticality of the task area. All values ​​need to be normalized, and the range is 0 to 1. It is worth noting that... The settings can be configured based on experience before the drone performs a mission. This requires calculation by a specific algorithm, which can be determined based on the ratio of unfinished tasks to all tasks. For example, the task execution can be measured from the perspective of the amount of data involved in the task. For instance, if the task requires capturing 10M of images and 5M has already been captured, then the task execution amount W1 is 0.5.

[0065] The specific function form can be expressed as:

[0066]

[0067] In the formula, , , Let be the weighting coefficients of each component, satisfying... This is used to adjust the impact of different dimensions on task urgency. In the initial stage, , , It can be set based on experience and can be further optimized and adjusted in subsequent reinforcement learning phases.

[0068] The environmental and locational conditions are denoted as P, and their quantitative indicator is the difficulty of reaching the destination. ,Depend on The decision is The function, where Distance to the destination For wind speed, The density is the obstacle density. The value ranges from 0 to 1. Distance to the destination. Wind speed =min(current wind speed / maximum wind speed for safe drone flight, 1), obstacle density = (Number of obstacles on the path / Distance of the planned flight path).

[0069] It should be noted that the system has prior knowledge of the global map information before executing the mission. The UAV is equipped with a limited-range perception module (detection distance R). After planning the flight path, the UAV uses map data to count the number of obstacles within a width of R on both sides of the path, which is the number of obstacles on the path.

[0070] The specific function form can be expressed as:

[0071]

[0072] In the formula, , , Let be the weighting coefficients of each component, satisfying... In the initial stage, , , It can be set by expert experience and can be optimized and adjusted in subsequent reinforcement learning phases.

[0073] Furthermore, in another implementation, an overall weight set for the three core states can be set. That is, in the overall weight set, , , The weighting coefficients for energy state, task state, and environment and position state are respectively, satisfying... Depending on the type of task, the weight of each state varies and can be flexibly adjusted. , , In routine inspection tasks, the weight of task status may be relatively high; however, under severe weather conditions, the weights of energy status and environmental status will increase significantly. Initially, weights set using expert experience or the AHP method (Analytic Hierarchy Process) can be used as initial values. Subsequently, these weight allocations can be adjusted in real time by the system's reinforcement learning model and optimization adjustment strategies.

[0074] S202: Based on the multi-dimensional state information of the target UAV, index the preset membership function library and fuzzy rule library, and output the quantitative value of charging demand;

[0075] It should be noted that, in one embodiment, the present invention can introduce a fuzzy logic controller to address the uncertainty of state determination. When the centralized scheduling controller receives the three quantified state information indicators, it does not immediately make a judgment, but needs to input them into the fuzzy logic controller. After internal calculation by the fuzzy logic controller, it outputs the specific value of the charging demand and compares it with the charging threshold.

[0076] For example, if a drone is performing a mission with "low energy level" (0.25), "medium mission urgency" (0.5), and "low arrival difficulty" (0.3), the IF-THEN rule base of the fuzzy logic controller may calculate and output "high" charging demand. At the same time, the specific value of the charging demand is obtained through defuzzification operation.

[0077] As the core of the system's intelligent decision-making, the fuzzy logic controller's operational logic flow is as follows: Figure 5 As shown. First, based on the aforementioned overall weight set, the collected multi-dimensional state information of the UAV is preprocessed with weighted quantification indicators to obtain... This process yields precise values ​​for three state information quantification indicators. These precise values ​​are then transformed into fuzzy language descriptions such as "low," "medium," and "high" by a preset membership function to characterize their uncertainty. Subsequently, the fuzzy logic controller invokes the IF-THEN rule base, which encapsulates expert experience, for parallel reasoning. Each rule is activated based on the degree of satisfaction (membership) of the input conditions, resulting in a fuzzy output conclusion. Finally, the fuzzy logic controller employs common defuzzification strategies such as the centroid method to aggregate and weigh the fuzzy outputs of all activated rules, calculating a precise charging demand value. This transforms qualitative, human-like fuzzy judgments into quantitative, executable scheduling instructions, triggering the subsequent collaborative charging process.

