A lithium battery altitude capacity attenuation early warning method and system based on material characteristics

Abstract

The application relates to a lithium battery altitude capacity attenuation early warning method and system based on material characteristics, which comprises the following steps: configuring a lithium battery material type and a warning condition, calling corresponding historical attenuation reference data, and initializing a collection device. Altitude and battery capacity are synchronously collected and pretreated, historical reference data are fused for feature extraction, and standardized analysis data sets are obtained. A ternary correlation model of materials, altitude and capacity is constructed through machine learning, the altitude influence weight is quantified, and the capacity attenuation law under the current altitude is output. According to the law and the real-time altitude trend, the attenuation degree and type are predicted; if the warning condition is met, a warning signal is generated. The warning levels are divided and graded warning information is generated. According to user handling feedback, the historical reference data are updated and the model parameters are optimized. The application can accurately predict the capacity attenuation of a lithium battery in a high-altitude environment in real time, and improves the safety and adaptability of battery management.

Description

Technical Field

[0001] This application relates to the field of battery management systems and early warning technology, and in particular to a method and system for early warning of capacity decay of lithium batteries based on material properties at altitude. Background Technology

[0002] With the continuous improvement of lithium battery energy density and the sustained decrease in cost, lithium batteries have been widely used as a core power source in consumer electronics, electric vehicles, aerospace, and various portable industrial equipment. In particular, in scenarios such as drone logistics, mountain surveying, high-altitude tourism, and border patrol, mobile devices equipped with lithium batteries often need to perform tasks in geographical environments with drastic altitude changes. Increased altitude leads to decreased atmospheric pressure and air density. This environmental change directly affects the electrochemical processes inside lithium batteries, such as accelerating electrolyte solvent evaporation, reducing the stability of the electrode-electrolyte interface, and altering ion conduction rates. This results in a significant decrease in usable battery capacity and may even accelerate battery aging, induce thermal runaway risks, and seriously threaten the continuous operational safety and mission reliability of the equipment.

[0003] However, existing monitoring and early warning technologies for lithium batteries mostly focus on threshold alarms for basic parameters such as voltage, current, and temperature, or on estimations based on battery state of health (SOH) and remaining charge (SOC) under fixed scenarios. In practical applications, the lack of comprehensive consideration of environmental factors, especially the critical stress factor of continuously changing dynamic altitude, makes it impossible to predict the risk of capacity degradation when the battery is about to enter or is already in a high-altitude environment. The early warning behavior is lagging, often triggering only when the battery performance has been severely damaged, thus losing its preventative significance.

[0004] Furthermore, existing systems typically use general, static warning thresholds, ignoring the fundamental differences in altitude sensitivity among different material systems. They also fail to consider that the ability of a single cell to withstand environmental stress dynamically degrades throughout its entire lifespan as cycle aging deepens. This makes it difficult for the warning system to accurately reflect the true state of an individual battery, easily leading to false alarms or missed alarms. Consequently, this reduces equipment availability and triggers unnecessary maintenance actions, and may even directly ignore equipment risks, resulting in serious accidents such as thermal runaway. Summary of the Invention

[0005] To address the aforementioned technical issues, this application provides a method and system for early warning of lithium battery capacity degradation based on material properties at altitude.

[0006] Firstly, this application provides a method for early warning of lithium battery capacity degradation based on material properties at altitude, employing the following technical solution: Configure lithium battery material types and preset warning conditions, and obtain historical altitude capacity decay benchmark data that matches the lithium battery material type from the material property database; Initialize the communication protocol and acquisition frequency of the data acquisition device, and synchronously acquire real-time altitude data, battery capacity data and auxiliary data through the data acquisition device, add a unified timestamp, and generate the original dataset; Preprocessing is performed on the original dataset, and feature extraction is performed based on the historical elevation capacity decay benchmark data to generate a dataset that integrates benchmark features; Perform a standardization transformation on the merged dataset to generate a standardized analysis dataset; Based on the standardized analysis dataset, a ternary correlation model of lithium battery material type, altitude and battery capacity is constructed through machine learning algorithms to quantify the influence weight of altitude change on battery capacity and output the capacity decay law of the current altitude range. Based on the capacity decay law and the real-time altitude change trend, the degree and type of battery capacity decay are predicted, and a warning signal is generated when the capacity prediction result meets the preset warning conditions. Based on the warning signals, warning levels are divided, graded warning information is generated and pushed to terminal devices; wherein, the graded warning information includes altitude status, capacity prediction results and response instructions; Obtain user response feedback data after receiving the graded early warning information, update the historical altitude capacity attenuation benchmark data, and optimize the parameters of the ternary correlation model.

[0007] By adopting the above technical solution, the performance management of lithium batteries in high-altitude environments is upgraded from passive monitoring to proactive, material-characteristic-based, predictive, and self-learning intelligent early warning. This technical solution can not only accurately quantify the altitude sensitivity differences of batteries with different materials, predict capacity decay risks in advance, and distinguish between reversible and irreversible losses, but also provide clear response strategies through graded early warning. Finally, through a closed-loop learning mechanism, it continuously improves the system's adaptability and accuracy, thereby enhancing the operational safety and mission reliability of mobile equipment such as drones and high-altitude operation equipment in dynamic altitude environments.

[0008] Optionally, the steps of preprocessing the original dataset and extracting features based on the historical elevation capacity decay benchmark data to generate a dataset fused with benchmark features include: Obtain the raw dataset, including timestamped altitude data, battery capacity data, and auxiliary data; Perform timestamp alignment processing on the original dataset to generate a time-synchronized dataset; Based on preset anomaly detection rules, outliers are removed from the time synchronization dataset to generate a cleaned dataset. Material sensitivity characteristics and degradation trend characteristics are extracted from the historical altitude capacity degradation benchmark data; The material sensitivity features and attenuation trend features are fused into the cleaned dataset to generate a dataset with fused baseline features.

[0009] By adopting the above technical solution, the problems of low data quality, limited information dimensions, and disconnect between historical experience and real-time status in lithium battery monitoring under altitude change scenarios are systematically solved. The final dataset not only ensures accurate time synchronization and physical authenticity of the data, but also deeply embeds the inherent altitude sensitivity of the material and historical statistical attenuation patterns into the real-time observation data in a calculable and quantifiable feature form through feature engineering. This provides subsequent early warning models with a much richer and more discriminative input than the original signal or simply cleaned data, fundamentally improving the accuracy and reliability of subsequent capacity attenuation pattern mining, trend prediction, and risk assessment. It is a solid data foundation for the entire intelligent early warning system to achieve accurate analysis and early warning.

[0010] Optionally, based on the standardized analysis dataset, the steps of constructing a ternary correlation model between lithium battery material type, altitude, and battery capacity using machine learning algorithms, quantifying the impact weight of altitude changes on battery capacity, and outputting the capacity decay pattern within the current altitude range include: Obtain standardized analysis datasets, including lithium battery material types, altitude data, battery capacity data, and auxiliary data; Based on the standardized analysis dataset, a ternary correlation model of lithium battery material type, altitude data, and battery capacity data is trained using machine learning algorithms. The machine learning algorithm is either a gradient boosting tree or a support vector machine. The training includes fitting the nonlinear relationship between material type, altitude, and battery capacity through an iterative optimization algorithm, and using auxiliary data as input feature variables to optimize the fitting process. The ternary correlation model is used to analyze the capacity decay rate corresponding to the unit of altitude change, and to calculate the weight of the impact of altitude change on battery capacity; wherein, the unit of altitude change is preset to a fixed value; Based on the aforementioned influence weights, the altitude sensitivity thresholds for lithium batteries of different material types are determined, and the capacity decay pattern within the current altitude range is output.

