Unmanned aerial vehicle battery health management prediction method and system

Abstract

The invention discloses an unmanned aerial vehicle battery health management prediction method and system, and the system collects battery data in real time through a plurality of sensors, transmits the battery data to a cloud through an encryption channel for processing, and detects abnormal data through an Isolation Forest algorithm. Then, samples similar to abnormal data are generated by using a generative adversarial network to enhance the diversity of a data set and the robustness of a model. And analyzing the time sequence data by adopting a long-short-term memory network, and capturing the long-term change trend of the battery state. And calculating a comprehensive health score by combining normal and abnormal data through a fuzzy logic and decision tree method, and dividing the health state of the battery according to the score. And detailed maintenance suggestions are provided according to the health state of the battery. The user interface module provides real-time monitoring and visualization tools. The method effectively improves the management efficiency and safety of the unmanned aerial vehicle battery, and can be widely applied to the field of unmanned aerial vehicles and other battery management applications.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle (UAV) operation and maintenance technology, and in particular to a method and system for predicting the health management of UAV batteries. Background Art

[0002] With the rapid development of drone technology, its application in various fields is becoming more and more extensive. However, the endurance and health status of drone batteries are directly related to the operating efficiency and safety of drones. Traditional battery health management systems have shortcomings in prediction accuracy, real-time performance, and intelligent management. Therefore, this project aims to develop a drone battery health management prediction and management system based on generative adversarial networks (GANs) to improve the reliability of drone batteries and extend their service life.

[0003] As a key component of energy storage, drone batteries play a vital role in many drone fields. The advantages of lithium batteries are their high energy density, long cycle life and low self-discharge rate. However, as the use time increases, the capacity and performance of the battery gradually decline, affecting the efficiency and safety of the entire system. Therefore, it is very important for the battery management system to accurately predict the state of health (SOH) of the battery in order to optimize its performance and extend its life.

[0004] The SOH prediction methods of lithium batteries can be divided into two main categories: model-based methods and data-driven methods. Model-based methods are based on models established based on physical or chemical principles, which can intuitively explain the causes and processes of battery performance degradation. Although these methods can provide accurate predictions in theory, they are limited in practical applications. In order to accurately predict the battery life, complex mathematical models need to be established, which increases the difficulty and computational cost of modeling. The accuracy of model parameters has a strong dependence on parameters, and the acquisition and calibration of parameters are usually difficult. The data-driven method is a prediction strategy that does not rely on detailed physical or chemical models inside the battery, but evaluates its health state (SOH) and predicts future performance by analyzing various data generated by the battery during use. The advantage of this type of method is that it does not require an in-depth understanding of the internal reaction mechanism of the battery, reducing the difficulty and cost of prediction. It can handle a large amount of data in nonlinear systems and has efficient fitting capabilities. With the increase in data volume and continuous optimization of algorithms, the prediction accuracy can be continuously improved. Real-time online prediction can be achieved to provide timely health status assessment for the battery management system. This helps to optimize charging strategies, warn battery failures, etc., and improve battery efficiency and safety. However, the performance of these methods depends heavily on the quality and quantity of data. If the data is defective or insufficient in magnitude, it may lead to a decrease in prediction accuracy. In recent years, deep learning techniques have been widely used in SOH prediction. The current challenge lies in how to effectively use these different deep learning techniques, as well as how to process and interpret the results generated by these complex models. Future research may focus on optimizing the structure of these models, reducing the dependence on large amounts of training data, and improving the interpretability of the models to make them more reliable and efficient in practical applications.

[0005] The related existing technologies have the following disadvantages: The prediction accuracy of existing equipment is not high and there may be certain errors. The test method measures the battery capacity by charging and discharging, which will have a certain impact on the battery life. In addition, certain calibration and adjustments are required according to different test environments and requirements. The test time is long and requires a complete charge and discharge cycle. The complexity and cost of the equipment may be high, and it is impossible to accurately derive the health status of the battery and its remaining life directly through the data.

[0006] The current battery management systems on the market have limitations, such as single monitoring parameters and insufficient data analysis capabilities. Summary of the invention

[0007] The main purpose of the embodiment of the present invention is to propose a high-precision drone battery health management prediction method and system, which can accurately and directly derive the battery health status and its remaining life through data. This system aims to comprehensively monitor, analyze and predict the health status of drone batteries by integrating a variety of advanced sensing technologies and machine learning algorithms. The system includes data collection, processing, analysis, health assessment and decision support functions, aiming to improve the accuracy and safety of battery management.

[0008] To achieve the above object, an embodiment of the present invention provides a method for predicting the health management of a drone battery, comprising the following steps:

[0009] Data collection and preprocessing: The sensor network of the multi-source data collection and transmission module collects battery status data in real time, and constructs the training set, validation set, and test set for model training through wavelet transform and normalization processing;

[0010] Anomaly detection: Perform anomaly detection on the collected battery status data and identify abnormal data points through the Isolation Forest algorithm;

[0011] Data augmentation: Use GAN to detect potential anomalies in the data and generate virtual data augmentation training sets. The generator generates samples with a distribution similar to the real data, and the discriminator distinguishes the generated data from the real data.

[0012] Time series prediction: The enhanced training set is processed by the LSTM model to predict future battery health indicators;

[0013] Comprehensive health status assessment: Combine fuzzy logic and decision tree to conduct comprehensive health status assessment of the battery, classify the battery status according to the comprehensive health score, and generate maintenance recommendations and health reports;

[0014] Data visualization: Real-time data visualization is achieved through the front-end framework and back-end data interface, providing visual display of time series graphs, health score trends and anomaly detection results, and setting alarm thresholds and notification methods according to user instructions.