[0078] To further clarify the specific implementation of the fuzzy logic controller, this embodiment provides the following detailed explanation of the membership function and the defuzzification method:

[0079] (1) Definition of membership function

[0080] In this embodiment, the three input variables—energy state E, task urgency W, and arrival difficulty P—are respectively divided into three fuzzy sets: "low" (L), "medium" (M), and "high" (H), and fuzzification is performed using a triangular membership function. Taking energy state E as an example, its membership function can be defined as follows:

[0081]

[0082]

[0083]

[0084] Similarly, the task urgency W and the difficulty of arrival P adopt the same triangular membership function form and parameters, with their normalized values ​​used as inputs. The above parameters (0.2, 0.4, 0.6) can be adjusted according to the actual application scenario, and this implementation method does not impose specific limitations.

[0085] (2) Defuzzification method

[0086] This implementation method can use the Center of Gravity (COG) method to defuzzify the fuzzy output, and the calculation formula is as follows:

[0087]

[0088] in, To determine the precise output value after defuzzification (i.e., the quantified value of charging demand), n represents the total number of activated fuzzy rules. Let be the activation strength of the i-th rule (i.e., the minimum value of all membership degrees in the antecedent of the rule). Let be the centroid value of the consequent fuzzy set of the i-th rule. In this embodiment, the centroid value of "low" is preset to 0.2, "medium" is preset to 0.5, and "high" is preset to 0.85. Their specific values ​​can also be flexibly adjusted based on the actual use scenario.

[0089] The IF-THEN rule base can be generated using expert experience in the initial stage. In a specific embodiment, it can be, for example:

[0090] IF Energy_Level IS L AND Work_Urgency IS L AND Arrival_Difficulty IS L THEN Charging_Demand IS H (If the energy level is low, the task urgency is low, and the arrival difficulty is low, then the charging demand is high)

[0091] If Energy_Level IS M AND Work_Urgency IS H AND Arrival_Difficulty IS SL THEN Charging_Demand IS M (If the energy level is medium, the task urgency is high, and the arrival difficulty is low, then the charging demand is medium)

[0092] IF Energy_Level IS H AND Work_Urgency IS H AND Arrival_Difficulty IS H THEN Charging_Demand IS L (If the energy level is high, the task urgency is high, and the arrival difficulty is high, then the charging demand is low)

[0093] IF Energy_Level IS L AND Arrival_Difficulty IS H THEN Charging_Demand IS H (If the energy level is low and the arrival difficulty is high, then the charging demand is high)

[0094] It should also be noted that in the subsequent reinforcement learning stage, the fuzzy judgments of the IF-THEN rule base can be modified by collecting task completion data through the reinforcement learning model, so as to achieve higher completion rate and efficiency for specific types of tasks.

[0095] It is important to note that the remaining "return battery power" of a drone at the end of a mission is not equivalent to "absolutely safe battery power." For example, even after the mission is completed, environmental conditions still exist. During mission execution, redundant energy may have been exhausted to complete the mission. If the drone encounters a sudden increase in crosswinds or needs to navigate around obstacles during the return journey, a drone in a low battery state is highly susceptible to crashing. Therefore, in an exemplary embodiment of the present invention, to ensure a safe return, any drone, after completing all missions, still needs to determine its charging needs through the fuzzy logic controller of the centralized scheduling controller. The process is the same as described above. However, since the charging need determination at this time is unrelated to the mission state, the weights need to be adjusted. Set to 0.

[0096] S203: When the quantified value of the charging demand is greater than or equal to the preset charging threshold, the charging mode of the target drone and the rendezvous coordinates of the target drone and the mobile charging platform are determined based on the real-time environmental status information, and the target drone is replenished with energy at the rendezvous coordinates.

[0097] In this invention, the centralized scheduling controller no longer maintains a fixed role division, but performs dynamic scheduling, that is, it sends a "return to charging" command to the target drone that needs charging, sends a "task takeover" command to the cooperating drone, and sends a "welcome to charge" command to the intelligent mobile charging platform.

[0098] The centralized scheduling controller, based on the location information of the target drone and the intelligent mobile charging platform, and comprehensively considering movement-related information such as their energy and speed, uses mathematical optimization or deep learning to find a suitable intermediate position, denoted as (x0, y0, z0), towards which both the target drone and the intelligent mobile charging platform move. The mathematical optimization algorithm or deep learning model used to calculate the intermediate position is not limited here, as long as it outputs a position coordinate.