[0011] By adopting the above technical solution, a high-order correlation model integrating material properties, altitude, environmental covariates and battery capacity was established. The influence weight of altitude changes can be dynamically extracted from the model, and then output the capacity decay law, which provides a reliable and quantifiable decision basis for realizing a forward-looking lithium battery altitude capacity decay early warning based on material properties. This improves the accuracy and intelligence of battery status assessment and risk management in dynamic altitude application scenarios such as drones and plateau equipment.

[0012] Optionally, based on the capacity decay law and real-time altitude change trend, the step of predicting the degree and type of battery capacity decay, and generating a warning signal when the capacity prediction result meets the preset warning conditions, includes: The capacity decay pattern and real-time altitude change trend are obtained; wherein, the real-time altitude change trend is obtained by performing time series analysis on real-time altitude data, including the trend types of continuous increase, maintenance at a high level, or decrease in altitude; Based on the real-time altitude change trend and combined with the capacity decay law, the degree and type of battery capacity decay are predicted, and a capacity prediction result is generated; wherein, the degree of battery decay includes the future battery capacity value, and the decay type includes reversible decay and irreversible decay. Determine whether the capacity prediction result meets the preset warning conditions; if so, generate a warning signal; wherein, the preset warning conditions include the future battery capacity value being lower than the capacity safety threshold or the real-time capacity decay rate exceeding the decay rate threshold.

[0013] By adopting the above technical solution, the passive and delayed alarms in traditional battery management systems based on current voltage or remaining charge are upgraded to proactive and forward-looking early warnings that integrate material properties, altitude change dynamics, and degradation mechanisms. This solution achieves advance risk prediction through trend extrapolation, provides differentiated response guidance by distinguishing between reversible and irreversible degradation, and ensures comprehensive risk coverage through a dual-threshold triggering mechanism of absolute capacity and degradation rate. The resulting standardized early warning signal provides reliable and timely decision input for subsequent accurate, tiered, and actionable user warnings, fundamentally enhancing the proactive defense capabilities of mobile devices against battery performance degradation risks in dynamic altitude environments.

[0014] Optionally, the step of classifying warning levels based on the warning signals, generating graded warning information, and pushing it to terminal devices includes: Obtain early warning signals and associated battery material types; Based on the warning signal and battery material type, a preset classification rule is used to determine the warning level; Based on the warning level, a graded warning information is generated, which includes altitude status, capacity prediction results and response instructions. If the attenuation type is reversible attenuation, a response instruction to reduce equipment power consumption or adjust altitude is generated. If the attenuation type is irreversible attenuation, a response instruction to suspend operation or replace the battery is generated. The graded early warning information is pushed to the corresponding terminal device according to the communication protocol type of the terminal device.

[0015] By adopting the above technical solutions, the accuracy of risk assessment is ensured by utilizing differentiated classification rules based on material sensitivity. The effectiveness of the response is enhanced by combining attenuation types to generate corresponding instructions. Finally, protocol adaptation ensures the robustness and coverage of early warning information transmission. This enables end-users such as drone pilots and high-altitude equipment operators to obtain action guidance most relevant to the actual risk situation of the battery at the right time through the most convenient channels. This effectively transforms the results of complex algorithmic analysis into on-site execution capabilities to ensure safe equipment operation and prevent accidents, thus improving the practical value and user experience of the entire early warning system.

[0016] Optionally, the steps of obtaining user response feedback data after receiving the graded early warning information, updating the historical altitude capacity attenuation benchmark data, and optimizing the ternary correlation model parameters include: After obtaining user response to graded early warning information, the feedback data includes altitude change information, battery capacity recovery status, and capacity decay type verification results; Based on the lithium battery material type in the disposal feedback data, match the corresponding entry in the historical altitude capacity decay benchmark data; Based on the altitude change information and battery capacity recovery status in the disposal feedback data, the historical altitude capacity decay benchmark data is updated based on the capacity decay type verification results. Based on the updated historical altitude capacity decay baseline data, the optimization operation of the ternary correlation model parameters is triggered.

[0017] By adopting the above technical solution, user responses to warnings and their objective results are transformed into learning material for the system. This method allows the historical benchmark database to continuously approximate real and complex operating conditions, while also driving the synchronous iteration of the core warning model. This feedback-driven closed-loop optimization mechanism not only gradually corrects cognitive biases in the initial model but also enables the system to adapt to battery aging, the discovery of new environmental patterns, and the operating habits of different users. This achieves a qualitative leap in warning capabilities, from static presets to dynamic growth, ensuring reliability, adaptability, and foresight in long-term operation.

[0018] Optionally, the method further includes: Obtain the charge / discharge cycle count and individual identifier of the lithium battery to generate an aging parameter dataset; Based on the individual identifiers and lithium battery material types in the aging parameter dataset, the corresponding aging sensitivity parameter values ​​are retrieved from the material property database. Based on the charge-discharge cycle number and aging sensitivity parameter value, the aging compensation factor is calculated and generated by calling the preset attenuation rate calculation formula. The aging compensation factor is applied to the output of the ternary correlation model to generate an adjusted capacity decay law. Based on the adjusted capacity decay law, the deviation rate between the predicted capacity decay value and the actual measured capacity value is calculated. If the deviation rate exceeds the preset tolerance threshold, then the warning trigger threshold in the preset warning conditions is updated.

[0019] By adopting the above technical solution, the initial model is personalized for aging correction using the battery's cycle history and material property database, so as to more accurately reflect the degradation characteristics of the battery in the current life cycle. By continuously comparing the predicted value of the corrected model with the actual measured value, and using this deviation as a basis, the final warning trigger threshold is dynamically adjusted, thereby constructing an intelligent early warning system that can continuously optimize itself as the battery's performance deteriorates throughout its entire life cycle.

[0020] Secondly, this application provides a lithium battery altitude capacity decay early warning system based on material properties, employing the following technical solution: The parameter configuration module is used to configure the lithium battery material type and preset warning conditions, and to obtain historical altitude capacity decay benchmark data that matches the lithium battery material type from the material property database. The initialization module is used to initialize the communication protocol and acquisition frequency of the data acquisition device. The data acquisition module is used to synchronously acquire real-time altitude data, battery capacity data, and auxiliary data through the data acquisition device, and add a unified timestamp to generate the raw dataset; The data processing and feature extraction module is used to preprocess the original dataset and extract features based on the historical altitude capacity decay benchmark data to generate a dataset that integrates benchmark features. The data standardization and transformation module is used to perform standardization transformation on the merged dataset to generate a standardized analysis dataset. The model building and analysis module is used to build a ternary correlation model of lithium battery material type, altitude and battery capacity based on the standardized analysis dataset and through machine learning algorithms, quantify the influence weight of altitude change on battery capacity, and output the capacity decay law of the current altitude range. The capacity degradation trend prediction module predicts the degree and type of battery capacity degradation based on the capacity degradation law and the real-time altitude change trend. The early warning signal generation module is used to generate an early warning signal when the capacity prediction result meets the preset early warning conditions. The graded early warning module is used to classify early warning levels based on the early warning signal, generate graded early warning information, and push it to the terminal device; wherein, the graded early warning information includes altitude status, capacity prediction results, and response instructions; The closed-loop feedback optimization module is used to obtain the handling feedback data after the user responds to the graded early warning information, update the historical altitude capacity attenuation benchmark data, and optimize the parameters of the ternary correlation model.

[0021] Thirdly, this application provides a computer device, which adopts the following technical solution: A computer device includes a memory, a processor, and a computer program stored in the memory, the processor executing the computer program to perform the steps of the method as described in the first aspect.

[0022] Fourthly, this application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium storing a computer program that can be loaded by a processor and executed as in any of the methods in the first aspect. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the first process of a lithium battery altitude capacity decay early warning method based on material properties, according to one embodiment of this application.

[0024] Figure 2 This is a schematic diagram of the second process of a lithium battery altitude capacity decay early warning method based on material properties, according to one embodiment of this application.