[0015] In some embodiments, the data collection and preprocessing process includes the following steps:

[0016] The battery status data is collected in real time through the sensor network of the multi-source data acquisition and transmission module, wherein the sensor network includes but is not limited to: voltage sensor, current sensor, temperature sensor, internal resistance sensor, acoustic sensor, infrared sensor; wherein, voltage sensor V(t): used to record the voltage change of the battery; current sensor I(t): used to monitor the discharge and charging current of the battery; temperature sensor T(t): used to measure the temperature of the battery to prevent overheating; internal resistance measurement Z(t): used to obtain the battery internal resistance through the AC impedance method; infrared imaging IR(t): used to detect the temperature distribution on the battery surface; in addition, the battery status data also includes discharge rate, battery capacity decay rate, electrochemical impedance spectrum; the expression of the collected battery status data is Among them, t i is the timestamp, V i is the voltage, I i is the current, T i is the temperature, Z i is the internal resistance, IR i is infrared data, R i is the discharge rate, C i is the capacity decay rate, EIS i is the electrochemical impedance spectroscopy; N is the number of samples collected;

[0017] According to the flight characteristics of the drone and the speed at which the battery health status changes, define the data collection function collect_data(t0, T, f), where t0 is the start time, T is the collection cycle length, and f is the collection frequency;

[0018] The data acquisition function returns the data set D collected at frequency f within the time period [t0, t0+T] collected It is represented as a mapping between timestamps and data points, and its expression is:

[0019]

[0020] Among them, the data set D collected is a mapping that converts the timestamp t of each data collection into k Mapped to the corresponding data point x k , D collected Contains multiple tuples (t k ,x k ), where t k is the timestamp, x k is at timestamp t k The data points collected at

[0021] After the data enters the cloud or edge computing terminal, it is first denoised and standardized to convert the raw data into purified data. clean ;

[0022] Denoising is performed by wavelet transform, and the wavelet transform formula is:

[0023]

[0024] Among them, x(t) is the signal, ψ is the mother wavelet, and a and b are scale and translation parameters;

[0025] Normalize the stored data set and construct the training set, validation set, and test set for model training;

[0026] The normalization expression is: Where μ and σ are the mean and standard deviation of the data respectively; X normalized is the normalized data set;

[0027] The data set is filtered by a Boolean function, and the expression of the process is:

[0028] D cleaned ={(t, x)∈D collected |is_valid(x)=1}

[0029] Among them, the data set D cleaned It is from D collected Data points that do not meet the validity conditions are deleted; D collected Represents the collected unfiltered data set; the validity condition is determined by the Boolean function is_valid(x). If is_valid(x) returns true, it means that the data point (t, x) meets the validity condition and should be included in D cleaned middle;

[0030] The training set, validation set and test set meet the following conditions:

[0031] D processed =D train ∪D val ∪D test and

[0032] Among them, D train represents the training set, D val represents the validation set and D test Represents the test set.

[0033] In some embodiments, the anomaly detection process includes the following steps:

[0034] Use the Isolation Forest algorithm to perform anomaly detection and data enhancement on the data:

[0035] First, a random tree is constructed by randomly selecting subsamples from the data set. Multiple random trees are created, and each tree is used to split the data points. Among them, the abnormal data points are isolated at a shallower segmentation level. The anomaly score s(x) is formulated as: Where: E(h(x)) is the average path length of data point x in the random tree; c(n) is the normalization coefficient, which depends on the number of samples n;

[0036] According to the anomaly score S(x), a threshold θ is set. If S(x)>θ, the sample x is marked as an anomaly.

[0037] In some embodiments, the data enhancement process includes the following steps:

[0038] A generator for constructing a model, the generator being used to generate simulated data corresponding to real battery health data;

[0039] A discriminator of the model is constructed, wherein the discriminator is used to distinguish whether the input data is real battery health data or simulated data;

[0040] Design objective functions of the generator and the discriminator, and define an optimizer;

[0041] According to the objective function and the optimizer, the generator and the discriminator are trained through the training set, the validation set and the test set to obtain a trained GAN model;

[0042] GAN is used to detect potential anomalies in the data and generate virtual data to enhance the training set, and a generative adversarial network is used to generate samples similar to abnormal data to enhance the diversity of the dataset and the robustness of the model.

[0043] In some embodiments, the process of time series prediction includes the following steps:

[0044] Retraining the battery health prediction model based on LSTM using the enhanced training set to predict the future health indicators of the battery, where the health indicators include but are not limited to capacity decay trend, voltage change, and internal resistance change; the prediction results of the health indicators are used to update the health scoring model;

[0045] Process time series data through LSTM model;

[0046] Arrange the preprocessed data in chronological order to construct a time series data set;

[0047] Use the training set to train the LSTM model, use the validation set to fine-tune the model, and use the test set to evaluate the model performance;

[0048] The training goal is to enable the model to capture the long-term trend of battery health status by minimizing the prediction error;

[0049] The LSTM model is constructed using the following modules, including:

[0050] Input Gate:

[0051] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0052] Forget Gate:

[0053] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0054] Output Gate:

[0055] o t =σ(W o ·[h t-1 ,x t ]+b o )

[0056] Cell status update:

[0057]

[0058] Among them, x t is the input vector at the current time step t, h t-1 is the hidden state at the previous moment, C t is the memory cell state at the current time step; C t-1 is the cell state at the previous moment; σ is the standard deviation; Represents the candidate memory cell state, which is the candidate information that is added to the memory cell through the input gate to determine whether to add it; W i , W f , W o Represents the weight matrices of the input gate, forget gate, and output gate respectively; b i 、b f 、b o 、b C Represent the bias vectors of the input gate, forget gate, and output gate respectively; i t 、f t , o t Represent the activation values ​​of the input gate, forget gate, and output gate respectively;

[0059] Use the trained LSTM model to input current and historical battery status data and predict health indicators for multiple time steps in the future to update the health score model.