[0099] In one implementation, to achieve seamless task handover, before moving to (x0, y0, z0), the target UAV can transmit task context data, including its current "electronic fence" boundary, unfinished inspection path waypoints, and any abnormal points of interest, to the collaborating UAV at high speed via beamforming technology in wireless communication. Upon receiving the data, the collaborating UAV's flight controller temporarily generates an expanded working area, incorporating the target UAV's unserved working area. By dynamically adjusting its own flight path, it ensures that the target UAV's unserved working area is covered and the task can be executed. The path planning algorithm used to implement the collaborative UAV's flight path adjustment is not limited in this invention; for example, the D* algorithm can be used to plan the flight path after changing destinations. Simultaneously, the intelligent mobile charging platform moves to (x0, y0, z0) and reserves a charging position for the target UAV, preparing to enter the charging process.

[0100] It should be noted that when the specific value of the charging demand exceeds the set charging threshold, the centralized scheduling controller triggers the charging scheduling process. The charging threshold can be initially obtained from the actor-critic model, and subsequently, a new threshold can be derived from the experience collected by the reinforcement learning model to directly replace the original threshold.

[0101] Once the collaborative drone reaches the designated location (x0, y0, z0), it will hover and await guidance from the intelligent mobile charging platform. The intelligent mobile charging platform will then stop upon reaching the designated location (x0, y0, z0), guiding the drone to land on the platform and providing two charging modes.

[0102] Among them, Mode A (wired charging) uses a robotic arm to connect the charging cable to the drone's charging port to provide wired charging for the drone; Mode B (wireless charging) requires the drone to be within a fixed location range and uses magnetic resonance to perform wireless charging.

[0103] The centralized scheduling controller records all energy-related data for the drone, including charging demand assessments, energy consumption data, and task completion status, forming a historical data database. The centralized scheduling controller selects a machine learning model capable of solving the binary classification problem of charging, such as Support Vector Machine (SVM) or Random Forest, and trains it based on this database. Based on the current data, the centralized scheduling controller determines the charging demand using the trained machine learning model.

[0104] In one implementation, if the charging demand judgment obtained by the machine learning model is the same as the judgment of the fuzzy logic controller, then the information weighting value of the three states (i.e., ...) is maintained. , , , , , , ), weighted preprocessing weighted values ​​( , , ), charging threshold, and weight parameters, membership functions, and fuzzy rules in the fuzzy logic controller.

[0105] In another implementation, if the charging demand judgment obtained by the machine learning model differs from the judgment of the fuzzy logic controller, the charging threshold used by the fuzzy logic, the weight coefficient of the weighted fusion function, or the rules in the fuzzy rule base can be updated based on the judgment result of the neural network model; or the relevant parameters of the neural network model can be updated based on the judgment result of the fuzzy logic.

[0106] Specifically, in the first implementation, the information of the three states can be weighted (i.e., , , , , , , ), weighted preprocessing weighted values ​​( , , The charging threshold is determined, and a reinforcement learning model is selected for modification and optimization. The centralized scheduling controller uses the aforementioned database to train a specific reinforcement learning model and then uses the trained model to optimize the weighted values ​​of the information from the three states (i.e., charging threshold). , , , , , , ), weighted preprocessing weighted values ​​( , , The charging threshold is also considered. In the second implementation, a fuzzy neural network (FNN) can be used to modify and optimize the weight parameters, membership functions, and fuzzy rules in the fuzzy logic controller. The centralized scheduling controller will use the aforementioned database to train a specific fuzzy neural network model and then use the trained model to optimize the weight parameters, membership functions, and fuzzy rules in the fuzzy logic controller. Through these modifications and optimizations, the charging scheduling strategy can gradually adapt to specific task modes and environmental characteristics, becoming increasingly intelligent and efficient, and increasingly suited to the user's specific task modes.

[0107] To more clearly illustrate the implementation steps of the technical solution of the present invention, the following is a complete and specific embodiment of the present invention.

[0108] This embodiment is applied to a scenario of 24-hour uninterrupted safety inspection in a chemical plant area. The system is configured with an intelligent mobile charging platform, a centralized dispatch controller, and multiple drones. Specifically, the drones are equipped with sensing and communication modules and can connect to the charging interface of the intelligent mobile charging platform. The intelligent mobile charging platform has a sensing and positioning system and supports both wired conductive charging and hovering wireless charging modes. The centralized dispatch controller is deployed on the edge computing server in the plant control center and integrates a fuzzy logic controller and a reinforcement learning module.

[0109] Step 1: UAV-A is responsible for inspecting area A, and UAV-B is responsible for inspecting area B. Both drones collect three types of core status information in real time through their sensing and communication modules, and upload this information to the centralized dispatch controller via wireless communication.