[0025] Figure 3 This is a schematic diagram of the third process of a lithium battery altitude capacity decay early warning method based on material properties according to one embodiment of this application.

[0026] Figure 4 This is a schematic diagram of the fourth process of a lithium battery altitude capacity decay early warning method based on material properties according to one embodiment of this application.

[0027] Figure 5 This is a schematic diagram of the fifth step of a lithium battery altitude capacity decay early warning method based on material properties, according to one embodiment of this application.

[0028] Figure 6 This is a schematic diagram of the sixth process of a lithium battery altitude capacity decay early warning method based on material properties according to one embodiment of this application.

[0029] Figure 7 This is a schematic diagram of the seventh process of a lithium battery altitude capacity decay early warning method based on material properties according to one embodiment of this application. Detailed Implementation

[0030] To make the purpose, technical solution, and advantages of this application clearer, the following description is provided in conjunction with the appendix. Figures 1-7 The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.

[0031] This application discloses a method for early warning of lithium battery capacity degradation based on material properties at altitude.

[0032] Reference Figure 1 A method for early warning of capacity degradation of lithium batteries based on material properties at altitude, specifically including: Step S101: Configure the lithium battery material type and preset warning conditions, and obtain historical altitude capacity decay benchmark data that matches the lithium battery material type from the material characteristic database; Lithium-ion batteries with different electrochemical systems (such as ternary lithium NCM / NCA and lithium iron phosphate LFP) exhibit drastically different response mechanisms and tolerances to low-pressure environments due to differences in their positive and negative electrode active materials, electrolyte formulations, and internal structures. For example, ternary lithium batteries have high energy density, but the electrode-electrolyte interface may be more unstable at high altitudes, leading to accelerated capacity decay; while lithium iron phosphate batteries have better thermal stability, and their decay curve may be smoother. Without distinguishing material types, all warnings will be based on a fuzzy, generic model, resulting in inaccurate predictions. Therefore, the system must first clearly identify the material type.

[0033] Secondly, the configuration of preset warning conditions includes setting a capacity safety threshold and a degradation rate threshold. The capacity safety threshold is set as a percentage of the lithium battery's rated capacity, and the degradation rate threshold is set as the capacity degradation rate corresponding to a unit change in altitude. These thresholds are not fixed values, but rather strategic parameters that can be dynamically adjusted according to the specific equipment's safety redundancy requirements.

[0034] In addition, the historical altitude capacity degradation benchmark data stored in the material properties database comes from the verified altitude-capacity correlation curves and statistical data accumulated from a large number of historical operations of lithium batteries of the same model or material system. It provides a reliable normal degradation profile or reference baseline for the current analysis for this specific material type.

[0035] Step S102: Initialize the communication protocol and acquisition frequency of the data acquisition device, and synchronously acquire real-time altitude data, battery capacity data and auxiliary data through the data acquisition device, and add a unified timestamp to generate the original dataset; Initializing the communication protocol and acquisition frequency are technical guarantees to ensure that the subsequent data acquisition process can be executed stably, efficiently, and with low latency, providing a high-quality, synchronous data source for the entire analysis chain.

[0036] Specifically, altitude change is the core environmental driver variable, measured by a high-precision barometric pressure sensor; battery capacity data (including remaining capacity, degradation rate, and cycle count) characterizes the battery's "health status," calculated and provided by the battery management system (BMS); auxiliary data (such as temperature, current, voltage, and charge / discharge cycle count) are important covariates or confounding factors used to distinguish whether capacity changes originate from altitude effects, the charging / discharging process, or temperature variations. Only by adding strictly synchronized timestamps to all these data points sampled at different physical interfaces and frequencies can the accuracy of the fact that "at time T, at altitude H, the battery capacity is C, and the temperature is T" be ensured. If the data is not synchronized, it will be impossible to distinguish whether the capacity change stems from the immediate effects of altitude, the hysteresis effect of temperature, or a natural consequence of the charging / discharging process.

[0037] Ultimately, the generated original dataset is a multivariate time series containing time, environmental, electrical, and state dimensions, which truly records the complete operational profile of lithium batteries under dynamic altitude conditions.

[0038] Step S103: Perform preprocessing on the original dataset and extract features based on historical altitude capacity decay benchmark data to generate a dataset that integrates benchmark features. The preprocessing operations include aligning timestamps, removing outliers caused by sensor transient failures or communication interference, and filling in reasonable missing values. This ensures the cleanliness and consistency of the data and eliminates noise interference for subsequent analysis.

[0039] Next, leveraging the prior knowledge inherent in historical altitude capacity degradation benchmark data, more discriminative fusion features are constructed from the current real-time data. For example, the system can calculate the deviation between the current real-time capacity degradation rate and the median of the historical benchmark degradation rate for the same material and temperature as a feature; or, it can map the current altitude to "sensitive altitude range" markers (such as "low-sensitivity zone," "transition zone," and "high-sensitivity zone") in the historical benchmark data as a classification feature. The resulting dataset not only contains the original voltage, current, and altitude values ​​but also higher-order features derived from historical knowledge, such as "the degree of deviation from historical normal levels" and "the typical risk range in which it is located."

[0040] Step S104: Perform a standardization transformation on the merged dataset to generate a standardized analysis dataset; The fused dataset may contain features with entirely different scales: altitude may be in the thousands of meters, battery voltage in the single or double digits of volts, capacity decay rate in decimal percentages, and temperature in degrees Celsius. If these features are directly input into the model, features with large numerical ranges (such as altitude) may dominate the model calculations, masking the influence of features with small numerical ranges but potentially important effects (such as a specific voltage fluctuation pattern).

[0041] In this embodiment, the standardization transformation can map all features to the same scale (e.g., a distribution with a mean of 0 and a variance of 1, or the interval [0,1]) through mathematical transformations (such as Z-score standardization, max-min normalization). For example, the altitude can be mapped from the actual [0,5000] meters to the range [0,1], and the capacity decay rate can be converted from a percentage to a numerical value of the same scale. The standardized analysis dataset generated by this step has all features on a comparable order of magnitude. This ensures that subsequent machine learning algorithms can learn and make decisions based on the true intrinsic patterns of the features, rather than their original numerical values, thereby improving the convergence speed, stability, and final prediction accuracy of the model.

[0042] Step S105: Based on the standardized analysis dataset, a ternary correlation model of lithium battery material type, altitude and battery capacity is constructed through machine learning algorithm to quantify the influence weight of altitude change on battery capacity and output the capacity decay law of the current altitude range. In this embodiment, the ternary correlation model constructed based on machine learning algorithms (such as Gradient Boosting Tree (GBDT)) takes "lithium battery material type" (as a category feature), "altitude," and other standardized covariate features as inputs, and trains on "battery capacity" or "capacity decay rate" as the prediction target. The model can evaluate the weight of the impact of a unit change in altitude on battery capacity decay, while keeping other factors (such as temperature and material) constant. This weight is not a single linear coefficient, but may be a function that dynamically changes with altitude range and material type.

[0043] Finally, the system inputs the real-time collected and processed data into this trained "ternary correlation model." The model, combining the current material type and specific conditions such as altitude and temperature, "calculates" or "retrieves" the "degradation law" that the battery capacity should follow under this specific scenario. This degradation law describes a predictive function or probability distribution of how the capacity will dynamically evolve with future altitude changes in the current state, providing a mathematical basis for further prediction.

[0044] Step S106: Based on the capacity decay law and the real-time altitude change trend, predict the degree and type of battery capacity decay, and generate a warning signal when the capacity prediction result meets the preset warning conditions. The model outputs a capacity decay pattern that provides an immediate correlation between capacity and variables such as altitude. The system, combined with real-time altitude change trends (e.g., a drone climbing at 5 meters per second, or a vehicle heading towards a higher altitude), can extrapolate or simulate this pattern over time. Through this trend extrapolation, the system can predict the future capacity level within a certain timeframe (next minute, next waypoint), rather than simply reporting the current value.