[0060] In some embodiments, the process of comprehensive health status assessment includes the following steps:

[0061] Conduct a comprehensive health status assessment, classify the battery status according to the comprehensive health score, and generate maintenance recommendations and health reports;

[0062] The process of conducting a comprehensive health status assessment is as follows:

[0063] Uncertainty is handled through fuzzy logic, and the fuzzy membership function of fuzzy logic is: Among them, μ A (x) is the fuzzy membership; α controls the slope; c is the center value;

[0064] Combining fuzzy logic and decision tree to analyze various battery parameters, calculate the comprehensive health score S: Where: w i is the weight of each health parameter; x i is the parameter value; w anomaly is the weight of abnormal data; f anomaly is the comprehensive impact function of abnormal data;

[0065] Generate maintenance recommendations based on comprehensive health scores, abnormal detection results, and trend analysis results. Maintenance recommendations include, but are not limited to: adjusting charging and discharging strategies, conducting in-depth inspections, and replacing batteries;

[0066] Based on the comprehensive health score, the battery status is classified into:

[0067] Health: S>θ high ;

[0068] Sub-health: θ mid <S≤θ hihh ;

[0069] Note: θ low <S≤θ mid ;

[0070] Failure: S≤θ low ;

[0071] where θ high ,θ mid ,θ low is the preset threshold.

[0072] In some embodiments, the training of the generator and the discriminator according to the objective function and the optimizer through the training set, the validation set and the test set to obtain a trained GAN model comprises the following steps:

[0073] Let G(z;θG ) is the generator function, D(x;θ D ) is the discriminator function, where z is the noise vector, θ G and θ D are the parameters of the generator and the discriminator respectively. The training of GAN is to alternately update the parameters θ of the generator and the discriminator. G and θ D conduct;

[0074] The goal of GAN is to solve the following minimum-maximum optimization problem:

[0075]

[0076] Among them, V(D,G) is the loss function of GAN, p data (x) is the real data distribution, p z (z) is the distribution of the noise vector;

[0077] G(z;θ G ): Generator function, input is the noise vector z, output is the generated data sample, parameter is θ G ;

[0078] D(x;θ D ): Discriminator function, the input is data sample x, the output is the probability that the data sample comes from the real data distribution, and the parameter is θ D ;

[0079] The Gaussian noise z is used as the input of the generator G, which generates virtual samples similar to the real data distribution. And combined with the original training set to form an enhanced training set;

[0080] The discriminator network receives the augmented training set as input and outputs a probability indicating the probability that the input sample is from the true data distribution; during the training process, any sample that significantly deviates from the true data distribution is considered as potential abnormal data.

[0081] Another aspect of the embodiment of the present invention further provides a drone battery health management prediction system, the system comprising: a multi-source data acquisition and transmission module, a data processing and analysis module, a health status assessment module, a decision support and suggestion module, a user interface and visualization module;

[0082] The multi-source data acquisition and transmission module integrates a variety of sensors to collect battery status data in real time; the sensors include but are not limited to voltage sensors, current sensors, temperature sensors, internal resistance sensors, and infrared sensors;

[0083] The data processing and analysis module is used to: when the data is based on the local edge computing terminal or enters the cloud, it is first pre-processed, and the pre-processing includes denoising, normalization, anomaly detection and data filling; finally, GAN is used to detect potential anomalies in the data and generate a virtual data enhancement training set; then LSTM is used to perform time series analysis to capture the long-term change trend of the battery status;

[0084] The health status assessment module uses intelligent algorithms combined with fuzzy logic and decision trees to calculate the battery health score;

[0085] The decision support and recommendation module generates battery maintenance recommendations and detailed health reports based on health scores and trend analysis;

[0086] The user interface and visualization module provides real-time battery status monitoring, historical data analysis and trend prediction; the visualization tool displays include but is not limited to time series graphs, health score trends, anomaly detection results, and customized alarm thresholds and notification settings.

[0087] Another aspect of an embodiment of the present invention further provides an electronic device, including a processor and a memory;

[0088] The memory is used to store programs;

[0089] The processor executes the program to implement the method described above to achieve the above purpose.

[0090] Another aspect of an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned method is implemented.

[0091] The embodiment of the present invention also discloses a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device can read the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the above method.

[0092] The embodiments of the present invention include at least the following beneficial effects: the present invention provides a method and system for predicting the health management of unmanned aerial vehicle batteries. The scheme collects battery status data in real time through a sensor network of a multi-source data acquisition and transmission module, and constructs a training set, a validation set and a test set for model training through wavelet transform and normalization processing; performs anomaly detection on the collected battery status data, and identifies abnormal data points through the Isolation Forest algorithm; uses GAN to detect potential anomalies in the data and generate a virtual data enhanced training set, the generator generates samples similar to the real data distribution, and the discriminator distinguishes the generated data from the real data; enhances the training set through LSTM model processing to predict future battery health indicators; combines fuzzy logic and decision trees to evaluate the comprehensive health status of the battery, classifies the battery status according to the comprehensive health score, and generates maintenance suggestions and health reports; realizes real-time visualization of data through the front-end framework and the back-end data interface, provides a visual display of time series graphs, health score trends and abnormal detection results, and sets the alarm threshold and notification method according to user instructions. The present invention has high accuracy and can accurately obtain the health status of the battery and its remaining life directly through data. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] Figure 1 It is a schematic diagram of an implementation environment provided by an embodiment of the present invention;

[0094] Figure 2 is a flow chart of the overall steps provided by an embodiment of the present invention;

[0095] Figure 3 It is a schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0096] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present invention. They are only examples of devices and methods consistent with some aspects of the embodiments of the present invention as detailed in the attached claims.

[0097] It is understood that the terms "first", "second", etc. used in the present invention may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present invention, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0098] The terms "at least one", "multiple", "each", "any", etc. used in the present invention, at least one includes one, two or more, multiple includes two or more, each refers to each of the corresponding multiple, and any refers to any one of the multiple.

[0099] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein are only for the purpose of describing the embodiments of the present invention and are not intended to limit the present invention.

