Intelligent safety monitoring management method and system for charging state of unmanned aerial vehicle

Abstract

The invention provides an intelligent safety monitoring management method and system for the charging state of an unmanned aerial vehicle, and the method comprises the steps: firstly collecting the environment and battery data in the charging process of the unmanned aerial vehicle, and carrying out the abnormal cleaning processing, and obtaining effective charging data; based on the effective charging data, analyzing a coupling relationship of key parameters of the battery through an electrothermal coupling model, and continuously monitoring an internal safety state of the battery; meanwhile, an environment risk model is constructed in an off-line mode, the external environment safety state is monitored in combination with effective data, and electromagnetic interference parameters are extracted so that power frequency harmonic waves and burst pulse interference can be restrained in a targeted mode. And finally, integrating the internal and external safety states to construct a safety association diagram, outputting an overall risk level, and generating and executing a corresponding charging instruction. According to the invention, the effects of safe and accurate monitoring of charging of the unmanned aerial vehicle and effective suppression of external interference are realized.

Description

Technical Field

[0001] This invention belongs to the field of drone charging safety, specifically relating to an intelligent safety monitoring and management method and system for the charging status of drones. Background Technology

[0002] With the rapid development of drone technology, it has been widely used in many fields such as power line inspection, agricultural plant protection, logistics distribution and geographic surveying. Among them, the power line inspection scenario requires routine monitoring of facilities such as high-voltage lines and substations, which puts forward higher requirements for the endurance and charging safety of drones. In this scenario, drones are mostly charged through charging cabinets, and the safety and stability of the charging process directly determines the reliability of subsequent inspection tasks.

[0003] Current drone charging safety monitoring technology still faces numerous challenges. Existing methods often struggle to capture early, subtle anomalies when identifying safety hazards during charging, and lack the ability to deeply analyze the dynamic changes of complex risk factors. This leads to one-sided and delayed risk assessments, making it difficult to comprehensively and accurately reflect the true safety situation during charging. Furthermore, facing multi-source potential risks, existing technologies have limitations in integrating information from different dimensions to form a holistic safety judgment, resulting in less than ideal safety warnings and risk management. Therefore, the industry urgently needs an intelligent monitoring method capable of comprehensively sensing, deeply analyzing, and dynamically assessing charging risks to effectively improve the reliability and safety of drone charging processes. Summary of the Invention

[0004] This invention provides an intelligent safety monitoring and management method and system for the charging status of drones, in order to solve the problems of existing drone charging safety monitoring that make it difficult to capture early subtle anomalies, lack of dynamic analysis of complex risks, and poor safety early warning management.

[0005] In a first aspect, the present invention provides an intelligent safety monitoring and management method for the charging status of a drone, the method comprising the following steps: The system collects environmental and battery data during the drone's charging process as charging data, and performs anomaly cleaning on the charging data to obtain valid charging data. The system processes effective charging data through a pre-defined charging coupling model to continuously monitor the internal safety status of the battery during the drone charging process. An offline environmental risk model is built, and the external safety status of the external environment during the drone charging process is continuously monitored based on effective charging data and the environmental risk model. By combining the monitoring results of internal and external security status, a security correlation diagram is constructed, the overall risk level of drone charging is output, and corresponding charging instructions are generated and executed.

[0006] Optionally, collecting environmental and battery data during the drone's charging process as charging data, and performing anomaly cleaning on the charging data to obtain valid charging data includes the following steps: Collect environmental and battery data during the drone's charging process as charging data; The isolated forest algorithm is used to identify isolated points in the charging data that deviate from the normal distribution as potential outliers. The LOF algorithm is used to calculate the local outlier factor of the charging data. When the value of the local outlier factor exceeds a preset threshold, the corresponding charging data is marked as outlier data. The intersection of the potential abnormal data and the outlier data is taken as the abnormal data, and the effective charging data is obtained after removing the abnormal data.

[0007] Optionally, processing effective charging data through a preset charging coupling model and continuously monitoring the internal safety status of the battery during drone charging includes the following steps: Extract key battery parameters from valid charging data and input the key battery parameters into a preset charging coupling model; The charging coupling model is used to analyze the coupling relationship between the key battery parameters in real time, and the internal state parameters of the battery during the charging process of the drone are calculated based on the coupling relationship. The internal state parameters are compared with the preset safety state boundary to obtain the comparison result; Based on the charging coupling model and the comparison results, an online monitoring algorithm is executed to determine and output the internal safety status of the battery.

[0008] Optionally, the charging coupling model is an electrothermal coupling model. The step of analyzing the coupling relationship between the key battery parameters in real time through the charging coupling model and calculating the internal state parameters of the battery during the UAV charging process based on the coupling relationship includes the following steps: Identify the electrothermal and electrochemical parameters among the key parameters of the battery, and input the electrothermal and electrochemical parameters into the electrothermal and electrochemical modules of the electrothermal-chemical coupling model, respectively; The electrothermal coupling relationship between the electrothermal parameters is analyzed using the electrothermal module, and the electrochemical coupling relationship between the electrochemical parameters is analyzed using the electrochemical module. Based on the pre-set correlation model of the electrothermal coupling model, the electrothermal coupling relationship and the electrochemical coupling relationship are integrated to determine the interaction law between the electrothermal coupling relationship and the electrochemical coupling relationship; Based on the aforementioned interaction law, the internal state parameters of the battery during the drone charging process are deduced.

[0009] Optionally, integrating the electrothermal coupling relationship and the electrochemical coupling relationship based on the pre-defined correlation model of the electrothermal coupling model to determine the interaction law between the electrothermal coupling relationship and the electrochemical coupling relationship includes the following steps: Extract charging current and charging voltage data from key battery parameters, and determine the constant current charging stage and constant voltage charging stage of the UAV charging process based on the changing characteristics of the charging current and charging voltage data. The interaction coefficients of the electrothermal coupling relationship and the electrochemical coupling relationship are set based on the constant current charging stage and the constant voltage charging stage; The heat generation rate is analyzed using an electrothermal module, and the electrochemical reaction polarization is analyzed using an electrochemical module. The influence factor of the heat generation rate on the electrochemical reaction polarization and the correction factor of the electrochemical reaction polarization on the heat generation rate are calculated. The interaction coefficient, the influence factor, and the correction factor are input into a preset correlation model to establish a mapping relationship between electrothermal coupling and electrochemical coupling. Based on the mapping relationship, the interaction strength of the electrothermal coupling relationship and the electrochemical coupling relationship under the constant current charging stage and the constant voltage charging stage is quantified, and the interaction law of the electrothermal coupling relationship and the electrochemical coupling relationship is determined.

[0010] Optionally, an offline environmental risk model is constructed, and based on effective charging data, the external safety status of the external environment during the drone charging process is continuously monitored through the environmental risk model, including the following steps: An environmental risk model is constructed offline, and the model is used to identify the associated variables of the external environment during the charging process. The degree of impact of the associated variables on the safety of the charging process is quantified, and the dynamic evolution law of the interaction relationship and the degree of safety impact of the associated variables is integrated. Extract key environmental parameters from the valid charging data and input the key environmental parameters into the environmental risk model; Based on the environmental risk model, the key environmental parameters, and the interaction relationships, the safety impact of the external environment on the charging process is evaluated in real time, and the current external safety status is determined according to the dynamic evolution law. Based on the current external security status, the environmental risk model is used to predict the subsequent external security status of the external environment, and the evolution trajectory from the current external security status to the subsequent external security status is continuously monitored.