[0045] At the same time, it is also necessary to distinguish between different types of capacity reduction: "Reversible capacity reduction" mainly refers to the capacity decrease caused by temporary changes in the physical properties of the electrolyte and increased internal resistance of the battery due to low air pressure. This capacity can be partially or fully recovered when the altitude decreases and the air pressure recovers. "Irreversible capacity reduction," on the other hand, involves chemical reactions such as damage to the electrode material structure and permanent loss of active materials, and cannot be recovered even when returning to a lower altitude. Differentiating between different types of capacity reduction is crucial for the urgency of early warnings and for subsequent handling recommendations.

[0046] Next, the system compares the predicted "future capacity value" with the "capacity safety threshold" set in the preset warning conditions, or compares the calculated "real-time capacity decay rate" with the "decay rate threshold". Once the prediction result touches or exceeds these preset warning lines, that is, "the preset warning conditions are met", the system generates a clear warning signal, which indicates that it has entered the alarm response state from the analysis state.

[0047] Step S107: Based on the warning signal, the warning level is divided, a graded warning information is generated and pushed to the terminal device; wherein, the graded warning information includes altitude status, capacity prediction results and response instructions; The logic behind this step lies in differentiating risk responses and specifying operational guidelines. A single alarm method (such as a buzzer) can easily confuse operators or lead to inappropriate responses in complex operating conditions. A tiered early warning mechanism, however, classifies warning signals into different levels based on the severity of the risk they represent (typically based on the gap between predicted capacity and safety thresholds, attenuation rate, and whether the attenuation is reversible). These levels include, for example, alert level (Level 1), attention level (Level 2), and emergency level (Level 3). Each level corresponds to different levels of information detail and operational urgency.

[0048] Ultimately, the generated tiered early warning information is a structured message package containing at least three elements: 1. Altitude status: current and predicted altitude, clearly identifying the source of risk; 2. Capacity prediction results: current capacity, predicted future capacity, and attenuation type (reversible / irreversible), clearly conveying the nature and extent of the risk; 3. Response instructions: this is the ultimate manifestation of the information's value, ranging from simple "suggest paying attention to power consumption" (Level 1) to explicit "suggest reducing operating altitude or power" (Level 2), and finally to mandatory "immediately stop operations, return to a lower altitude area, or prepare emergency power" (Level 3). This tiered early warning information, accompanied by specific instructions, is pushed to users through terminal devices such as displays and mobile apps, transforming complex model analysis conclusions into clear, immediately understandable, and executable operational guidelines, greatly improving the practicality of the early warning and the efficiency of human-computer interaction.

[0049] Step S108: Obtain the handling feedback data after the user responds to the graded early warning information, update the historical altitude capacity attenuation benchmark data and optimize the ternary correlation model parameters.

[0050] The system continuously collects and acquires feedback data generated after the warning is issued, based on the user's (or the automatic control system's) actions (such as reducing flight altitude, reducing load, or returning to base for charging). This includes the actual trajectory of battery capacity changes under the new operating strategy, subsequent changes in altitude, and the final equipment status.

[0051] Next, the system cleans and labels this new data as necessary, then updates the historical altitude capacity degradation benchmark database. Simultaneously, this new data is also used as incremental training samples, input into the machine learning model to optimize the parameters of the ternary correlation model. Through continuous training, the model can constantly adjust its internal understanding of the "material-altitude-capacity" correlation, making its predictions more closely aligned with the constantly changing state of battery health (SOH) and new environmental patterns encountered in actual use.

[0052] Understandably, this closed-loop mechanism enables the system to learn from actual operation, becoming more accurate with use and gradually adapting to long-term changes such as battery aging and different climatic conditions, thereby realizing the evolution from a static, preset early warning system to a dynamic, self-learning intelligent early warning system.

[0053] In the above embodiments, the performance management of lithium batteries in high-altitude environments is upgraded from passive monitoring to proactive, material-characteristic-based, and predictive with self-learning capabilities. This technical solution can not only accurately quantify the altitude sensitivity differences of batteries with different materials, predict capacity decay risks in advance, and distinguish between reversible and irreversible losses, but also provide clear response strategies through graded early warning. Finally, through a closed-loop learning mechanism, it continuously improves the system's adaptability and accuracy, thereby enhancing the operational safety and mission reliability of mobile equipment such as drones and high-altitude operation equipment in dynamic altitude environments.

[0054] Reference Figure 2 As one implementation of step S103, the steps of preprocessing the original dataset and extracting features based on historical elevation capacity attenuation benchmark data to generate a dataset fused with benchmark features include: Step S201: Obtain the raw dataset, including timestamped altitude data, battery capacity data, and auxiliary data; Among these data, altitude data, collected by high-precision barometric pressure sensors, is the core environmental driving variable; battery capacity data, typically estimated by the battery management system (BMS) through coulomb counters or models, includes remaining capacity and state of health (SOH), and is the core state variable reflecting battery performance; auxiliary data such as temperature, current, and voltage are important covariates or confounding factors. Adding timestamps to all these data points to record the precise moment each data point was generated is a prerequisite for subsequent multi-source data fusion and causal correlation analysis.

[0055] Step S202: Perform timestamp alignment processing on the original dataset to generate a time-synchronized dataset; Because altitude sensors, BMS, and temperature sensors may have different sampling start times, sampling frequencies, and data reporting delays, the timestamps of altitude, capacity, and temperature data generated at the same physical moment (e.g., T0) in the directly acquired "raw dataset" may have slight but not negligible differences. If these unaligned data are used directly for analysis, it will lead to an incorrect association between "altitude at time A" and "capacity at time B," severely interfering with the discovery of the "altitude-capacity" correlation.

[0056] In the embodiments of the present application, the timestamp alignment process usually takes the time series of a certain type of data (such as altitude data as the core driving variable) as the reference time axis, and through the linear interpolation algorithm, interpolates the values of other data series (such as battery capacity, temperature) to each timestamp of the altitude data points. For example, if the BMS reports capacity values C1 and C2 at times T1 and T2 (T1 < T0 < T2), and the capacity value C0 at time T0 is required, then C0 can be calculated by linear interpolation. The time synchronization data set generated after this processing contains the values of altitude, capacity, and all auxiliary data that are completely synchronized at this moment.

[0057] Step S203, based on the preset anomaly detection rules, remove the outliers from the time synchronization data set to generate a cleaned data set; Among them, the preset anomaly detection rules need to take into account both general statistical laws and specific domain knowledge. For example, for altitude data, the 3σ (three-sigma) rule based on a sliding time window is adopted: calculate the mean and standard deviation of the data within the window. If the current value deviates from the mean by more than 3 times the standard deviation, it is considered an extremely unlikely outlier statistically and is very likely to be caused by a transient sensor failure. This method can adapt to the fluctuation level of the data itself and is more robust than a fixed threshold.

[0058] For battery capacity data, the knowledge of material characteristics is introduced for constraint: within a fixed time interval, if the change rate of the battery capacity (such as an instantaneous jump) exceeds the "historical maximum change rate" of this type of battery in the material characteristic database, it is regarded as an abnormal point that violates the electrochemical physical law. For example, the capacity of a lithium battery cannot suddenly drop by 20% within 1 second. Such data points, regardless of their numerical values, should be removed.

[0059] By executing these rules, the untrustworthy noise points are filtered out from the time synchronization data set, and the generated cleaned data set has a higher signal-to-noise ratio and physical consistency. This is a crucial step to ensure that subsequent machine learning models or analysis algorithms are not biased by abnormal data, thus obtaining stable and reliable conclusions.