[0100] The unmanned aerial vehicle battery health management prediction method and system provided in the embodiment of the present invention relate to the field of unmanned aerial vehicle operation and maintenance technology. The unmanned aerial vehicle battery health management prediction method provided in the embodiment of the present invention can be applied to a terminal, can also be applied to a server, and can also be software running in a terminal or a server. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, and a car terminal, etc., but is not limited to this; the server side can be configured as an independent physical server, or it can be configured as a server cluster or a distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network; the software can be an application that implements the unmanned aerial vehicle battery health management prediction method, etc., but is not limited to the above forms.

[0101] The present invention can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present invention can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present invention can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0102] like Figure 1 FIG. 1 is a schematic diagram of an implementation environment provided by an embodiment of the present invention. Figure 1 The implementation environment includes at least one terminal 102 and a server 101. The terminal 102 and the server 101 can be connected to a network wirelessly or wired to complete data transmission and exchange.

[0103] Server 101 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), as well as big data and artificial intelligence platforms.

[0104] In addition, the server 101 can also be a node server in the blockchain network. Blockchain is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, and encryption algorithm.

[0105] The terminal 102 may be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, etc. The terminal 102 may also be a vehicle-mounted terminal of various device types described above, but is not limited thereto. The terminal 102 and the server 101 may be directly or indirectly connected via wired or wireless communication, which is not limited in this embodiment of the present invention.

[0106] Based on the example Figure 1In the implementation environment shown, an embodiment of the present invention provides a method for predicting the battery health management of a UAV. The following is an example of applying the method for predicting the battery health management of a UAV to a server 101. It can be understood that the method can also be applied to a terminal 102.

[0107] Reference Figure 2 , Figure 2 The flowchart of the method for predicting the health management of a drone battery applied to a server provided in an embodiment of the present invention, the execution subject of the method can be any of the aforementioned computer devices (including a server or a terminal). Figure 2 , the method may include the following steps:

[0108] Data collection and preprocessing: The sensor network of the multi-source data collection and transmission module collects battery status data in real time, and constructs the training set, validation set, and test set for model training through wavelet transform and normalization processing;

[0109] Anomaly detection: Perform anomaly detection on the collected battery status data and identify abnormal data points through the Isolation Forest algorithm;

[0110] Data augmentation: Use GAN to detect potential anomalies in the data and generate virtual data augmentation training sets. The generator generates samples with a distribution similar to the real data, and the discriminator distinguishes the generated data from the real data.

[0111] Time series prediction: The enhanced training set is processed by the LSTM model to predict future battery health indicators;

[0112] Comprehensive health status assessment: Combine fuzzy logic and decision tree to conduct comprehensive health status assessment of the battery, classify the battery status according to the comprehensive health score, and generate maintenance recommendations and health reports;

[0113] Data visualization: Real-time data visualization is achieved through the front-end framework and back-end data interface, providing visual display of time series graphs, health score trends and anomaly detection results, and setting alarm thresholds and notification methods according to user instructions.

[0114] The following takes a specific application scenario as an example to describe the specific implementation process of the method of the present invention in detail:

[0115] 1. Collect data: Collect the usage data of the drone battery under various flight conditions, including the battery's charging cycle, discharging cycle, temperature, voltage, current and other parameters.

[0116] 2. Data screening: remove useless data and abnormal data to ensure data quality.

[0117] 3. Data labeling: Label the data to clarify which data intervals represent healthy status and which data intervals represent potential failures or performance degradation.

[0118] 4. Data unification: Adjust the data to the same module to facilitate model training.

[0119] 5. Data enhancement: Design the generator, discriminator, and objective function of the GAN model. The goal of the generator is to generate simulated data that is closer and closer to the real data, while the goal of the discriminator is to identify the authenticity of the data more and more accurately.

[0120] 6. Real-time data access: Realize the system's access to the real-time data of the drone battery, including real-time monitoring of parameters such as voltage, current, and temperature.

[0121] 7. Time Series Forecasting: Process the enhanced dataset through the LSTM model to predict future battery health indicators.

[0122] 8. Integrated GAN-LSTM model: Integrate the trained GAN model into the drone battery health management prediction and management system to predict the battery health status, including remaining life, potential failures, etc.

[0123] 9. Health status prediction: Combine fuzzy logic and decision tree to conduct comprehensive health status assessment of the battery, and classify the battery status according to the comprehensive health score.

[0124] 12. Management decision support: Generate maintenance recommendations and health reports based on prediction results to provide management decision support, such as reminding to replace batteries, adjust flight plans, etc.

[0125] The embodiment of the present invention collects battery data in real time through multiple sensors and uses the Isolation Forest algorithm to detect abnormal data. Subsequently, the long short-term memory network is used to analyze the time series data to capture the long-term change trend of the battery state. At the same time, by using the generative adversarial network to generate samples similar to abnormal data, new and simulated battery health data can be generated based on the existing drone battery health data to enhance the diversity of the data set and the robustness of the model. The health state of the drone battery can be predicted using the generated simulated data and the real battery health data. The prediction model can predict key indicators such as the remaining life of the battery and the performance decay trend. Based on the prediction results, the system can manage the health state of the drone battery. Through fuzzy logic and decision tree methods, combined with normal and abnormal data, a comprehensive health score (S) is calculated, and the battery health state is divided according to the score. Detailed maintenance suggestions are provided based on the battery health state. The user interface module provides real-time monitoring and visualization tools. The invention effectively improves the management efficiency and safety of drone batteries and can be widely used in the field of drones and other battery management applications. For example, when it is predicted that the battery performance is about to decline significantly, the system can remind the user to replace the battery or take other maintenance measures. In addition, the system can also adjust the flight strategy of the drone according to the real-time status of the battery to maximize the battery life. The model can use GAN to generate simulated data to expand the training data set and improve the accuracy of the prediction model. It can also predict and manage the health status of the drone battery in real time to improve the safety and reliability of the drone.