[0011] Optionally, the key environmental parameters include electromagnetic interference parameters, and the method further includes a step of suppressing electromagnetic interference: Electromagnetic interference parameters are extracted from the valid charging data, and power frequency harmonic interference characteristics and burst pulse interference characteristics are obtained from the electromagnetic interference parameters. An interference filtering strategy is constructed based on the characteristics of power frequency harmonic interference to dynamically suppress power frequency harmonic interference. Based on the characteristics of the sudden pulse interference, a monitoring frequency band is determined, and time-frequency analysis is performed on the electromagnetic interference within the monitoring frequency band to identify the sudden pulse interference and the pulse interval characteristics of the sudden pulse interference. When the sudden pulse interference is detected, the transmission signal of the charging command is instantly switched to the preset backup frequency band and the switching sequence is determined. Based on the results of the time-frequency analysis and the pulse interval characteristics, a timing control method is used to predict the interval pattern of the sudden pulse interference. The switching timing is then optimized based on the interval pattern to suppress electromagnetic interference.

[0012] Optionally, a safety correlation diagram is constructed by combining the monitoring results of internal and external safety status, outputting the overall risk level of drone charging, and generating and executing the corresponding charging instructions, including the following steps: A set of security statuses is formed by combining the monitoring results of internal and external security statuses; A safety association graph is constructed based on the set of safety states. The overall risk probability of drone charging is calculated through the safety association graph, and the overall risk probability is converted into an overall risk level. Match the charging strategy corresponding to the overall risk level, generate and execute the charging command corresponding to the charging strategy.

[0013] Optionally, constructing a security association graph based on the security state set includes the following steps: Identify the state variables in the set of safe states as root causal nodes, and define the optional charging strategies as intervention nodes; Construct a probabilistic causal path connecting the root causal node and the intervention node; Perform a hypothetical simulation on the optional charging strategy corresponding to any of the intervention nodes to simulate the propagation of charging safety risks in the probabilistic causal path and calculate the corresponding risk results; Integrate all the aforementioned risk outcomes to generate a decision matrix that correlates charging strategies with charging safety risk status, and use this decision matrix as the core decision-making basis for the safety correlation diagram.

[0014] In a second aspect, the present invention also provides an intelligent safety monitoring and management system for the charging status of unmanned aerial vehicles (UAVs), including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the intelligent safety monitoring and management method for the charging status of UAVs as described in the first aspect.

[0015] The beneficial effects of this invention are: This invention collects drone charging data and cleans it to obtain effective charging data, providing a high-quality data foundation for subsequent safety monitoring and effectively avoiding misjudgments caused by abnormal data interference. By using an electrothermal coupling model to analyze the coupling relationship of key battery parameters, this invention can accurately capture early and subtle abnormal signals inside the battery during the charging process, thus solving the problems of insufficient dynamic analysis of complex risks and one-sided and lagging assessments in traditional monitoring, and realizing in-depth and real-time monitoring of the battery's internal safety status. At the same time, by constructing an offline environmental risk model and working in conjunction with internal safety monitoring, it breaks the limitation of isolated internal and external safety risk assessments, forming a comprehensive safety perception of the drone charging scenario. For electromagnetic interference in power inspection scenarios, this invention extracts interference features and suppresses them in a targeted manner, further improving the stability and accuracy of safety monitoring. Finally, by integrating internal and external safety status to construct a safety correlation graph and outputting the overall risk level, generating and executing corresponding charging commands, it transforms scattered monitoring information into holistic safety decisions, effectively improving the poor performance of traditional technology in safety early warning and risk management, and comprehensively improving the safety, reliability, and intelligence level of the drone charging process. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the intelligent safety monitoring and management method for the charging status of a drone in one embodiment of this application.

[0017] Figure 2 This is a security association diagram and intelligent decision-making diagram of an intelligent safety monitoring and management method for the charging status of a drone in one embodiment of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0019] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0020] Figure 1 This is a flowchart illustrating an intelligent safety monitoring and management method for the charging status of a drone in one embodiment. It should be understood that, although... Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps. For example Figure 1 As shown, the intelligent safety monitoring and management method for the charging status of a drone disclosed in this invention specifically includes the following steps: S101. Collect environmental and battery data of the drone during the charging process as charging data, and perform abnormal data cleaning on the charging data to obtain valid charging data.

[0021] The process begins with collecting environmental and battery data during drone charging. Environmental data includes temperature, humidity, and electromagnetic interference (EMI) intensity. Temperature and humidity are collected using temperature and humidity sensors, while EMI intensity is collected using an EMI detector. Battery data includes charging current, charging voltage, battery temperature, and remaining battery power, collected by the drone's battery management system. The environmental and battery data are then integrated to form charging data, which undergoes anomaly cleaning. First, the Isolation Forest algorithm is used. This algorithm randomly selects feature dimensions of the charging data and randomly assigns thresholds within the selected dimension's value range to split the data samples. Isolated data deviating from the normal range is identified based on the average isolated path length of the samples in the forest model constructed by the algorithm. Next, the LOF algorithm is used. The number of neighboring samples is set, and the reachability distance between each data sample and its neighbors is calculated. The local reachability density of the sample is obtained based on the reachability distance. Local outliers are determined by comparing the local reachability density of the sample with that of its neighbors. Data with local outliers exceeding a preset threshold are identified as local outliers. Finally, the intersection of isolated data and local outliers is taken as abnormal data and removed from the charging data to obtain valid charging data.

[0022] S102. Process effective charging data through a preset charging coupling model to continuously monitor the internal safety status of the battery during the drone charging process.

[0023] First, a pre-defined charging coupling model is established, which can be an electrothermal coupling model. This model can correlate key parameters such as battery charging current, charging voltage, and battery temperature, establishing a dynamic mapping relationship between parameters to reflect the internal electrothermal changes of the battery. Then, the previously obtained valid charging data is input into this electrothermal coupling model. Based on a pre-defined parameter correlation algorithm, the input real-time charging current and charging voltage data are calculated, and the changes in battery temperature are simultaneously considered to analyze the heat generation and conduction process inside the battery, determining whether the changes in each parameter conform to the internal operating logic during normal charging. Simultaneously, the internal safety status of the battery during drone charging is monitored through the model's continuous output of internal state evaluation results. This includes monitoring for internal overcurrent caused by abnormally increased charging current, internal overvoltage caused by charging voltage exceeding the rated range, and signs of thermal runaway due to abnormally high battery temperature that is mismatched with current and voltage changes.

[0024] S103. Build an environmental risk model offline, and continuously monitor the external safety status of the external environment during the drone charging process based on effective charging data and the environmental risk model.