[0110] (1) UAV-A status data:

[0111] State of Energy (E): Real-time battery level (E_1=0.45) (normalized), battery health status (E_2=0.95), estimated remaining range (E_3=0.2) (maximum range 60 minutes, 12 minutes remaining), effective real-time battery level (E_effective=E_1×E_2=0.45×0.95=0.4275), energy level (Energy_Level=0.7×0.4275+0.3×0.2=0.29925+0.06=0.35925).

[0112] Task status (W): Task execution volume (W_1=0.65), Task priority (W_2=0.8), Area criticality (W_3=0.7), Task urgency (Work_Urgency=0.4×(1-0.65)+0.3×0.8+0.3×0.7=0.14+0.24+0.21=0.59).

[0113] Environmental and location conditions (P): Distance normalized (P_1=0.4) (800m / 2000m), Wind speed normalized (P_2=0.533) (8m / s / 15m / s), Obstacle density (P_3=1.0), Arrival Difficulty (Arrival_Difficulty=0.3×0.4+0.3×0.533+0.4×1.0=0.12+0.1599+0.4=0.6799)

[0114] (2) UAV-B status data:

[0115] Energy state (E): (E_1=0.45), (E_2=0.95), (E_3=0.333), (E_effective=0.4275), (Energy_Level=0.7×0.4275+0.3×0.333=0.29925+0.0999=0.39915);

[0116] Task status (W): (W_1=0.40), (W_2=0.6), (W_3=0.5), (Work_Urgency=0.4×0.6+0.3×0.6+0.3×0.5=0.24+0.18+0.15=0.57);

[0117] Environmental and location status (P): (P_1=0.6), (P_2=0.4), (P_3=1.0), (Arrival_Difficulty=0.3×0.6+0.3×0.4+0.4×1.0=0.18+0.12+0.4=0.70)

[0118] The centralized scheduling controller performs weighted fusion of the above information to generate three types of quantitative indicators for subsequent fuzzy logic judgment.

[0119] Step 2: The controller inputs the UAV-B's Energy_Level=0.399, Work_Urgency=0.57, Arrival_Difficulty=0.70 into the fuzzy logic controller, and sets the overall weights W_E=0.4, W_W=0.4, and W_P=0.2 based on the task type.

[0120] First, perform fuzzy processing on the three input variables through the triangular membership functions defined above, and calculate the membership values corresponding to each input variable:

[0121] Energy state

[0122] Precise calculation:

[0123] Actually, piecewise calculation is adopted: When When

[0124] falls within

[0125]

[0126]

[0127] Task urgency

[0128]

[0129] When When, it falls within the range of 0.4 < W < 0.6,

[0130]

[0131]

[0132] Arrival difficulty

[0133]

[0134]

[0135]

[0136] After being transformed by the membership function, the input variables are described as: Energy level: close to "low"; Task urgency: close to "medium"; Arrival difficulty: close to "high";

[0137] The following rules are matched in the fuzzy rule base:

[0138] >IF Energy_Level IS L AND Arrival_Difficulty IS H THEN Charging_Demand IS H

[0139] Adopt the Center of Gravity (COG) method for defuzzification calculation:

[0140]

[0141] Substitute the values ​​from the activation rules above:

[0142]

[0143]

[0144] After defuzzification (centroid method) calculation, the charging demand value is 0.82, which exceeds the system's set charging threshold of 0.70, and the controller triggers the charging scheduling process.

[0145] Step 3: The controller sends a "return to charging" command to UAV-B, a "task takeover" command to UAV-A, and a "welcome charging" command to the intelligent mobile charging platform.

[0146] Before leaving the mission area, UAV-B transmits mission context data, such as the incomplete inspection path and points of concern in area B, to UAV-A at high speed using beamforming technology. After receiving the data, UAV-A dynamically adjusts its flight path, uses the D* algorithm to generate a new inspection route, and temporarily expands its working area to cover key points in area B.

[0147] Meanwhile, the controller calculates the optimal rendezvous point coordinates (x_0, y_0, z_0) based on the current position, speed, energy, and other information of the UAV-B and the charging platform, and sends them to the UAV-B and the charging platform respectively, so that the two move towards the point synchronously.