[0126] The specific implementation steps are as follows:

[0127] 1. Multi-source data acquisition and transmission: The status data of the drone battery is collected in real time through the integrated multiple sensors, including voltage (V(t)), current (I(t)), temperature (T(t)), internal resistance (Z(t)), acoustic signal (A(t)) and infrared signal (IR(t)); the frequency of data acquisition is f s , the collected data is transmitted to the cloud through an encrypted channel.

[0128] 2. Data collection: According to the flight characteristics of the drone and the speed of battery health status change, the data collection function collect_data(t0, T, f) can be defined, where t0 is the start time, T is the collection cycle length, and f is the collection frequency. This function returns the data set D collected at frequency f in the time period [t0, t0+T] collected A dataset can be represented as a mapping of timestamps to data points:

[0129]

[0130] Dataset D collectedis a mapping that converts the timestamp t of each data collection into k Mapped to the corresponding data point x k Therefore, D collected Can be represented as a set containing multiple tuples (t k ,x k ), where t k is the timestamp, x k is at timestamp t k The data points collected.

[0131] 3. Data format and transmission: collected data It is initially stored locally and then transmitted to the cloud through an encrypted channel to ensure the security and integrity of the data.

[0132] Among them, t i is the timestamp, V i is the voltage, I i is the current, T i is the temperature, Z i is the internal resistance, IR i is infrared data, R i is the discharge rate, C i is the capacity decay rate, EIS i is the electrochemical impedance spectroscopy. N is the number of samples collected for data.

[0133] 4. Data preprocessing After the data enters the cloud or edge computing terminal, it is first subjected to denoising (using wavelet transform) and standardization to convert the raw data into purified data. clean .

[0134] Denoising is performed by wavelet transform, and the wavelet transform formula is:

[0135]

[0136] Where x(t) is the signal, ψ is the mother wavelet, and a and b are the scale and translation parameters.

[0137] Normalize the stored data set and construct the training set, validation set, and test set for model training;

[0138] The normalization expression is: Where μ and σ are the mean and standard deviation of the data respectively; X normalized is the normalized data set.

[0139] Delete the data points that do not meet the validity conditions and use the Boolean function is_valid to filter them.

[0140] D cleaned ={(t, x)∈Dcollected |is_valid(x)=1}

[0141] Dataset D cleaned It is from D collected The data points that do not meet the validity condition are deleted. The validity condition is determined by the Boolean function is_valid(x). If is_valid(x) returns true (or equal to 1), it means that the data point (t, x) meets the validity condition and should be included in D cleaned middle.

[0142] D cleaned is a subset that includes D collected The data point (t, x) that meets the validity condition. Therefore, D cleaned It can be represented as a set containing multiple tuples (t, x), where each tuple (t, x) satisfies is_valid(x) = 1.

[0143] SOH is an important indicator for evaluating the degree of battery aging or decline, reflecting the health of the battery, and is usually expressed as a percentage, with 100% for new batteries when they leave the factory. When defined by capacity reduction, SOH is the ratio of the current remaining capacity of the battery to the rated capacity.

[0144]

[0145] Among them C n is the current remaining capacity, C o is the rated capacity of the battery. The SOH of the battery is calculated based on the acquired data.

[0146] 5. Data unification: use Where μ and σ are the mean and standard deviation of the data respectively. Dataset creation: The processed data set D processed Divide into training set D train , validation set D val and the test set D test , satisfying the conditions of mutual exclusion and the union being the whole set. That is, D processed =D train ∪D val ∪D test and

[0147] 6. Training the GAN model: The training process is performed by alternately optimizing θ D and θ G To minimize max D V(D,C) and min G max D V(D,G). This is usually done via gradient descent or its variants.

[0148] The training set is used to initially learn and train the model. The model learns to recognize patterns and features from this data.

[0149] Use the validation set to tune the model and select hyperparameters. Evaluate the model performance and make adjustments.

[0150] Use the test set to test the model's performance on unknown data to evaluate its generalization ability.

[0151] Training the discriminator: Fix the parameters of the generator and sample a batch of real data x from the real data distribution. Sample a batch of noise z from the noise distribution and generate fake data G(z) through the generator. Use real data and fake data to train the discriminator to maximize the discriminator part in V(D,G) (i.e. maximize

[0152] Train the generator: fix the parameters of the discriminator and only update the parameters of the generator. Sample a batch of noise z from the noise distribution and generate fake data G(z) through the generator. Use fake data to train the generator to minimize the generator part in V(D,G) (i.e. minimize

[0153] Repeat the above steps: alternately train the discriminator and the generator until the predetermined training round is reached or other stopping conditions are met.

[0154] 7. Add LSTM model: LSTM model is particularly suitable for processing time series data, which captures long-term dependencies through memory and forgetting mechanisms. The key components of LSTM include:

[0155] Input Gate:

[0156] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0157] Forget Gate:

[0158] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0159] Output Gate:

[0160] o t =σ(W o ·[h t-1 ,x t ]+b o )

[0161] Cell status update:

[0162]

[0163] Among them, x t is the input vector at the current time step t, h t-1 is the hidden state at the previous moment, C t is the memory cell state at the current time step; C t-1 is the cell state at the previous moment; σ is the standard deviation; Represents the candidate memory cell state, which is the candidate information that is added to the memory cell through the input gate to determine whether to add it; W i , W f , W o Represents the weight matrices of the input gate, forget gate, and output gate respectively; b i 、b f 、b o 、b C Represent the bias vectors of the input gate, forget gate, and output gate respectively; i t 、f t , o t Represent the activation values ​​of the input gate, forget gate, and output gate respectively.

[0164] LSTM is used to predict future battery health indicators, such as capacity decay trends, changes in voltage and internal resistance, etc. These prediction results are used to update the health score model.

[0165] 8. Health status prediction: Use the GAN model to process real-time data, and the prediction function accepts f predict Accepts new data points and outputs predictions Right now Where θ* is the trained model parameter, which can predict the health status of the battery, including remaining life, potential failures, etc.