[0025] The process begins with the offline construction of an environmental risk model. The model is built around a safety overlay model, first defining a safe environmental zone. This zone encompasses a reasonable range of parameters including temperature, humidity, electromagnetic interference intensity, and air pressure, with each parameter exhibiting a gradual transition. Next, an impact overlay mechanism is established. During this mechanism, experimental data can be used to calibrate the degree of impact of a single environmental factor deviating from the safe zone. For example, when the temperature increases from its optimal value of 25℃, the impact value is calculated using a non-linear relationship. Factor interaction rules can be established to quantify combined effects; for example, when high temperature and high humidity are combined, the impact value is amplified by a synergistic effect, and the impact value is adjusted when a specific combination of factors produces a counteracting effect. An environmental impact load variable can be introduced to reflect the time effect. This variable accumulates with the intensity and duration of the impact and gradually diminishes after the environment returns to normal, thus completing the offline construction of the environmental risk model. Finally, real-time environmental data from the effective charging data is input into the constructed environmental risk model. Real-time environmental data includes the current charging environment's temperature, humidity, electromagnetic interference intensity, and air pressure. Based on the environmental safety zone boundary and influence superposition mechanism, an environmental risk model is used to calculate the total impact index of the current environment in real time, identify the main sources of influence, assess the risk level of factor combinations, and continuously output multi-dimensional environmental status results. Based on these results, the external safety status during the drone charging process is continuously monitored. If the total impact index exceeds a preset safety threshold or the combined risk level is high, the external safety status is determined to be abnormal.

[0026] S104. Combine the monitoring results of internal and external security status to construct a security correlation diagram, output the overall risk level of drone charging, and generate and execute the corresponding charging instructions.

[0027] First, a safety correlation diagram is constructed by combining monitoring results of internal and external safety states. The core nodes of the correlation diagram are determined, with key indicators from the internal safety state (including battery overcurrent, overvoltage, and thermal runaway precursor signals) and key information from the external safety state (including total impact index, main impact sources, and combined risk levels) designated as independent nodes. Based on historical charging safety cases and experimental data, the correlations between nodes are identified. By analyzing historical correlations and experimental verification data, the strength of interactions between nodes is determined, thus constructing a safety correlation diagram that reflects the mutual influence of internal and external risks. Next, the overall risk level of drone charging is output based on the safety correlation diagram. By counting the number of high-risk nodes in the correlation diagram and calculating the proportion of node pairs with high correlation strength, and referring to preset risk classification rules, the overall risk is divided into low, medium, and high levels. Finally, corresponding charging instructions are generated and executed according to the overall risk level. A continuous normal charging instruction is generated for low risk, a command to reduce charging current and increase monitoring frequency in real time is generated for medium risk, and a command to immediately stop charging, disconnect the charging circuit, and trigger an alarm is generated for high risk. This ensures that charging operations are accurately matched with the risk level.

[0028] In one implementation, collecting environmental and battery data from the drone during the charging process as charging data, and performing anomaly cleaning on the charging data to obtain valid charging data includes the following steps: Collect environmental and battery data during the drone's charging process as charging data; The isolated forest algorithm is used to identify isolated points in the charging data that deviate from the normal distribution as potential outliers. The LOF algorithm is used to calculate the local outlier factor of the charging data. When the value of the local outlier factor exceeds the preset threshold, the corresponding charging data is marked as outlier data. The intersection of potential outlier data and outlier data is taken as the outlier data. After removing the outlier data, the effective charging data is obtained.

[0029] In this embodiment, firstly, temperature, humidity, and electromagnetic interference (EMI) intensity data of the charging environment are collected. For example, the real-time temperature range is between -20°C and 50°C, the relative humidity range is between 10% and 95%, and the EMI intensity data is between 9 kHz and 30 MHz. In actual power line inspection scenarios, the background noise of EMI in a specific charging area is typically between 40 and 60 dBμV / m. Simultaneously, combined with external weather forecast API data, information such as rainfall, wind speed, and air pressure for the next 24 hours is obtained. Then, battery charging current, charging voltage, battery surface temperature, internal temperature gradient, remaining charge, and battery health status data are acquired. For example, the real-time charging current ranges from 0 amps to 100 amps, the charging voltage ranges from 20 volts to 60 volts, the battery surface temperature ranges from -10°C to 60°C, the internal temperature gradient ranges from 0°C to 10°C, and the battery health status can be expressed as 0% to 100%. The collected environmental and battery data are timestamped and integrated, for example, collected once per second and accompanied by a timestamp accurate to milliseconds, to form charging data.

[0030] Next, abnormal data cleaning is performed on the charging data. First, an isolated forest model consisting of multiple independent isolated trees is constructed using the Isolation Forest algorithm, with a preset number of 100 isolated trees. When constructing each isolated tree, a subsample is randomly selected from the charging data, for example, 60% of the charging data samples, and this subsample is recursively segmented until each data point is completely isolated. For each segmentation, a feature dimension is randomly selected, such as charging current, battery temperature, or electromagnetic interference intensity, and the segmentation point is randomly chosen between the maximum and minimum values ​​of that dimension. After all isolated trees are constructed, the average path length of each charging data point across all trees in the forest is calculated. Since abnormal data points are typically "few and different," their average path length is significantly shorter than that of normal data points. An anomaly score is calculated for each charging data point based on the average path length, and data points with anomaly scores exceeding a preset contamination ratio threshold are identified as potentially abnormal data. The preset contamination ratio threshold can be 0.1. For example, if the average path length of a certain charging current data point in 100 isolation trees is 5, and the anomaly score corresponding to this path length is significantly higher than the anomaly score threshold determined by the contamination ratio of 0.1, then this data point is identified as potentially anomalous data.

[0031] Subsequently, the LOF algorithm was used to perform density analysis on the charging data. First, the number of neighboring points, K, was set, with a preset limit of 20. For each data point in the charging data, its distance to the Kth nearest neighbor (K-distance) was calculated, and its K-distance neighborhood was determined. Based on the K-distance and neighboring points, the local reachability density of each data point was calculated. Finally, the local outlier factor (LOF) of the data point was obtained by calculating the ratio of the average local reachability density of other points within the K-distance neighborhood to the local reachability density of the data point itself. When the LLF of a data point is significantly greater than 1 and exceeds a preset outlier threshold, such as 1.5, it indicates that the density of that point is much lower than the density of its neighboring points, thus marking the data point as an outlier. For example, if a battery internal temperature data point has a LLF of 2.0, significantly greater than the preset outlier threshold of 1.5, it indicates that the density of that data point is much lower than the density of its neighboring points, and therefore, this data point is marked as an outlier.

[0032] Finally, the intersection of the potential outlier data set identified by the Isolation Forest algorithm and the outlier data set marked by the LOF algorithm is taken, and the data points in this intersection are defined as the outliers that need to be removed. All data points defined as outliers are removed from the charging data, and the remaining data set constitutes the valid charging data. This process implements a high-confidence outlier cleaning mechanism with dual algorithm verification. By combining two different detection principles—path length-based and density-based—the credibility of the data input into subsequent models is ensured.

[0033] In one implementation, processing effective charging data through a preset charging coupling model and continuously monitoring the internal safety status of the battery during drone charging includes the following steps: Extract key battery parameters from valid charging data and input them into a preset charging coupling model; The coupling relationship between key battery parameters is analyzed in real time using a charging coupling model, and the internal state parameters of the battery during the drone charging process are calculated based on the coupling relationship. The internal state parameters are compared with the preset safety state boundary to obtain the comparison results; Based on the charging coupling model and combined with the comparison results, an online monitoring algorithm is executed to determine and output the internal safety status of the battery.

[0034] In this embodiment, key battery parameters are first extracted from the valid charging data. These key parameters include charging current, charging voltage, battery surface temperature, battery internal temperature, battery internal impedance, state of charge (SOC), and state of health. These key battery parameters are then input into a preset charging coupling model.