[0148] Step 4: Optimization of charging strategy based on machine learning

[0149] The UAV-B and the charging platform met at the rendezvous point. Due to the high wind speed at the scene, the charging platform automatically activated the "hover wireless charging" mode, allowing the UAV-B to recharge while hovering. After 45 minutes, the battery level was restored to 85%, and the UAV-B returned to area B to take over the mission.

[0150] The centralized scheduling controller records energy data, task completion status, environmental parameters, and charging decision results throughout the entire scheduling process and stores them in a historical database. The system uses this data to train a support vector machine (SVM) model for auxiliary judgment of subsequent charging needs.

[0151] If the SVM model's judgment is consistent with the output of the fuzzy logic controller, the current parameters remain unchanged; otherwise, the system starts the reinforcement learning module (used to optimize the weight coefficients and charging thresholds) and the fuzzy neural network module (used to optimize the membership function and fuzzy rules) to gradually adjust the decision logic to better adapt it to the actual operating characteristics of the plant inspection task.

[0152] Please refer to Figure 3As shown, based on the same inventive concept as the aforementioned UAV energy replenishment method, one embodiment of the present invention provides a UAV energy replenishment system 300, including: an acquisition module 301, a demand determination module 302, and a replenishment module 303.

[0153] It should be noted that the drone energy replenishment method of this embodiment can be applied to the drone energy replenishment system 300 of this embodiment. The drone energy replenishment system 300 can be configured on a terminal. The terminal may include, but is not limited to, a PC (Personal Computer), a PDA (Tablet PC), a smartphone, a smart wearable device, etc.

[0154] Specifically, the acquisition module 301 is used to acquire multi-dimensional state information of the target UAV. The multi-dimensional state information includes at least energy state information representing its energy level, task state information representing its mission execution urgency, and environmental state information representing the difficulty of its arrival at the charging platform.

[0155] The demand determination module 302 is used to output a quantitative value of charging demand based on the multi-dimensional state information of the target UAV, indexing a preset membership function library and a fuzzy rule library.

[0156] The replenishment module 303 is used to determine the charging mode of the target drone and the rendezvous coordinates of the target drone and the mobile charging platform based on real-time environmental status information when the quantified value of the charging demand is greater than or equal to a preset charging threshold, and to replenish the energy of the target drone at the rendezvous coordinates.

[0157] Please refer to Figure 4 As shown, embodiments of the present invention also provide an electronic device 400, which includes at least one processor 401, a memory 402 (e.g., non-volatile memory), a main memory 403, and a communication interface 404, and the at least one processor 401, memory 402, main memory 403, and communication interface 404 are connected together via an internal bus 405. The at least one processor 401 is used to invoke at least one program instruction stored or encoded in the memory 402 to cause the at least one processor 401 to perform various operations and functions of the UAV energy replenishment method described in the various embodiments of this specification.

[0158] In the embodiments of this specification, electronic device 400 may include, but is not limited to: personal computer, server computer, workstation, desktop computer, laptop computer, notebook computer, mobile electronic device, smartphone, tablet computer, cellular phone, personal digital assistant (PDA), handheld device, messaging device, wearable electronic device, consumer electronic device, etc.

[0159] This invention also provides a computer-readable medium carrying computer-executable instructions. When executed by a processor, these instructions can be used to implement various operations and functions of the UAV energy replenishment methods described in the various embodiments of this specification.

[0160] The computer-readable medium in this invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0161] In this invention, the computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. This propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wireline, optical fiber, RF, etc., or any suitable combination thereof.

[0162] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0163] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus, systems, and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0164] The foregoing description of specific exemplary embodiments of the invention is for illustrative and explanatory purposes. These descriptions are not intended to limit the invention to the precise forms disclosed, and it will be apparent that many changes and variations can be made in accordance with the foregoing teachings. The exemplary embodiments were chosen and described in order to explain the specific principles of the invention and its practical application, thereby enabling those skilled in the art to implement and utilize various different exemplary embodiments of the invention, as well as various different choices and variations. The scope of the invention is intended to be defined by the claims and their equivalents.

[0165] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0166] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for replenishing the energy of a drone, characterized in that, include: Acquire multi-dimensional state information of the target drone, including at least energy state information representing its energy level, task state information representing its mission urgency, and environmental state information representing the difficulty of its arrival at the charging platform. Based on the multi-dimensional state information of the target UAV, the preset membership function library and fuzzy rule library are indexed to output a quantitative value of charging demand. When the quantified value of the charging demand is greater than or equal to the preset charging threshold, the charging mode of the target drone and the rendezvous coordinates of the target drone and the mobile charging platform are determined based on the real-time environmental status information, and the target drone is replenished with energy at the rendezvous coordinates.