[0166] Fuzzy logic is used to deal with uncertainty. Typical fuzzy membership functions are:

[0167]

[0168] μ A (x) is the fuzzy membership; α controls the slope; c is the center value.

[0169] Combining fuzzy logic and decision tree to analyze various battery parameters, calculate the comprehensive health score S:

[0170]

[0171] Where: w i is the weight of each health parameter; x iis the parameter value; w anomaly is the weight of abnormal data; f anomaly It is the comprehensive influence function of abnormal data.

[0172] 9. Management decision support: Based on the forecast results Make decisions based on preset decision rules (such as threshold comparison), state classification and decision support;

[0173] Based on the health score S, anomaly detection results, and trend analysis results, the system generates maintenance recommendations.

[0174] 10. Battery status classification

[0175] Based on the comprehensive health score, the battery status is classified into:

[0176] Health: S>θ high

[0177] Sub-health: θ mid <S≤θ high

[0178] Note (including abnormal status): θ low <S≤θ mid

[0179] Failure: S≤θ low

[0180] where θ high ,θ mid ,θ low is the preset threshold.

[0181] 11Decision support and maintenance recommendations

[0182] Based on the battery status, the system generates specific maintenance recommendations (such as adjusting the charging and discharging strategy, performing deep testing, replacing the battery, etc.). Abnormal detection can also trigger an emergency alarm mechanism.

[0183] Based on the battery health score and status classification, the system provides maintenance recommendations, including but not limited to:

[0184] Adjust the charging and discharging strategy to extend battery life.

[0185] Remind users to check and maintain batteries regularly.

[0186] When the score falls below the threshold, it is recommended to replace the battery.

[0187] Provide management decision support, such as reminders to replace batteries, adjust flight plans, etc.

[0188] 12. User interface and visualization module

[0189] Real-time monitoring: Real-time visualization of data is achieved through the front-end framework and back-end data interface.

[0190] Historical data analysis: Provides visualization of time series graphs, health score trends, and anomaly detection results.

[0191] User-defined settings: Allow users to set alarm thresholds and notification methods.

[0192] In summary, the embodiment of the present invention collects battery data in real time through multiple sensors and uses the IsolationForest algorithm to detect abnormal data. Subsequently, the long short-term memory network is used to analyze the time series data to capture the long-term change trend of the battery state. Through the training of the long short-term memory network, the system can learn the complexity and nonlinear characteristics of the battery degradation process and achieve high-precision remaining life and performance degradation prediction. At the same time, the generative adversarial network is used to generate samples similar to abnormal data to enhance the diversity of the data set and the robustness of the model. The adversarial training mechanism of the generator and the discriminator is adopted to improve the performance of both through continuous game, so as to achieve accurate simulation and prediction of the complex battery degradation process. Combining the output of the long short-term neural network and the real battery data, a high-precision prediction model is constructed to achieve accurate prediction of the remaining life and performance degradation trend of the drone battery. Through fuzzy logic and decision tree methods, the comprehensive health score (S) is calculated by combining normal and abnormal data, and the battery health status is divided according to the score. Based on the prediction results, the system can manage the health status of the drone battery and provide detailed maintenance suggestions based on the battery health status. The user interface module provides real-time monitoring and visualization tools. The invention effectively improves the management efficiency and safety of drone batteries and can be widely used in the field of drones and other battery management applications. The invention can be widely used in various drone platforms, such as aerial photography drones, agricultural drones, environmental monitoring drones and other drone-related industries. By improving the reliability of drone batteries and extending their service life, the operating and maintenance costs of drones can be reduced, promoting the further development and application of drone technology.

[0193] The embodiment of the present invention also provides a drone battery health management prediction system, including:

[0194] Multi-source data acquisition and transmission module, data processing and analysis module, health status assessment module, decision support and suggestion module, user interface and visualization module;

[0195] The multi-source data acquisition and transmission module integrates a variety of sensors to collect battery status data in real time; the sensors include but are not limited to voltage sensors, current sensors, temperature sensors, internal resistance sensors, and infrared sensors;

[0196] The data processing and analysis module is used to: when the data is based on the local edge computing terminal or enters the cloud, it is first pre-processed, and the pre-processing includes denoising, normalization, anomaly detection and data filling; finally, GAN is used to detect potential anomalies in the data and generate a virtual data enhancement training set; then LSTM is used to perform time series analysis to capture the long-term change trend of the battery status;

[0197] The health status assessment module uses intelligent algorithms combined with fuzzy logic and decision trees to calculate the battery health score;

[0198] The decision support and recommendation module generates battery maintenance recommendations and detailed health reports based on health scores and trend analysis;

[0199] The user interface and visualization module provides real-time battery status monitoring, historical data analysis and trend prediction; the visualization tool displays include but is not limited to time series graphs, health score trends, anomaly detection results, and customized alarm thresholds and notification settings.

[0200] It can be understood that the contents of the above method embodiments are all applicable to the present system embodiments, the functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0201] The embodiment of the present invention further provides an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned drone battery health management prediction method when executing the computer program. The electronic device can be any intelligent terminal including a tablet computer, a vehicle-mounted computer, etc.

[0202] It can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0203] See also Figure 3 , Figure 3 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:

[0204] The processor 301 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention;

[0205] The memory 302 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 302 can store an operating system and other applications. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 302, and the processor 301 calls and executes the drone battery health management prediction method of the embodiment of the present invention;

[0206] Input / output interface 303, used to implement information input and output;

[0207] The communication interface 304 is used to realize the communication interaction between the device and other devices. The communication can be realized through a wired manner (such as USB, network cable, etc.) or a wireless manner (such as mobile network, WIFI, Bluetooth, etc.);

[0208] A bus 305 that transmits information between the various components of the device (e.g., the processor 301, the memory 302, the input / output interface 303, and the communication interface 304);

[0209] The processor 301 , the memory 302 , the input / output interface 303 and the communication interface 304 are connected to each other in communication within the device via the bus 305 .