[0035] Subsequently, the coupling relationships between key battery parameters are analyzed in real time using a charging coupling model. By solving a set of partial differential equations describing ion transport, electrochemical reaction kinetics, and heat generation and transfer within the battery, such as based on a P2D model or a lumped parameter model, accurate simulations of the battery's internal physicochemical processes are achieved. The analytical process involves correlating external inputs such as charging current and voltage with state variables within the battery, such as the electrochemical reaction rate, overpotential, heat generation rate, and temperature distribution. Based on these coupling relationships, internal state parameters of the battery during drone charging are calculated. These internal state parameters include the battery's internal temperature distribution, such as the cell center temperature and local hot spot temperatures, local current density distribution, electrolyte concentration distribution, solid electrolyte interface layer growth rate, and battery capacity decay rate.

[0036] Next, the calculated internal state parameters are compared in real time with preset safety state boundaries. These safety state boundaries include the maximum permissible cell center temperature (60 degrees Celsius), the maximum permissible local hot spot temperature (70 degrees Celsius), the maximum solid electrolyte interface layer growth rate (5% thickness increase per year), and the minimum healthy state (80%). The comparison determines whether the internal state parameters exceed the safety boundaries. For example, if the calculated cell center temperature is 62 degrees Celsius, it exceeds the safety boundary of the maximum permissible cell center temperature of 60 degrees Celsius.

[0037] Finally, based on the charging coupling model and comparison results, an online monitoring algorithm is executed. This online monitoring algorithm can be an extended Kalman filter or a particle filter algorithm. The charging coupling model is used to predict the internal state, and the prediction results are corrected by combining real-time measurement data. Simultaneously, the comparison results are used as an anomaly signal input. The internal safety state of the battery is determined and output, specifying a safety level such as "Normal," "Warning: Local Overheating," or "Alarm: Precursor to Thermal Runaway." For example, if the online monitoring algorithm detects that the calculated local hotspot temperature exceeds 65 degrees Celsius for five consecutive minutes, and the comparison results show that it exceeds the safety boundary, then a "Warning: Local Overheating" command is output.

[0038] In one implementation, the charging coupling model is an electrothermal coupling model. The coupling relationship between key battery parameters is analyzed in real time using this model. Based on this coupling relationship, the internal state parameters of the battery during the drone charging process are calculated, including the following steps: Identify the electrothermal and electrochemical parameters among the key parameters of the battery, and input the electrothermal and electrochemical parameters into the electrothermal and electrochemical modules of the electrothermal-chemical coupling model, respectively. The electrothermal coupling relationship between electrothermal parameters is analyzed using the electrothermal module, and the electrochemical coupling relationship between electrochemical parameters is analyzed using the electrochemical module. Based on the pre-defined correlation model of the electrothermal coupling model, the electrothermal coupling relationship and the electrochemical coupling relationship are integrated to determine the interaction law between the electrothermal coupling relationship and the electrochemical coupling relationship. The internal state parameters of the battery during the drone charging process are calculated based on the interaction law.

[0039] In this embodiment, the electrothermal and electrochemical parameters among the key battery parameters are first identified. Electrothermal parameters include charging current, charging voltage, internal battery temperature, surface battery temperature, and total heat generation rate. Electrochemical parameters include electrode current density, exchange current density, electrolyte ion concentration, and overpotential. The electrothermal parameters are input into the electrothermal module of the electrothermal-chemical coupling model, and the electrochemical parameters are input into the electrochemical module.

[0040] Next, the electrothermal coupling relationship between electrothermal parameters is analyzed using an electrothermal module. During this process, the real-time acquired battery surface temperature serves as a key boundary condition or calibration signal for the model, improving the accuracy of the battery's internal temperature estimation. The heat conduction equation is solved using the electrothermal module to describe the temperature distribution inside the battery.

[0041] in, For density, For specific heat capacity, Let t be the internal temperature of the battery, t be time, and k be the thermal conductivity. The total heat generation rate, which includes ohmic heat, reaction heat, and entropy change heat, is calculated from parameters such as the input charging current and charging voltage.

[0042] Simultaneously, the electrochemical coupling relationships between electrochemical parameters were analyzed using the electrochemical module. The Butler-Wolmer equation and the mass conservation equation were solved using the electrochemical module to describe the reaction kinetics *i* at the electrode interface and the lithium-ion concentration in the electrolyte. Distribution:

[0043] Where i is the electrode current density. For exchange current density, The anode charge transfer coefficient, Where is the cathode charge transfer coefficient, and F is the Faraday constant. This is an overpotential, where R is the ideal gas constant. This is the absolute temperature, which is the internal temperature of the battery calculated and fed back by the heating module.

[0044] Next, based on the pre-defined correlation model of the electrothermal coupling model, the electrothermal coupling relationship and the electrochemical coupling relationship are integrated to determine the interaction law between the electrothermal coupling relationship and the electrochemical coupling relationship. The pre-defined correlation model can be a set of coupled partial differential equations, which will incorporate the heat generation rate calculated by the electrochemical module. As input to the heating module, it also includes the battery's internal temperature calculated by the heating module. As the exchange current density in the electrochemical module With the input of the diffusion coefficient D(T), a bidirectional iterative solution for the electrothermal and electrochemical processes is achieved.

[0045] Finally, based on the interaction law, the internal state parameters of the battery during the drone charging process are calculated. These internal state parameters include the battery's internal temperature, local current density, electrolyte concentration, and the thickness of the solid electrolyte interface layer.

[0046] In one embodiment, the process of integrating electrothermal coupling and electrochemical coupling relationships based on a pre-defined correlation model of the electrothermal coupling model to determine the interaction law between the electrothermal coupling and electrochemical coupling relationships includes the following steps: Extract charging current and charging voltage data from key battery parameters, and determine the constant current charging stage and constant voltage charging stage of the UAV charging process based on the changing characteristics of the charging current and charging voltage data. The interaction coefficients of the electrothermal coupling relationship and the electrochemical coupling relationship are set based on the constant current charging stage and the constant voltage charging stage; The heat generation rate is analyzed using the electrothermal module, and the polarization of the electrochemical reaction is analyzed using the electrochemical module. The influence factor of the heat generation rate on the polarization of the electrochemical reaction and the correction factor of the polarization of the electrochemical reaction on the heat generation rate are calculated. The interaction coefficient, influence factor, and correction factor are input into a pre-defined correlation model to establish a mapping relationship between electrothermal coupling and electrochemical coupling. Based on the mapping relationship, the interaction strength of the electrothermal coupling and electrochemical coupling relationships under the constant current charging stage and constant voltage charging stage is quantified, and the interaction law of the electrothermal coupling and electrochemical coupling relationships is determined.

[0047] In this embodiment, charging current and charging voltage data from the key battery parameters are first extracted. Based on the changing characteristics of the charging current and charging voltage data, such as the charging current remaining constant while the charging voltage gradually increases, or the charging current gradually decreasing while the charging voltage remains constant, the constant current charging stage and constant voltage charging stage of the drone charging process are determined. For example, when the charging current remains constant at 0.2C (C is the battery capacity) and the voltage of a single cell rises from 3.0V to 4.2V, it is determined to be the constant current charging stage; when the voltage of a single cell remains constant at 4.2V and the charging current gradually decreases from 0.2C to 0.05C, it is determined to be the constant voltage charging stage. Here, the voltage value specifically refers to the voltage of a single cell in the battery pack, not the terminal voltage of the entire battery pack. The overall voltage range of the battery pack (e.g., 20V to 60V) is composed of multiple cells connected in series.