[0060] Step S204, extract the material sensitivity feature and decay trend feature from the historical altitude-capacity decay reference data; Among them, the material sensitivity feature aims to quantify the difference in the response intensity of different types of lithium batteries to altitude changes. From the reference data, features such as the altitude sensitivity threshold (for example, the critical point at which the capacity decay of a certain type of ternary lithium battery starts to accelerate significantly after the altitude exceeds 3000 meters) and the sensitive interval grading parameter (for example, defining 0 - 2500 meters as the "insensitive zone", 2500 - 4000 meters as the "transition zone", and above 4000 meters as the "highly sensitive zone") can be extracted. These features provide prior labels about the inherent properties of battery materials for the model.

[0061] The decay trend feature aims to characterize the dynamic law of capacity variation with altitude. This can be achieved by extracting the "capacity decay rate distribution curves corresponding to different altitude ranges" from the benchmark data. For example, for a certain material, it can be modeled that "within the altitude range [3500, 4000) meters, the capacity decay rate follows a normal distribution with a mean of 0.7% / 100m and a standard deviation of 0.1%". More complex models can extract the decay curve function (such as the coefficients of piecewise linear or polynomial fitting) across the entire altitude range. These trend features encode historical statistical laws into mathematical expressions.

[0062] Step S205: The material sensitivity features and attenuation trend features are fused into the cleaned dataset to generate a dataset with fused baseline features.

[0063] Specifically, firstly, material sensitivity features can be directly added as dimensional identifiers. For example, the sensitivity range classification of the current battery (e.g., "high sensitivity range") can be incorporated as a new classification feature field into each record of the cleaned dataset, providing the model with important contextual labels.

[0064] Next, the extracted "capacity decay rate distribution across different altitude ranges" is transformed into an altitude-capacity decay weighting coefficient matrix. Each row (or element) of this matrix corresponds to an altitude range, and its value represents the "expected intensity" or "risk weight" of capacity decay at that altitude based on historical patterns. During feature fusion, the system reads the altitude values ​​of real-time data points in the cleaned dataset, maps them to the aforementioned weighting coefficient matrix, and generates a weighted decay feature value through matrix multiplication or table lookup operations. This feature value represents the expected degree of capacity decay (or risk index) at the current altitude based on historical experience.

[0065] Finally, this calculated weighted decay feature value is added as a new field and merged into the original cleaned dataset. The resulting dataset, which integrates baseline features, contains not only the original real-time observations (altitude, volume, temperature, etc.) for each record, but also contextual features derived from historical knowledge that are strongly correlated with the current state (such as sensitivity level and historically based expected decay weights). This enhances the information density and interpretability of the dataset, enabling subsequent machine learning models to learn from both real-time data and historical experience, thereby improving their predictive accuracy and generalization ability.

[0066] The above implementation systematically solves the problems of low data quality, limited information dimensions, and disconnect between historical experience and real-time status in lithium battery monitoring scenarios under altitude change conditions. The resulting dataset not only ensures precise time synchronization and physical authenticity of the data, but also deeply embeds the inherent altitude sensitivity of the material and historical statistical attenuation patterns into the real-time observation data in a calculable and quantifiable feature form through feature engineering. This provides subsequent early warning models with a much richer and more discriminative input than the original signal or simply cleaned data, fundamentally improving the accuracy and reliability of subsequent capacity attenuation pattern mining, trend prediction, and risk assessment. It is a solid data foundation for the entire intelligent early warning system to achieve accurate analysis and early warning.

[0067] Reference Figure 3 As one implementation of step S105, the steps of constructing a ternary correlation model between lithium battery material type, altitude, and battery capacity based on a standardized analysis dataset and a machine learning algorithm, quantifying the influence weight of altitude changes on battery capacity, and outputting the capacity decay law of the current altitude range include: Step S301: Obtain a standardized analysis dataset, including lithium battery material type, altitude data, battery capacity data, and auxiliary data; Among these factors, the type of lithium battery material is a prerequisite for the model to conduct differential analysis, ensuring that the model can learn the intrinsic differences in characteristics of different electrochemical systems; altitude, as a core environmental driving factor, is a key independent variable that the model needs to interpret; battery capacity is the target variable or dependent variable that the model predicts; auxiliary data such as battery surface temperature, ambient temperature, charge-discharge cycle number, current, and voltage can be used as covariates, which are crucial for separating the influence of non-altitude factors (such as temperature effect and battery aging) on ​​capacity, and can help the model capture the "altitude-capacity" relationship more purely.

[0068] It should be noted that "standardization" means that the numerical features in the dataset (such as altitude, temperature, and volume) have been processed to eliminate differences in units and orders of magnitude, in order to improve the convergence speed and stability of subsequent machine learning algorithms and avoid some algorithms from being biased due to excessive differences in feature scale.

[0069] Step S302: Based on the standardized analysis dataset, a ternary correlation model of lithium battery material type, altitude data and battery capacity data is trained using a machine learning algorithm; wherein, the machine learning algorithm is a gradient boosting tree or support vector machine, and the training includes fitting the nonlinear relationship between material type, altitude and battery capacity through an iterative optimization algorithm, and using auxiliary data as input feature variables to optimize the fitting process; The underlying principle of this step is to leverage the powerful nonlinear fitting and pattern recognition capabilities of machine learning algorithms to automatically mine and mathematically represent the complex, interactive, and nonlinear mapping function relationships between the three core elements of "battery material type," "altitude," and "battery capacity" from massive, multi-dimensional standardized data.

[0070] Specifically, machine learning algorithms (such as gradient boosting trees or support vector machines) adjust a large number of parameters within the model through iterative optimization algorithms (such as gradient descent) to minimize the difference between the model's predicted values ​​and the actual battery capacity values ​​in the dataset (i.e., the loss function).

[0071] In this process, the algorithm learns autonomously that for "material type A", an elevation change from H1 to H2 may result in a capacity decrease of ΔC1; while for "material type B", the same elevation change may only lead to a decrease of ΔC2 (ΔC1 ≠ ΔC2). Simultaneously, it also learns how auxiliary parameters such as temperature modulate this decrease relationship.

[0072] Ultimately, the trained ternary correlation model is essentially a set of mathematical functions or decision rules that encapsulate all these complex learned patterns. When a new data sample is input (containing material type, current altitude, temperature, etc.), this model can output an accurate estimate or prediction of the battery capacity under those conditions.

[0073] Step S303: Use a ternary correlation model to analyze the capacity decay rate corresponding to the unit of altitude change and calculate the weight of the impact of altitude change on battery capacity; wherein, the unit of altitude change is preset to a fixed value. Specifically, all other variables in the model input can be fixed (e.g., material type set to a specific model, temperature set to the current value), and then the value of the "altitude" feature can be systematically changed (e.g., simulating an increase in altitude from the current value H by a preset unit ΔH, such as 100 meters, in the model input), and the magnitude of change (ΔC) in the predicted battery capacity output by the model can be observed. This change ΔC (or the normalized ratio ΔC / ΔH) can be quantified as the "weight of the impact of altitude change on battery capacity," that is, "under the current comprehensive conditions, if the altitude increases by another ΔH meters, the battery capacity is expected to decrease by what percentage?"

[0074] It should be noted that this weight is not a global constant, but dynamic: it varies with the current altitude, material type, and temperature (for example, the influence weight is greater at an altitude of 4000 meters than at an altitude of 1000 meters).

[0075] Step S304: Based on the influence weight, determine the altitude sensitivity threshold of lithium batteries of different material types, and output the capacity decay law of the current altitude range.

[0076] The capacity decay law can be specified in several forms. One key form is determining the altitude sensitivity threshold for lithium batteries of different materials. This threshold is defined as the critical altitude point at which the capacity decay rate significantly increases. The system can analyze the curve of the "influence weight" changing with altitude to find the inflection point where the weight value changes abruptly. The altitude corresponding to this inflection point can be defined as the "altitude sensitivity threshold," that is, the critical altitude at which the capacity decay rate begins to accelerate significantly. For example, for a certain type of ternary lithium battery, the system may output its sensitivity threshold as 3200 meters.