[0210] An embodiment of the present invention further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned drone battery health management prediction method is implemented.

[0211] It can be understood that the contents of the above method embodiments are all applicable to the present storage medium embodiments, the functions specifically implemented by the present storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0212] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0213] The embodiments described in the embodiments of the present invention are intended to more clearly illustrate the technical solutions of the embodiments of the present invention, and do not constitute a limitation on the technical solutions provided by the embodiments of the present invention. Those skilled in the art can appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present invention are also applicable to similar technical problems.

[0214] Those skilled in the art will appreciate that the technical solutions shown in the figures do not limit the embodiments of the present invention and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0215] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0216] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0217] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0218] It should be understood that in the present invention, "at least one (item)" refers to one or more, and "plurality" refers to two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can represent: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0219] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0220] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0221] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0222] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including multiple instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store programs.

[0223] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the embodiments of the present invention is not limited thereby. Any modification, equivalent substitution and improvement made by those skilled in the art without departing from the scope and essence of the embodiments of the present invention shall be within the scope of the rights of the embodiments of the present invention.

Claims

1. A method for predicting the health management of drone batteries, characterized in that: The following steps are involved: Data collection and preprocessing: The sensor network of the multi-source data collection and transmission module collects battery status data in real time, and constructs the training set, validation set, and test set for model training through wavelet transform and normalization processing; Anomaly detection: Perform anomaly detection on the collected battery status data and identify abnormal data points through the Isolation Forest algorithm; Data augmentation: Use GAN to detect potential anomalies in the data and generate virtual data augmentation training sets. The generator generates samples with a distribution similar to the real data, and the discriminator distinguishes the generated data from the real data. Time series prediction: The enhanced training set is processed by the LSTM model to predict future battery health indicators; Comprehensive health status assessment: Combine fuzzy logic and decision tree to conduct comprehensive health status assessment of the battery, classify the battery status according to the comprehensive health score, and generate maintenance recommendations and health reports; Data visualization: Real-time data visualization is achieved through the front-end framework and back-end data interface, providing visual display of time series graphs, health score trends and anomaly detection results, and setting alarm thresholds and notification methods according to user instructions.

2. The method for predicting the health management of a drone battery according to claim 1, characterized in that: The process of data collection and preprocessing includes the following steps: The battery status data is collected in real time through the sensor network of the multi-source data acquisition and transmission module, wherein the sensor network includes but is not limited to: voltage sensor, current sensor, temperature sensor, internal resistance sensor, acoustic sensor, infrared sensor; wherein, voltage sensor V(t): used to record the voltage change of the battery; current sensor I(t): used to monitor the discharge and charging current of the battery; temperature sensor T(t): used to measure the temperature of the battery to prevent overheating; internal resistance measurement Z(t): used to obtain the battery internal resistance through the AC impedance method; infrared imaging IR(t): used to detect the temperature distribution on the battery surface; in addition, the battery status data also includes discharge rate, battery capacity decay rate, electrochemical impedance spectrum; the expression of the collected battery status data is Among them, t i is the timestamp, V i is the voltage, I i is the current, T i is the temperature, Z i is the internal resistance, IR i is infrared data, R i is the discharge rate, C i is the capacity decay rate, EIS i is the electrochemical impedance spectroscopy; N is the number of samples collected; According to the flight characteristics of the drone and the speed at which the battery health status changes, define the data collection function collect_data(t0, T, f), where t0 is the start time, T is the collection cycle length, and f is the collection frequency; The data acquisition function returns the data set D collected at frequency f within the time period [t0, t0+T] collected It is represented as a mapping between timestamps and data points, and its expression is: Among them, the data set D collected is a mapping that converts the timestamp t of each data collection into k Mapped to the corresponding data point x k , D collected Contains multiple tuples (t k ,x k ), where t k is the timestamp, x k is at timestamp t k The data points collected at After the data enters the cloud or edge computing terminal, it is first denoised and standardized to convert the raw data into purified data. clean ; Denoising is performed by wavelet transform, and the wavelet transform formula is: Among them, x(t) is the signal, ψ is the mother wavelet, and a and b are scale and translation parameters; Normalize the stored data set and construct the training set, validation set, and test set for model training; The normalization expression is: Where μ and σ are the mean and standard deviation of the data respectively; X normalized is the normalized data set; The data set is filtered by a Boolean function, and the expression of the process is: D cleaned ={(t,x)∈D collected |is_valid(x)=1} Among them, the data set D cleaned It is from D collected Data points that do not meet the validity conditions are deleted; D collected Represents the collected unfiltered data set; the validity condition is determined by the Boolean function is_valid(x). If is_valid(x) returns true, it means that the data point (t, x) meets the validity condition and should be included in D cleaned middle; The training set, validation set and test set meet the following conditions: D processed =D train ∪D val ∪D test and Among them, D train represents the training set, D val represents the validation set and D test Represents the test set.

3. The method for predicting the health management of a drone battery according to claim 1, characterized in that: The anomaly detection process includes the following steps: Use the Isolation Forest algorithm to perform anomaly detection and data enhancement on the data: First, a random tree is constructed by randomly selecting subsamples from the data set. Multiple random trees are created, and each tree is used to split the data points. Among them, the abnormal data points are isolated at a shallower segmentation level. The anomaly score s(x) is formulated as: Where: E(h(x)) is the average path length of data point x in the random tree; c(n) is the normalization coefficient, which depends on the number of samples n; According to the anomaly score S(x), a threshold θ is set. If S(x)>θ, the sample x is marked as an anomaly.