[0048] Subsequently, interaction coefficients for the electrothermal and electrochemical coupling relationships were established based on the constant current charging and constant voltage charging stages. In the constant current charging stage, due to the constant current, the contributions of Joule heating and reaction heat are relatively stable, and the electrochemical reaction is dominant. In the constant voltage charging stage, due to the decrease in current, heat generation decreases, and the electrochemical reaction rate is limited by voltage. Therefore, the interaction coefficient for the constant current charging stage was set as follows: and The interaction coefficient during the constant voltage charging phase is and The coefficients are obtained through experimental calibration and training with historical data, for example... =0.7, =0.3.

[0049] Next, the heat generation rate was analyzed using the electrothermal module. Heat generation rate Covering Joule heating, electrochemical reaction heat, and entropy change heat:

[0050] Where I is the charging current, This refers to the battery's internal resistance. This represents the total overpotential of the battery (where V is the actual battery voltage). (Battery equilibrium potential). This refers to the internal temperature of the battery. This is the rate of change of the battery's equilibrium potential with temperature (i.e., the temperature coefficient). This formula calculates the total heat generation power inside the battery (unit: watts). If the heat conduction equation in the electrothermal module requires the volumetric heat generation rate (unit: W / m³), then... 3 If the total heat generation power is less than the effective volume of the battery, then the total heat generation power needs to be divided by the effective volume of the battery.

[0051] Subsequently, the electrochemical reaction polarization was analyzed using the electrochemical module. Here, the polarization ( ) refers to the total overpotential of the battery, including activation polarization ( ) and concentration polarization ( The two main components are activation polarization and concentration polarization. Activation polarization is the overpotential caused by charge transfer resistance in the electrode reaction kinetics, while concentration polarization is the overpotential caused by the concentration gradient resulting from the limitation of ion transport rates in the electrolyte. The influence of the heat generation rate on the polarization degree of the electrochemical reaction is calculated. Correction factor for the rate of heat generation by electrochemical reaction polarization Impact Factor Quantifying the effect of changes in unit heat generation rate on electrochemical reaction polarization, correction factor The influence of a unit change in electrochemical reaction polarization on the rate of heat generation is quantified. The influence factor and correction factor can be calculated using the following formula:

[0052] in This represents the change in polarization of the electrochemical reaction. This represents the change in the rate of heat generation, which is obtained by calculating the model's response under a specific perturbation.

[0053] Subsequently, the interaction coefficient, influencing factor, and correction factor are input into a pre-defined correlation model. The pre-defined correlation model can be a nonlinear mapping model based on a neural network or a multinomial regression model, establishing a mapping relationship between electrothermal coupling and electrochemical coupling. This mapping relationship can be expressed as:

[0054] in For mapping functions, , , , These are the model weight parameters, obtained through training with historical charging data.

[0055] Finally, based on the mapping relationship, the interaction strength of the electrothermal coupling and electrochemical coupling relationships under the constant current charging and constant voltage charging stages is quantified, and the interaction law of the electrothermal coupling and electrochemical coupling relationships is determined. For example, in the constant current charging stage, if... A higher value indicates that heat has a significant impact on the electrochemical reaction, thus indicating a higher interaction strength.

[0056] In one implementation, building an environmental risk model offline and continuously monitoring the external safety status of the external environment during the drone charging process based on effective charging data includes the following steps: An environmental risk model is built offline, and the model is used to identify the associated variables of the external environment during the charging process. The degree of safety impact of related variables on the charging process is quantified, and the dynamic evolution law of the interaction relationship and safety impact of related variables is integrated. Extract key environmental parameters from valid charging data and input these parameters into the environmental risk model. The system assesses the safety impact of the external environment on the charging process in real time based on environmental risk models, key environmental parameters, and interaction relationships, and determines the current external safety status based on dynamic evolution patterns. Based on the current external security status, the environmental risk model is used to predict the subsequent external security status, and the evolution trajectory from the current external security status to the subsequent external security status is continuously monitored.

[0057] In this embodiment, firstly, an environmental risk model is constructed offline. This environmental risk model can be a safety overlay model, which identifies related variables of the external environment during charging, including parameters such as ambient temperature, ambient humidity, electromagnetic interference intensity, and air pressure. The safety overlay model defines a multi-dimensional safe charging niche space, containing reasonable parameter ranges for related variables. The boundaries of these parameter ranges exhibit a gradual transition characteristic. For example, a safe charging niche can be defined as: temperature between 5°C and 35°C, humidity between 20% and 70%, electromagnetic interference intensity below 60 dBμV / m, and air pressure between 95 kPa and 105 kPa.

[0058] Next, the impact of correlated variables on the safety of the charging process is quantified. The quantification process treats each environmental factor deviating from the optimal safe charging niche as a "stress source," with the intensity of the stress source calculated using a nonlinear relationship. For example, when the ambient temperature rises from the optimal value of 25 degrees Celsius to 38 degrees Celsius, its stress intensity will increase according to a preset nonlinear function, which can be calibrated through physical models or experimental data. Simultaneously, the interaction relationships of correlated variables are integrated. An interaction matrix or function is established to quantify the nonlinear interactions when multiple environmental stress sources coexist, including synergistic and antagonistic effects. For example, when high temperature and high humidity are superimposed, the combined risk will be far greater than the sum of the individual risks of the two, reflecting a synergistic effect. Furthermore, the dynamic evolution of the safety impact is integrated. An environmental stress load variable is introduced, which accumulates with the intensity and duration of the impact and gradually diminishes after the environment returns to normal. For example, the cumulative stress load of moderate-intensity electromagnetic interference lasting two hours will exceed that of strong electromagnetic interference lasting one minute.

[0059] Subsequently, key environmental parameters are extracted from the valid charging data. These parameters include real-time collected and cleaned data on the charging environment, such as temperature, humidity, electromagnetic interference intensity, and air pressure. These parameters are then input into the constructed safety overlay model. For example, the current ambient temperature (35 degrees Celsius), relative humidity (70%), electromagnetic interference intensity (55 dBμV / m), and air pressure (101 kPa) are input into the safety overlay model.

[0060] Next, the safety impact of the external environment on the charging process is assessed in real time based on the safety superposition model, key environmental parameters, and interaction relationships. Utilizing the built-in assessment mechanism of the safety superposition model, combined with the input real-time key environmental parameters and integrated interaction relationships, the total stress index of the current environment is calculated in real time.

[0061] in, The total stress index is N, where N is the number of related variables. Let i be the current value of the i-th associated variable. For a single stress intensity function, Let i be the interaction function between the related variables i and j. To accumulate environmental stress loads, This is the environmental stress load function.

[0062] The security overlay model determines the current external security status based on dynamic evolution patterns. For example, it converts the total stress index into an external risk index, which ranges from 0 to 100. An external risk index below 30 is considered "low risk," between 30 and 60 is "medium risk," and above 60 is "high risk." For instance, if the assessment shows a current total stress index of 0.6 and a corresponding external risk index of 45, then the current external security status is determined to be "medium risk."

[0063] Finally, based on the current external security status, the environmental risk model is used to predict the subsequent external security status. The security overlay model utilizes its built-in state transition probability or time series prediction module, combined with external weather forecasts and other information, to predict the external environmental status over the next few hours or charging cycles. For example, if the current external security status is "medium risk," the model predicts a 70% probability of remaining "medium risk," a 20% probability of transitioning to "high risk," and a 10% probability of transitioning to "low risk" in the next hour. The evolution trajectory from the current external security status to subsequent external security status is continuously monitored; for example, the changing trends of the external risk index and major stressors are recorded and visualized over future charging cycles.