[0077] Another key form is to describe a quantitative relationship or curve showing how capacity changes with altitude within a given altitude range. For example, outputting "Within the altitude range [3000, 4000) meters, the battery capacity degradation rate is approximately 0.6% per 100 meters" or providing a degradation fitting function for that range. Furthermore, the pattern can also include a trend indicating the type of degradation (reversible / irreversible).

[0078] In the above embodiments, a high-order correlation model integrating material properties, altitude, environmental covariates and battery capacity was established. The influence weight of altitude changes can be dynamically extracted from the model, and then output the capacity decay law, which provides a reliable and quantifiable decision basis for realizing a forward-looking lithium battery altitude capacity decay early warning based on material properties. This improves the accuracy and intelligence of battery status assessment and risk management in dynamic altitude application scenarios such as drones and plateau equipment.

[0079] Reference Figure 4 As one implementation of step S106, the step of predicting the degree and type of battery capacity decay based on the capacity decay law and real-time altitude change trend, and generating a warning signal when the capacity prediction result meets the preset warning conditions, includes: Step S401: Obtain the capacity decay pattern and real-time altitude change trend; wherein, the real-time altitude change trend is obtained by performing time series analysis on the real-time altitude data, including the trend type of continuous increase, maintenance at a high level or decrease in altitude. The specific form of the capacity decay law can be the quantitative decay rate of the current battery material in the relevant altitude range (such as the percentage decay per 100 meters), the altitude sensitivity threshold, or a functional relationship describing the change of capacity with altitude. It represents the "physical law" about the performance change of the battery in the current environment, based on historical data and real-time analysis.

[0080] Meanwhile, the acquired "real-time altitude change trend" is an immediate perception of the dynamic environment in which the equipment is located. It can be obtained by performing time series analysis (such as slope calculation and moving average) on the continuously collected altitude data sequence to determine whether the equipment is in a state of "continuous climbing", "lingering at a high level" or "descending".

[0081] Understandably, by combining the static "capacity decay law" that reflects general correlations with the dynamic "altitude change trend" that describes the real-time movement of specific equipment, a framework for dynamic deduction under known rules is constructed.

[0082] Step S402: Based on the real-time altitude change trend and combined with the capacity decay law, predict the degree and type of battery capacity decay and generate capacity prediction results; wherein, the degree of battery decay includes the future battery capacity value, and the decay type includes reversible decay and irreversible decay. Specifically, the system integrates or extrapolates the capacity decay pattern over time based on real-time altitude changes (e.g., the drone is climbing at a constant speed of 2 meters per second). For example, if the pattern indicates that the capacity decay rate is approximately 0.8% per 100 meters within the current altitude range, and the device is continuously ascending, the system can predict the specific percentage by which the battery capacity will decrease as the altitude increases by a certain cumulative value in the future (to the next waypoint or a few minutes later), thus generating a quantitative prediction of the degree of decay.

[0083] Furthermore, predicting the type of degradation—that is, distinguishing between reversible and irreversible degradation—requires a comprehensive assessment combining deeper knowledge of underlying principles (such as triggering condition models for different degradation mechanisms) and real-time operating conditions (such as current temperature and historical battery stress). Reversible degradation is mainly related to physical processes such as changes in electrolyte properties and temporary increases in interfacial impedance under low atmospheric pressure; predicting its occurrence means that capacity is likely to recover once altitude decreases. Irreversible degradation, on the other hand, is related to material loss such as structural damage to electrode active materials and irreversible chemical reactions; predicting its occurrence means that capacity loss is permanent.

[0084] It should be noted that the prediction of the attenuation type provides a key decision-making basis for subsequent early warning classification, because it directly affects the urgency and content of the response strategy (whether to temporarily avoid or to maintain and repair).

[0085] Step S403: Determine whether the capacity prediction result meets the preset warning conditions; if yes, proceed to step S404; if no, return to re-execute step S401. Step S404: Generate a warning signal; wherein the preset warning conditions include the future battery capacity value being lower than the capacity safety threshold or the real-time capacity decay rate exceeding the decay rate threshold.

[0086] Among them, the preset early warning conditions are usually set as a dual threshold mechanism to ensure the comprehensiveness and sensitivity of the early warning.

[0087] First, the first condition is that the battery capacity is predicted to be lower than the capacity safety threshold. This is an absolute bottom-line protection to ensure that the device will not suddenly shut down due to power depletion under any circumstances. The system will compare the predicted battery capacity (or capacity percentage) at a future point in time with the "capacity safety threshold" (such as 20% of the rated capacity) set in advance according to the device's minimum operating requirements.

[0088] Secondly, the second condition is that the real-time capacity decay rate exceeds the decay rate threshold. This is an anomaly monitoring of the relative rate of change, designed to capture dangerous situations where the absolute capacity value has not yet bottomed out, but the decay rate is abnormally accelerating. For example, even if the current remaining capacity is 50%, if the system detects that the instantaneous decay rate is far greater than the normal model prediction value due to a sharp increase (which may indicate the onset of abnormalities such as thermal runaway), an early warning must be triggered immediately.

[0089] Understandably, this dual judgment logic combining "absolute value" and "rate of change" enables the system to both prevent gradual risks and respond to sudden risks, greatly enhancing the robustness and foresight of the early warning system.

[0090] Next, when any or all of the above triggering conditions are met, the system no longer simply records a state, but actively generates a warning signal. This signal is typically designed as a binary trigger identifier or a digital signal containing a basic event code.

[0091] The above embodiments upgrade the passive and delayed alarms in traditional battery management systems based on current voltage or remaining charge to proactive and forward-looking early warnings that integrate material properties, altitude change dynamics, and degradation mechanisms. This technical solution achieves advance risk prediction through trend extrapolation, provides differentiated response guidance by distinguishing between reversible and irreversible degradation, and ensures comprehensive risk coverage through a dual-threshold triggering mechanism of absolute capacity and degradation rate. The resulting standardized early warning signals provide reliable and timely decision input for subsequent accurate, tiered, and actionable user warnings, fundamentally enhancing the proactive defense capabilities of mobile devices against battery performance degradation risks in dynamic altitude environments.

[0092] Reference Figure 5 As one implementation of step S107, the steps of classifying warning levels based on warning signals, generating graded warning information, and pushing it to terminal devices include: Step S501: Obtain the warning signal and the associated battery material type; The warning signal is essentially a Boolean instruction, meaning "It has been determined that there is a risk requiring a warning." However, simply knowing that a warning is needed is insufficient for an effective response, because the severity and response strategies can vary drastically depending on the battery type facing the same risk. Therefore, it is essential to simultaneously obtain the relevant battery material type and input it along with the warning signal.

[0093] Step S502: Based on the warning signal and battery material type, match the preset classification rules to classify the warning level; In this embodiment, the system internally pre-defines one or more sets of hierarchical rules. These rules are not simply fixed thresholds, but rather conditional judgment logics deeply bound to the battery material type. A typical implementation is a three-level early warning system (Level 1: Reminder; Level 2: Attention; Level 3: Emergency). Specifically, Level 1 warning corresponds to scenarios where there is a slight change in altitude and the capacity decay has not reached the material-sensitive range; Level 2 warning corresponds to scenarios where the altitude enters the material-sensitive range and the capacity decay rate is close to the threshold; and Level 3 warning corresponds to scenarios where the altitude significantly exceeds the standard and the predicted capacity is about to fall below the safety threshold.