2. The method for replenishing the energy of a UAV according to claim 1, characterized in that, When the quantified value of the charging demand is greater than or equal to a preset charging threshold, the method further includes: Determine the quantified charging requirements of other drones in the cluster besides the target drone; Based on the inter-machine communication link, the context data corresponding to the unfinished task of the target drone is transmitted to the collaborating drones in the cluster; the collaborating drones are the other drones in the cluster with the smallest quantified charging demand value, and the quantified charging demand value of the collaborating drones is less than the charging threshold. The collaborative drone generates an extended working area based on context data sent from the target drone, and plans a flight path within the extended working area based on the unfinished tasks of the target drone.

3. The method for replenishing the energy of a UAV according to claim 1, characterized in that, The acquisition of multi-dimensional state information of the target UAV includes: Based on the target drone's real-time battery level, battery health, and estimated remaining flight time, the energy state information is calculated using a first weighted fusion function. Based on the amount of tasks already completed by the target UAV, the priority of the current task, and the criticality of the task area, the task status information is calculated using a second weighted fusion function. The environmental state information is calculated using a third weighted fusion function based on the relative distance between the target drone and the intelligent mobile charging platform, the current wind speed, and the density of obstacles along the path.

4. The method for replenishing the energy of a UAV according to claim 1, characterized in that, The step of indexing a preset membership function library and fuzzy rule library based on the multi-dimensional state information of the target UAV, and outputting a quantitative value of charging demand, includes: The precise values ​​of the energy state information, task state information, and environmental state information are mapped to corresponding fuzzy linguistic values ​​through their respective membership functions. The fuzzy language values ​​are input into a preset IF-THEN fuzzy rule base for parallel reasoning to obtain a fuzzy output result of charging demand. The fuzzy output result is defuzzified to obtain the quantified value of the target drone's charging requirement.

5. The method for replenishing the energy of a UAV according to claim 1, characterized in that, The charging mode of the target drone is determined based on real-time environmental status information, including: When the wind speed value in the environmental status information is less than or equal to the first preset threshold, the wired charging mode that transmits energy through physical contact is selected. When the wind speed value in the environmental status information is greater than the first preset threshold, a wireless charging mode that transmits energy through electromagnetic induction or magnetic resonance is selected.

6. The method for replenishing the energy of a UAV according to claim 1, characterized in that, The method further includes: Construct a neural network model and set relevant parameters for the neural network model, wherein the relevant parameters include at least the weight coefficients and biases of each network layer; Obtain a first training sample set, which includes multi-dimensional state data, charging decision data, and task completion effect evaluation data stored in a historical database; The neural network model is trained based on the training sample set, and the parameters of the neural network model are corrected until the loss function of the neural network model converges or reaches the preset number of training iterations.

7. The method for replenishing the energy of a UAV according to claim 6, characterized in that, The method further includes: Record the current multi-dimensional status information, decision results, and mission execution effects of the drone; When the judgment result of the charging demand based on the trained neural network model is inconsistent with the judgment result based on fuzzy logic, the judgment result based on the neural network model updates the charging threshold used by the fuzzy logic, the weight coefficient of the weighted fusion function, or the rules in the fuzzy rule base. Alternatively, the relevant parameters of the neural network model can be updated based on the judgment result of the fuzzy logic.

8. A drone energy replenishment system, employing the drone energy replenishment method as described in any one of claims 1-7, characterized in that, include: The acquisition module is used to acquire multi-dimensional state information of the target UAV. The multi-dimensional state information includes at least energy state information representing its energy level, task state information representing its mission execution urgency, and environmental state information representing the difficulty of its arrival at the charging platform. The demand determination module is used to output a quantitative value of charging demand based on the multi-dimensional state information of the target UAV, indexing a preset membership function library and a fuzzy rule library. The replenishment module is used to determine the charging mode of the target drone and the rendezvous coordinates of the target drone and the mobile charging platform based on real-time environmental status information when the quantified value of the charging demand is greater than or equal to a preset charging threshold, and to replenish the energy of the target drone at the rendezvous coordinates.

9. An electronic device, characterized in that, include: At least one processor; And a memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform the UAV energy replenishment method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to perform the UAV energy replenishment method according to any one of claims 1-7.

Images