[0064] In one embodiment, the key environmental parameters include electromagnetic interference parameters, and the method further includes a step of suppressing electromagnetic interference: Extract electromagnetic interference parameters from valid charging data, and obtain power frequency harmonic interference characteristics and burst pulse interference characteristics from electromagnetic interference parameters; An interference filtering strategy is constructed based on the characteristics of power frequency harmonic interference to dynamically suppress power frequency harmonic interference; The monitoring frequency band is determined based on the characteristics of sudden pulse interference. Time-frequency analysis is performed on the electromagnetic interference within the monitoring frequency band to identify sudden pulse interference and the pulse interval characteristics of sudden pulse interference. When sudden pulse-type interference is detected, the transmission signal of the charging command is instantly switched to the preset backup frequency band and the switching sequence is determined. Combining the results of time-frequency analysis and pulse interval characteristics, a timing control method is used to predict the interval pattern of sudden pulse interference. Based on the interval pattern, the switching timing is optimized to suppress electromagnetic interference.

[0065] In this embodiment, firstly, electromagnetic interference parameters are extracted from the valid charging data, and power frequency harmonic interference characteristics and burst pulse interference characteristics are obtained from these parameters. Electromagnetic interference parameters include real-time electromagnetic field strength, spectral distribution, and noise power density on the charging signal transmission link, such as spectral data in the 0.1 MHz to 1 GHz frequency band. Power frequency harmonic interference characteristics are obtained from the electromagnetic interference parameters through fast Fourier transform analysis or power spectral density analysis. These characteristics include the frequency, amplitude, and phase of the 50 Hz or 60 Hz fundamental wave and its integer multiples of harmonics. For example, harmonic components at frequencies such as 50 Hz, 150 Hz, and 250 Hz are identified, with amplitudes exceeding a preset threshold, such as 10 dBμV. Simultaneously, burst pulse interference characteristics are obtained through transient analysis or wavelet transform of the time-domain signal. These characteristics include the pulse's start time, duration, peak amplitude, and repetition frequency. For example, sharp pulses with durations in the microsecond range and peak amplitudes reaching 80 dBμV / m are identified.

[0066] Next, an interference filtering strategy is constructed based on the characteristics of power frequency harmonic interference to dynamically suppress it. This strategy can be an adaptive notch filter or a digital bandpass filter. Based on the acquired power frequency harmonic interference characteristics, the center frequency, bandwidth, and notch depth of the filter are adjusted in real time to dynamically match changes in interference. For example, if 50 Hz and 150 Hz harmonic interference are identified, a dual notch filter is constructed with center frequencies set to 50 Hz and 150 Hz respectively, and the notch depth is dynamically adjusted according to the harmonic amplitude to ensure that interference is suppressed while minimizing the impact on the normal portion of the charging command transmission signal.

[0067] Subsequently, the monitoring frequency band is determined based on the characteristics of burst-pulse interference. Time-frequency analysis is performed on the electromagnetic interference within the monitoring frequency band to identify burst-pulse interference and its pulse interval characteristics. Based on the spectral characteristics of burst-pulse interference, such as its broadband nature, a monitoring frequency band covering the main interference energy is determined, for example, the 2 MHz to 30 MHz band. Time-frequency analysis is performed on the electromagnetic interference signal within the monitoring frequency band, which can be done using short-time Fourier transform or wavelet transform. Through time-frequency analysis, the instantaneous concentrated region of signal energy on the time-frequency plane is identified, thereby identifying burst-pulse interference and extracting its pulse interval characteristics, including the time interval between pulses, pulse width, and periodicity of the pulse sequence. For example, through short-time Fourier transform analysis, a broadband signal with concentrated energy at a specific time point is identified, and its repetition period is measured to be 10 milliseconds.

[0068] Next, upon detecting sudden pulse-type interference, the charging command transmission signal is instantaneously switched to a preset backup frequency band, and the switching sequence is determined. Instantaneous switching is achieved through frequency hopping technology. The backup frequency band is one that does not overlap with the main transmission band and has a low interference level; for example, switching the transmission band from 2.4 GHz to 5.8 GHz. Determining the switching sequence involves initiating the switching mechanism with a millisecond-level response speed after detecting pulse interference, ensuring the integrity of the transmitted signal during the duration of the interference. For example, after detecting pulse interference, the system completes the frequency switching within 100 microseconds, transmitting the charging command to the backup frequency band.

[0069] Finally, combining the results of time-frequency analysis and pulse interval characteristics, a timing control method is used to predict the interval pattern of sudden pulse-type interference. Based on the interval pattern, the switching timing is optimized to suppress electromagnetic interference. The timing control method can be a prediction algorithm based on Kalman filtering or an autoregressive moving average model. Based on the pulse interval characteristics identified by time-frequency analysis, such as the periodicity or random distribution of pulses, the occurrence time of future pulses is predicted. Based on the predicted interval pattern, the switching timing of the charging command transmission signal is optimized to achieve pre-adaptive switching, that is, completing the frequency switching before the predicted pulse is about to occur. For example, if the pulse is predicted to occur in the next 5 milliseconds, the system switches to the backup frequency band 2 milliseconds in advance, and then switches back to the main frequency band after the pulse ends.

[0070] In one implementation, a safety correlation diagram is constructed by combining the monitoring results of internal and external safety statuses, outputting the overall risk level of drone charging, and generating and executing corresponding charging instructions, including the following steps: A set of security statuses is formed by combining the monitoring results of internal and external security statuses; A safety association graph is constructed based on a set of safety states. The overall risk probability of drone charging is calculated through the safety association graph, and the overall risk probability is converted into an overall risk level. Match the charging strategy with the overall risk level, and generate and execute the charging instructions corresponding to the charging strategy.

[0071] In this implementation, firstly, a safety state set is formed by combining the monitoring results of internal and external safety states. The monitoring results of internal safety states include specific judgments such as battery overcurrent state, overvoltage state, and precursor signals of thermal runaway. The monitoring results of external safety states include specific assessments such as the total stress index of the external environment, major stress sources, and combined risk levels. These internal and external monitoring results are integrated to form a multi-dimensional safety state set containing all relevant safety information. For example, if the internal monitoring outputs "Warning: Local overheating" and the external monitoring outputs "Medium risk: High temperature and high humidity," then this information is included in the safety state set.

[0072] Next, a safety association graph is constructed based on the set of safe states. This graph can be a Bayesian network or a causal graph model. The overall risk probability of drone charging is calculated using the safety association graph, and this overall risk probability is converted into an overall risk level. The safety association graph uses various indicators in the set of safe states as nodes, and the connections between nodes represent causal relationships or association strength. For example, the probability of a charging accident occurring under the current set of safe states is calculated using the conditional probability distribution in a Bayesian network. The overall risk probability can be represented as a value from 0% to 100%. The overall risk probability is then converted into an overall risk level; for example, a risk probability less than 10% is considered "low risk," 10% to 30% is "medium risk," and greater than 30% is "high risk."