[0094] In practical applications, the system first selects a specific rule set applicable to the battery material type (for example, the rule set for ternary lithium batteries might set 3000 meters as the sensitive threshold, while the rule set for lithium iron phosphate batteries might set it to 3500 meters). Then, it combines the original risk information implied in the warning signal (such as "predicted capacity below the safety threshold" or "excessive degradation rate" determined in previous steps) with any additional real-time data (such as the current precise altitude) to perform rule matching. For example, a rule might specify: "If the material is ternary lithium, the current altitude is >3000 meters, and the predicted battery capacity will fall below the safety threshold within 5 minutes, then a Level 3 warning is applied." Furthermore, the rules can include dynamic adjustment logic, such as: "If a Level 2 warning is applied, but the altitude is still continuously increasing, then the warning level is automatically upgraded to Level 3."

[0095] Through the above steps, a simple warning signal is accurately transformed into a "warning level identifier" (such as Level 1, Level 2, Level 3) with clear implications of urgency and severity. This provides a direct basis for generating differentiated information and instructions in the future.

[0096] Step S503: Based on the warning level, generate graded warning information including altitude status, capacity prediction results and response instructions; wherein, if the attenuation type is reversible attenuation, generate response instructions to reduce equipment power consumption or adjust altitude; if the attenuation type is irreversible attenuation, generate response instructions to suspend operation or replace battery. Once the warning level is determined, the system calls the corresponding template and fills in three core elements to generate the final graded warning information. The first element is the altitude status, which clearly informs the user of the source of the risk, typically including the current altitude, the trend of change (increasing / remaining the same / decreasing), and its position relative to the material's sensitivity threshold. The second element is the capacity prediction result, which is a quantitative description of the risk consequences, including the current remaining capacity, the predicted future capacity, the decay rate, and the key decay type (reversible or irreversible).

[0097] Finally, the third aspect is the response instructions. These instructions are generated with a high degree of intelligence, taking into account both the warning level and the type of attenuation. For example, for a "Level 2 warning" determined to be "reversible attenuation," the instruction might be "It is recommended to reduce the flight altitude to below 3000 meters and reduce the camera load." For a "Level 3 warning" involving "irreversible attenuation," the instruction might be "Immediately return to base, check battery health upon landing, and prepare for replacement." This structured information generation method ensures that the output is not a simple, panic-inducing alarm, but a decision support report containing a complete logic of "current situation-prediction-action," greatly improving the practicality of the warnings and user acceptance.

[0098] Step S504: Based on the communication protocol type of the terminal device, push the graded warning information to the corresponding terminal device.

[0099] The terminal forms of lithium battery equipment are diverse, including embedded displays for drone flight control, mobile apps for ground station personnel, cloud servers in remote monitoring centers, or central control consoles for vehicle-mounted equipment. These terminal devices use different communication protocols and data interfaces, such as graphics rendering commands for screens, HTTP / HTTPS or MQTT protocols for mobile apps, and CAN bus messages for vehicle systems. Therefore, the push module must have protocol adaptation capabilities.

[0100] Before pushing the message, the system encodes and converts the generated tiered warning information according to the "communication protocol type" registered or preset by the target terminal device, encapsulating it into a data packet that conforms to the protocol specification. For example, for a flight control display screen, the information is converted into specific display frame data; for an app, it is encapsulated in JSON format and sent via wireless network. This step converts the information from a general internal system format to a terminal-specific format, ensuring that regardless of whether the information is ultimately presented in a screen pop-up, app push notification, or alarm list on a monitoring dashboard, its core content (status, results, and instructions) can be accurately parsed and displayed.

[0101] In the above implementation, the accuracy of risk assessment is ensured by utilizing grading rules based on material sensitivity differences, the effectiveness of response is improved by combining attenuation types to generate corresponding instructions, and the robustness and coverage of early warning information transmission are ensured through protocol adaptation. This enables end users such as drone pilots and high-altitude equipment operators to obtain action guidance most relevant to the actual risk status of the battery at the right time through the most convenient channels. This effectively transforms the results of complex algorithmic analysis into on-site execution capabilities to ensure safe equipment operation and prevent accidents, enhancing the practical value and user experience of the entire early warning system.

[0102] Reference Figure 6 As one implementation of step S108, the steps of obtaining the handling feedback data after the user responds to the graded early warning information, updating the historical altitude capacity attenuation benchmark data, and optimizing the ternary correlation model parameters include: Step S601: Obtain the handling feedback data after the user responds to the graded early warning information, including altitude change information, battery capacity recovery status and capacity decay type verification results; When the user (or automatic control system) receives the tiered warning information generated in the aforementioned steps, it will take corresponding action, such as reducing flight altitude, decreasing equipment power consumption, returning to base, or suspending operations. The system's task at this moment is to synchronously monitor and structurally record this series of operations and their resulting data.

[0103] In this embodiment, the acquired feedback data includes not only the identifier of the user's action (e.g., "altitude reduction operation performed"), but more importantly, objective result data generated after the action is performed: including "altitude change information after the action" (e.g., descending to 2500 meters and maintaining it for 10 minutes), "battery capacity recovery status after the action" (e.g., capacity recovered from 65% at the time of the warning to 72%), and "capacity decay type verification result" based on before-and-after comparison (confirming that the decay was indeed reversible). The value of this data lies in clearly revealing "what actual effect the specific measures produced under a certain warning situation," thus providing a direct basis for assessing the accuracy of the warning and optimizing response strategies.

[0104] Step S602: Based on the lithium battery material type in the disposal feedback data, match the corresponding entry in the historical altitude capacity degradation benchmark data; Among them, the historical altitude capacity degradation benchmark data is a structured database organized according to the core attribute of "lithium battery material type". Each material system (such as ternary lithium NCM811 and lithium iron phosphate LFP) has its own independent data partition, which stores information such as the material's unique altitude sensitivity characteristics and degradation curves.

[0105] In this embodiment, upon obtaining a set of disposal feedback data, the system first extracts the lithium battery material type information and uses it as an index key to quickly and accurately locate the specific data partition or data entry set belonging to that material type in the historical benchmark database. This ensures that subsequent data update operations can be correctly matched, adding new experience data to the correct knowledge category. For example, "high-altitude capacity recovery data" obtained from a ternary lithium battery must be updated to the ternary lithium battery dataset and cannot be mistakenly mixed into the lithium iron phosphate battery dataset. Otherwise, it will cause confusion in subsequent model learning, destroy the purity of knowledge about different battery types, and thus affect the accuracy of the early warning.

[0106] Step S603: Based on the altitude change information and battery capacity recovery status in the disposal feedback data, update the historical altitude capacity decay benchmark data based on the capacity decay type verification results. The system analyzes and processes the altitude change information (such as the process of descending from the warning altitude to the stable altitude) and battery capacity recovery status in the feedback data, and then supplements the historical benchmark database accordingly. For example, for reversible degradation scenarios, it supplements the capacity recovery rate data after altitude reduction; for irreversible degradation scenarios, it supplements the permanent capacity degradation data caused by material loss.

[0107] Specifically, for reversible degradation scenarios, the system will focus on extracting and supplementing "capacity recovery rate data after altitude reduction," such as calculating what percentage of capacity is recovered and the recovery speed after returning from a high altitude to a low altitude, which enriches the knowledge about the elasticity of battery performance. For irreversible degradation scenarios, it is necessary to supplement "permanent capacity degradation data caused by material loss," recording the amount of permanent loss that cannot be recovered even if the environment recovers, which deepens the system's understanding of the limits of battery damage.

[0108] Furthermore, before updating, a validity verification mechanism for the handling feedback data can be introduced. For example, it can check whether the user's handling actions are logically consistent with the early warning suggestions, or whether the capacity change is within the theoretically possible range of materials. This filters out invalid data generated by misoperation or sensor malfunction, ensuring the quality of the data injected into the knowledge base. Through this step, historical benchmark data is transformed from a static reference library into a dynamic knowledge body that can continuously grow and self-correct with each actual early warning and handling.