4. The method for predicting the health management of a drone battery according to claim 1, characterized in that: The data enhancement process includes the following steps: A generator for constructing a model, the generator being used to generate simulated data corresponding to real battery health data; A discriminator of the model is constructed, wherein the discriminator is used to distinguish whether the input data is real battery health data or simulated data; Design objective functions of the generator and the discriminator, and define an optimizer; According to the objective function and the optimizer, the generator and the discriminator are trained through the training set, the validation set and the test set to obtain a trained GAN model; GAN is used to detect potential anomalies in the data and generate virtual data to enhance the training set, and a generative adversarial network is used to generate samples similar to abnormal data to enhance the diversity of the dataset and the robustness of the model.

5. The method for predicting the health management of a drone battery according to claim 1, characterized in that: The process of time series forecasting includes the following steps: Retraining the battery health prediction model based on LSTM using the enhanced training set to predict the future health indicators of the battery, where the health indicators include but are not limited to capacity decay trend, voltage change, and internal resistance change; the prediction results of the health indicators are used to update the health scoring model; Process time series data through LSTM model; Arrange the preprocessed data in chronological order to construct a time series data set; Use the training set to train the LSTM model, use the validation set to fine-tune the model, and use the test set to evaluate the model performance; The training goal is to enable the model to capture the long-term trend of battery health status by minimizing the prediction error; The LSTM model is constructed using the following modules, including: Input Gate: i t =σ(W i ·[h t-1 ,x t ]+b i ) Forget Gate: f t =σ(W f ·[h t-1 ,x t ]+b f ) Output Gate: the t =σ(W o ·[h t-1 ,x t ]+b o ) Cell status update: Among them, x t is the input vector at the current time step t, h t-1 is the hidden state at the previous moment, C t is the memory cell state at the current time step; C t-1 is the cell state at the previous moment; σ is the standard deviation; Represents the candidate memory cell state, which is the candidate information that is added to the memory cell through the input gate to determine whether to add it; W i , W f , W o Represents the weight matrices of the input gate, forget gate, and output gate respectively; b i 、b f 、b o 、b C Represent the bias vectors of the input gate, forget gate, and output gate respectively; i t 、f t , o t Represent the activation values ​​of the input gate, forget gate, and output gate respectively; Use the trained LSTM model to input current and historical battery status data and predict health indicators for multiple time steps in the future to update the health score model.

6. The method for predicting the health management of a drone battery according to claim 1, characterized in that: The process of comprehensive health status assessment includes the following steps: Conduct a comprehensive health status assessment, classify the battery status according to the comprehensive health score, and generate maintenance recommendations and health reports; The process of conducting a comprehensive health status assessment is as follows: Uncertainty is handled through fuzzy logic, and the fuzzy membership function of fuzzy logic is: Among them, μ A (x) is the fuzzy membership; α controls the slope; c is the center value; Combining fuzzy logic and decision tree to analyze various battery parameters, calculate the comprehensive health score S: Where: w i is the weight of each health parameter; x i is the parameter value; w anomaly is the weight of abnormal data; f anomaly is the comprehensive impact function of abnormal data; Generate maintenance recommendations based on comprehensive health scores, abnormal detection results, and trend analysis results. Maintenance recommendations include, but are not limited to: adjusting charging and discharging strategies, conducting in-depth inspections, and replacing batteries; Based on the comprehensive health score, the battery status is classified into: Health: S>θ high ; Sub-health: θ mid <S≤θ high ; Note: θ low <S≤θ mid ; Failure: S≤θ low ; where θ high ,θ mid ,θ low is the preset threshold.

7. The method for predicting the health management of a drone battery according to claim 4, characterized in that: The method of training the generator and the discriminator according to the objective function and the optimizer through the training set, the validation set and the test set to obtain a trained GAN model comprises the following steps: Let G(z;θ G ) is the generator function, D(x;θ D ) is the discriminator function, where z is the noise vector, θ G and θ D are the parameters of the generator and the discriminator respectively. The training of GAN is to alternately update the parameters θ of the generator and the discriminator. G and θ D conduct; The goal of GAN is to solve the following minimum-maximum optimization problem: Among them, V(D,G) is the loss function of GAN, p data (x) is the real data distribution, p z (z) is the distribution of the noise vector; G(z;θ G ): Generator function, input is the noise vector z, output is the generated data sample, parameter is θ G ; D(x;θ D ): Discriminator function, the input is data sample x, the output is the probability that the data sample comes from the real data distribution, and the parameter is θ D ; The Gaussian noise z is used as the input of the generator G, which generates virtual samples similar to the real data distribution. And combined with the original training set to form an enhanced training set; The discriminator network receives the augmented training set as input and outputs a probability indicating the probability that the input sample is from the true data distribution; during the training process, any sample that significantly deviates from the true data distribution is considered as potential abnormal data.

8. A drone battery health management prediction system, characterized in that: The system includes: multi-source data acquisition and transmission module, data processing and analysis module, health status assessment module, decision support and suggestion module, user interface and visualization module; The multi-source data acquisition and transmission module integrates a variety of sensors to collect battery status data in real time; the sensors include but are not limited to voltage sensors, current sensors, temperature sensors, internal resistance sensors, and infrared sensors; The data processing and analysis module is used to: when the data is based on the local edge computing terminal or enters the cloud, it is first pre-processed, and the pre-processing includes denoising, normalization, anomaly detection and data filling; finally, GAN is used to detect potential anomalies in the data and generate a virtual data enhancement training set; then LSTM is used to perform time series analysis to capture the long-term change trend of the battery status; The health status assessment module uses intelligent algorithms combined with fuzzy logic and decision trees to calculate the battery health score; The decision support and recommendation module generates battery maintenance recommendations and detailed health reports based on health scores and trend analysis; The user interface and visualization module provides real-time battery status monitoring, historical data analysis and trend prediction; the visualization tool displays include but is not limited to time series graphs, health score trends, anomaly detection results, and customized alarm thresholds and notification settings.

9. An electronic device, characterized in that: including a processor and a memory; The memory is used to store programs; The processor executes the program to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, wherein the computer program implements the method according to any one of claims 1 to 7 when executed by a processor.

Images