[0073] Finally, a charging strategy corresponding to the overall risk level is matched, and charging instructions corresponding to the strategy are generated and executed. The preset charging strategy library contains charging operation schemes for different risk levels. For example, if the overall risk level is "low risk," the "continuous normal charging" strategy is matched, generating and executing instructions to maintain the current charging current and voltage. If it is "medium risk," the "reduce charging current and increase monitoring frequency in real time" strategy is matched, generating and executing instructions to reduce the charging current by 20% and perform a status check every minute. If it is "high risk," the "immediately stop charging, disconnect the charging circuit, and trigger an alarm" strategy is matched, generating and executing instructions to disconnect the charging power supply and issue an audible and visual alarm signal.

[0074] In one implementation, constructing a security association graph based on a set of security states includes the following steps: Identify the state variables in the set of safe states as root causal nodes, and define the optional charging strategies as intervention nodes; Construct a probabilistic causal pathway connecting the root causal node and the intervention node; Perform a hypothetical simulation on the optional charging strategies corresponding to any intervention node, simulate the propagation of charging safety risks in probabilistic causal pathways, and calculate the corresponding risk results; All risk outcomes are integrated to generate a decision matrix that correlates charging strategies with charging safety risk status, and this decision matrix serves as the core decision-making basis for the safety correlation diagram.

[0075] In this embodiment, state variables in the set of safe states are first identified as root cause nodes, and optional charging strategies are defined as intervention nodes. Root cause nodes include battery overcurrent state, battery internal temperature, electromagnetic interference intensity, and ambient humidity. Intervention nodes include optional charging strategies such as "normal charging," "reduced current charging," "paused charging," and "stopped charging." For example, "excessively high battery internal temperature" is identified as a root cause node, and "reduced charging current" is defined as an intervention node.

[0076] Next, a probabilistic causal path connecting the root causal node and the intervention node is constructed. This probabilistic causal path is represented by a directed acyclic graph, where edges between nodes represent causal relationships and are accompanied by conditional probability distributions. These conditional probability distributions are trained using historical data and calibrated using expert knowledge. For example, a path is constructed from the intervention node "reduce charging current" to the root causal node "battery internal temperature," and the conditional probability P(battery internal temperature | reduce charging current) is calibrated, representing the probability of a change in battery internal temperature after reducing the charging current.

[0077] Subsequently, a hypothetical simulation is performed on the optional charging strategies corresponding to any intervention node. Hypothetical simulation is essentially counterfactual inference, simulating the propagation of charging safety risks in a probabilistic causal path and calculating the corresponding risk outcomes. For example, when the current internal battery temperature is already at a high level, a hypothetical simulation is performed on the intervention node of "reducing the charging current." The simulation simulates how the states of root causal nodes such as internal battery temperature, overvoltage risk, and precursors to thermal runaway change after this strategy is implemented, and calculates the overall risk probability after the simulation.

[0078] Finally, all risk outcomes are integrated to generate a decision matrix that correlates charging strategies with charging safety risk states, and this decision matrix serves as the core decision-making basis for the safety correlation graph. The decision matrix is ​​structured with optional charging strategies as rows and charging safety risk states (e.g., overall risk probability) as columns. The elements in the matrix represent the risk outcomes after executing the corresponding strategy. For example, the decision matrix might show that, under the current safety state, executing the "reduce charging current" strategy reduces the overall risk probability from 35% to 15%, while executing the "normal charging" strategy maintains it at 35%.

[0079] In one implementation, a hypothetical simulation is performed on the optional charging strategy corresponding to any intervention node to simulate the propagation of charging safety risks in a probabilistic causal path, and the corresponding risk results are calculated by the following steps: Based on the current safety status and available charging strategies, the risk propagation factor is dynamically calculated. By utilizing risk propagation factors, the conditional probability distribution between root causal nodes in probabilistic causal pathways can be adjusted in real time. Based on the adjusted conditional probability distribution, the propagation of charging safety risks in probabilistic causal pathways is simulated, and the corresponding risk outcomes are calculated.

[0080] In this embodiment, firstly, based on the safety status and available charging strategies, a risk propagation factor is dynamically calculated using a pre-defined effect model. This considers battery health status, external environmental risk levels, and historical charging data to quantify the suppression or enhancement effect of specific available charging strategies on risk propagation paths between different root-cause nodes. For example, when the battery health status is low, the suppression effect of implementing the "reduced charging current" strategy on the propagation of the "battery overheating" risk to "precursor to thermal runaway" will be lower than that of a healthy battery, and the effect model will output a lower risk propagation factor. Risk Propagation Factor By using optional charging strategies and safe status The eigenvectors are linearly combined and normalized to obtain the eigenvalues, which range from 0 to 1, representing the optional charging strategies. In a safe state The degree of suppression of risk propagation. The linear combination can be expressed as:

[0081] Where L is the result of a linear combination. This is the weight matrix. For optional charging strategies and safe status The feature vector includes strategy type, battery health status indicators, environmental risk index, etc. This is the bias vector. Based on the risk propagation factor. The function maps L to the range of 0 to 1 using a predefined function, such as a linear scaling or non-linear mapping function.

[0082] Next, the calculated risk propagation factor is applied to the conditional law between root causal nodes in the probabilistic causal pathway. For example, for the root causal node... arrive conditional probability When implementing optional charging strategies and safe status Below, its adjusted conditional probability It can be represented as:

[0083] in, The original conditional probability represents the probability of risk propagation without intervention. Risk propagation factor, representing optional charging strategies In a safe state Facing risks spread to The degree of inhibition.

[0084] Finally, the propagation of charging safety risks in a probabilistic causal path is simulated using the Markov chain Monte Carlo method or belief propagation algorithm. Risk propagation is then extrapolated within the adjusted probabilistic causal path, calculating the final risk probability of each root causal node and the overall risk probability. For example, in the adjusted causal path, the probability of the risk of "battery overcurrent" propagating to "precursor to thermal runaway" is simulated, and the overall risk probability after implementing the "reduced charging current" strategy is calculated to be 15%, lower than the 35% in the original simulation.

[0085] The processor can be a central processing unit (CPU). Of course, depending on the actual use, it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc., and this application does not limit it.

[0086] The memory can be an internal storage unit of a computer device, such as a hard disk or RAM, or an external storage device, such as a plug-in hard disk, smart memory card (SMC), secure digital card (SD), or flash memory card (FC) provided on the computer device. Furthermore, the memory can be a combination of internal storage units and external storage devices of a computer device. The memory is used to store computer programs and other programs and data required by the computer device. The memory can also be used to temporarily store data that has been output or will be output. This application does not limit this.

[0087] The present invention also discloses a computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the intelligent safety monitoring and management method for the charging status of a drone as described in any of the above embodiments.

[0088] The computer program can be stored in a machine-readable medium. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or certain middleware. The machine-readable medium includes any entity or device capable of carrying computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the machine-readable medium includes, but is not limited to, the above-mentioned components.

[0089] The intelligent safety monitoring and management method for the charging status of the UAV in the above embodiments is stored in the computer-readable storage medium and loaded and executed on the processor to facilitate the storage and application of the above method.

[0090] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of protection of this application is limited to these examples; under the concept of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of one or more embodiments of this application as described above, which are not provided in detail for the sake of brevity.

[0091] One or more embodiments in this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments in this application should be included within the protection scope of this application.

Claims

1. A method for intelligent safety monitoring and management of the charging status of a drone, characterized in that, Includes the following steps: The system collects environmental and battery data during the drone's charging process as charging data, and performs anomaly cleaning on the charging data to obtain valid charging data. The system processes effective charging data through a pre-defined charging coupling model to continuously monitor the internal safety status of the battery during the drone charging process. An offline environmental risk model is built, and the external safety status of the external environment during the drone charging process is continuously monitored based on effective charging data and the environmental risk model. By combining the monitoring results of internal and external security status, a security correlation diagram is constructed, the overall risk level of drone charging is output, and corresponding charging instructions are generated and executed.

2. The intelligent safety monitoring and management method for the charging status of a drone according to claim 1, characterized in that, The process of collecting environmental and battery data from the drone during charging as charging data, and cleaning outlier data to obtain valid charging data, includes the following steps: Collect environmental and battery data during the drone's charging process as charging data; The isolated forest algorithm is used to identify isolated points in the charging data that deviate from the normal distribution as potential outliers. The LOF algorithm is used to calculate the local outlier factor of the charging data. When the value of the local outlier factor exceeds a preset threshold, the corresponding charging data is marked as outlier data. The intersection of the potential abnormal data and the outlier data is taken as the abnormal data, and the effective charging data is obtained after removing the abnormal data.

3. The intelligent safety monitoring and management method for the charging status of a drone according to claim 1, characterized in that, The process of processing effective charging data through a preset charging coupling model and continuously monitoring the internal safety status of the battery during the drone charging process includes the following steps: Extract key battery parameters from valid charging data and input the key battery parameters into a preset charging coupling model; The charging coupling model is used to analyze the coupling relationship between the key battery parameters in real time, and the internal state parameters of the battery during the charging process of the drone are calculated based on the coupling relationship. The internal state parameters are compared with the preset safety state boundary to obtain the comparison result; Based on the charging coupling model and the comparison results, an online monitoring algorithm is executed to determine and output the internal safety status of the battery.

4. The intelligent safety monitoring and management method for the charging status of a drone according to claim 3, characterized in that, The charging coupling model is an electrothermal coupling model. The step of analyzing the coupling relationship between the key battery parameters in real time using the charging coupling model and calculating the internal state parameters of the battery during the drone charging process based on the coupling relationship includes the following steps: Identify the electrothermal and electrochemical parameters among the key parameters of the battery, and input the electrothermal and electrochemical parameters into the electrothermal and electrochemical modules of the electrothermal-chemical coupling model, respectively; The electrothermal coupling relationship between the electrothermal parameters is analyzed using the electrothermal module, and the electrochemical coupling relationship between the electrochemical parameters is analyzed using the electrochemical module. Based on the pre-set correlation model of the electrothermal coupling model, the electrothermal coupling relationship and the electrochemical coupling relationship are integrated to determine the interaction law between the electrothermal coupling relationship and the electrochemical coupling relationship; Based on the aforementioned interaction law, the internal state parameters of the battery during the drone charging process are deduced.

5. The intelligent safety monitoring and management method for the charging status of a drone according to claim 4, characterized in that, The process of integrating the electrothermal coupling relationship and the electrochemical coupling relationship based on the preset correlation model of the electrothermal coupling model, and determining the interaction law of the electrothermal coupling relationship and the electrochemical coupling relationship, includes the following steps: Extract charging current and charging voltage data from key battery parameters, and determine the constant current charging stage and constant voltage charging stage of the UAV charging process based on the changing characteristics of the charging current and charging voltage data. The interaction coefficients of the electrothermal coupling relationship and the electrochemical coupling relationship are set based on the constant current charging stage and the constant voltage charging stage; The heat generation rate is analyzed using an electrothermal module, and the electrochemical reaction polarization is analyzed using an electrochemical module. The influence factor of the heat generation rate on the electrochemical reaction polarization and the correction factor of the electrochemical reaction polarization on the heat generation rate are calculated. The interaction coefficient, the influence factor, and the correction factor are input into a preset correlation model to establish a mapping relationship between electrothermal coupling and electrochemical coupling. Based on the mapping relationship, the interaction strength of the electrothermal coupling relationship and the electrochemical coupling relationship under the constant current charging stage and the constant voltage charging stage is quantified, and the interaction law of the electrothermal coupling relationship and the electrochemical coupling relationship is determined.

6. The intelligent safety monitoring and management method for the charging status of a drone according to claim 1, characterized in that, The offline construction of the environmental risk model, based on effective charging data and continuously monitoring the external safety status of the external environment during the drone charging process through the environmental risk model, includes the following steps: An environmental risk model is constructed offline, and the model is used to identify the associated variables of the external environment during the charging process. The degree of impact of the associated variables on the safety of the charging process is quantified, and the dynamic evolution law of the interaction relationship and the degree of safety impact of the associated variables is integrated. Extract key environmental parameters from the valid charging data and input the key environmental parameters into the environmental risk model; Based on the environmental risk model, the key environmental parameters, and the interaction relationships, the safety impact of the external environment on the charging process is evaluated in real time, and the current external safety status is determined according to the dynamic evolution law. Based on the current external security status, the environmental risk model is used to predict the subsequent external security status of the external environment, and the evolution trajectory from the current external security status to the subsequent external security status is continuously monitored.

7. The intelligent safety monitoring and management method for the charging status of a drone according to claim 6, characterized in that, The key environmental parameters include electromagnetic interference parameters, and the method further includes a step to suppress electromagnetic interference: Electromagnetic interference parameters are extracted from the valid charging data, and power frequency harmonic interference characteristics and burst pulse interference characteristics are obtained from the electromagnetic interference parameters. An interference filtering strategy is constructed based on the characteristics of power frequency harmonic interference to dynamically suppress power frequency harmonic interference. Based on the characteristics of the sudden pulse interference, a monitoring frequency band is determined, and time-frequency analysis is performed on the electromagnetic interference within the monitoring frequency band to identify the sudden pulse interference and the pulse interval characteristics of the sudden pulse interference. When the sudden pulse interference is detected, the transmission signal of the charging command is instantly switched to the preset backup frequency band and the switching sequence is determined. Based on the results of the time-frequency analysis and the pulse interval characteristics, a timing control method is used to predict the interval pattern of the sudden pulse interference. The switching timing is then optimized based on the interval pattern to suppress electromagnetic interference.

8. The intelligent safety monitoring and management method for the charging status of a drone according to claim 1, characterized in that, The process of constructing a safety correlation diagram by combining the monitoring results of internal and external safety statuses, outputting the overall risk level of drone charging, and generating and executing corresponding charging instructions includes the following steps: A set of security statuses is formed by combining the monitoring results of internal and external security statuses; A safety association graph is constructed based on the set of safety states. The overall risk probability of drone charging is calculated through the safety association graph, and the overall risk probability is converted into an overall risk level. Match the charging strategy corresponding to the overall risk level, generate and execute the charging command corresponding to the charging strategy.

9. The intelligent safety monitoring and management method for the charging status of a drone according to claim 8, characterized in that, The process of constructing a security association graph based on the security state set includes the following steps: Identify the state variables in the set of safe states as root causal nodes, and define the optional charging strategies as intervention nodes; Construct a probabilistic causal path connecting the root causal node and the intervention node; Perform a hypothetical simulation on the optional charging strategy corresponding to any of the intervention nodes to simulate the propagation of charging safety risks in the probabilistic causal path and calculate the corresponding risk results; Integrate all the aforementioned risk outcomes to generate a decision matrix that correlates charging strategies with charging safety risk status, and use this decision matrix as the core decision-making basis for the safety correlation diagram.

10. A smart safety monitoring and management system for the charging status of a drone, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the intelligent safety monitoring and management method for the charging status of the drone as described in any one of claims 1 to 9.

Images