[0109] Step S604: Based on the updated historical altitude capacity attenuation baseline data, trigger the optimization operation of the ternary correlation model parameters.

[0110] The system will invoke machine learning training algorithms, using updated and richer historical altitude-based capacity degradation benchmark data as a new training set to rerun the model training process. During this process, the algorithm will recalculate and adjust the influence weights of various features (such as altitude, material type, and temperature) on battery capacity in the ternary correlation model based on the sum of the old and new data, thereby enabling the model to learn the recently verified patterns.

[0111] For example, new data might show that a certain material has a stronger recovery capability in a specific altitude range than previously thought, and the model will adjust the attenuation weight coefficients within that range accordingly. This optimization transforms the early warning model from a static snapshot at deployment into an intelligent agent capable of learning from experience and evolving throughout the entire equipment lifecycle. The optimized model parameters are immediately applied to the next correlation analysis and prediction process, forming an enhanced closed loop that continuously improves the accuracy, adaptability, and reliability of the entire system's early warning capabilities over time.

[0112] In the above implementation, user responses to warnings and their objective results are transformed into learning material for the system. This method allows the historical benchmark database to continuously approximate real and complex operating conditions, while also driving the synchronous iteration of the core warning model. This feedback-driven closed-loop optimization mechanism not only gradually corrects cognitive biases in the initial model but also enables the system to adapt to battery aging, the discovery of new environmental patterns, and the operating habits of different users. This achieves a qualitative leap in warning capabilities, from static presets to dynamic growth, ensuring reliability, adaptability, and foresight in long-term operation.

[0113] Reference Figure 7 As a further implementation of the lithium battery altitude capacity degradation early warning method, the method also includes: Step S701: Obtain the charge-discharge cycle number and individual identifier of the lithium battery, and generate an aging parameter dataset; Among them, the number of charge-discharge cycles is used to measure the degree of chemical aging of lithium batteries. Each complete charge-discharge cycle will trigger tiny, cumulative, irreversible electrochemical reactions inside the battery, such as micro-damage to the structure of positive and negative electrode active materials, continuous consumption of electrolyte, and thickening of solid electrolyte interface film. These micro-changes are directly manifested macroscopically as the gradual decay of the battery's maximum usable capacity and the increase of internal resistance.

[0114] However, relying solely on the global metric of cycle count is insufficient to differentiate between batteries of the same model and material that exhibit individual performance variations. Therefore, introducing an "individual identifier" (such as a battery serial number or a unique device ID) ensures that the system can strongly bind the abstract cycle count to a specific, physical battery individual.

[0115] By simultaneously collecting data from these two dimensions, the resulting aging parameter dataset is a tagged data record that can accurately point to "a specific aging stage of a particular battery in its life cycle".

[0116] Step S702: Based on the individual identifiers and lithium battery material types in the aging parameter dataset, retrieve the corresponding aging sensitivity parameter values ​​from the material property database; Among them, the aging sensitivity parameter value pre-stored in the material property database is a core parameter obtained by long-term cycle aging tests and statistical analysis on a large number of lithium batteries with the same material system. It is essentially the average slope of the capacity decay rate of the material system (such as high-nickel ternary NCM811, medium-nickel ternary NCM622, and lithium iron phosphate LFP) as the number of cycles increases, reflecting the inherent durability characteristics of the material chemical system.

[0117] For example, high-energy-density ternary materials may have higher aging sensitivity, meaning their capacity decays faster with cycling; while lithium iron phosphate materials, which have better stability, generally have lower aging sensitivity. The system uses an "individual identifier" and "lithium battery material type" as a composite query key to accurately locate and retrieve the material-level aging sensitivity parameter value applicable to this specific battery model from the database.

[0118] Step S703: Based on the number of charge-discharge cycles and the aging sensitivity parameter value, the preset attenuation rate calculation formula is called to calculate and generate the aging compensation factor; The logic behind this step lies in mathematically fusing the battery's actual usage history (number of charge-discharge cycles) with the inherent aging characteristics of the materials (sensitivity parameters) to dynamically generate a correction coefficient, or aging compensation factor, to quantify the impact of the current aging state on performance. This factor characterizes the factor by which the capacity decay rate exhibited by the battery under stress environments such as high altitude, due to aging, needs to be amplified relative to the expected decay rate of a brand-new (or reference-state) battery. For example, a calculated factor of 1.15 means that under the current aging state, any capacity decay effect caused by altitude needs to be additionally considered by 15%.

[0119] In the embodiments of this application, the aging compensation factor is typically a coefficient greater than or equal to 1, and its value increases with the number of cycles, with the rate of increase being adjusted by the aging sensitivity parameter. Specifically, for a battery that has undergone N cycles, its compensation factor F can be expressed as: F = 1 + α * N, where α is the aging sensitivity (material degradation coefficient).

[0120] Step S704: Apply the aging compensation factor to the output of the ternary correlation model to generate the adjusted capacity decay law. In ternary correlation models (such as machine learning models), the data samples used during training represent the average performance of batteries in their training set at their respective aging stages. When applied directly to a specific battery that has undergone many cycles and is significantly aged, the output capacity decay pattern (e.g., predicting a 0.5% capacity decrease for every 100 meters increase in altitude) may be overly optimistic because the model fails to fully account for the accelerated degradation caused by aging.

[0121] At this point, the calculated aging compensation factor is applied as a multiplicative weight to the degradation rate predicted by the model. For example, multiplying the original model output of 0.5% degradation rate by the compensation factor of 1.15 yields an adjusted degradation rate of 0.575%. The system obtains a new performance degradation prediction curve specifically weighted by the aging effect for this aging battery. This adjusted curve more accurately reflects the true trend of battery capacity change under the combined stress factors of "aging" and "altitude," providing a more reliable basis for subsequent early warning decisions.

[0122] Step S705: Based on the adjusted capacity decay law, calculate the deviation rate between the predicted capacity decay value and the actual capacity measurement value; Specifically, based on the adjusted patterns, the system calculates a predicted capacity decay value for the currently collected altitude data. Simultaneously, the battery management system (BMS) obtains an actual capacity measurement value in real-time or near real-time using methods such as coulomb counting or voltage-capacity integration.

[0123] Calculating the deviation rate between these two factors essentially quantifies the model's predictive accuracy. Even with aging compensation, predictions cannot be absolutely precise; the deviation rate reflects the model residuals. If the deviation rate remains consistently low, it indicates that the calculation and application of the "aging compensation factor" are effective. If the deviation rate systematically increases, it suggests that the current compensation model may still be insufficient to fully characterize the complex degradation behavior of the battery, or that new, unmodeled degradation factors have emerged. This provides a clear trigger signal and quantitative basis for the next step of system self-optimization.

[0124] Step S706: If the deviation rate exceeds the preset tolerance threshold, then update the warning trigger threshold in the preset warning conditions.

[0125] The preset tolerance threshold is the maximum error range allowed by the system for model prediction. When the calculated deviation rate exceeds this threshold, it means that even if the model has been compensated for aging, the degree of discrepancy between its prediction results and the actual situation has exceeded the acceptable range. At this point, simply adjusting the prediction formula within the model may be slow or overly complex. This step adopts a more direct strategy, namely, dynamically updating the warning trigger threshold.

[0126] For example, the system analyzes the direction of the deviation. If the actual degradation is always faster and more severe than the prediction (even after aging compensation), it indicates that the currently set "capacity safety lower limit threshold" or "degradation rate threshold" used to trigger the warning is too lenient and may not be able to provide timely warnings before a real danger occurs. Therefore, the system will automatically tighten these thresholds, such as raising the threshold for triggering a low-capacity warning from 70% to 75% of the rated capacity, thereby "offsetting" the optimistic bias of the model prediction and ensuring that the safety boundary is always effective.

[0127] It should be noted that this update is not arbitrary, but based on quantitative analysis of the deviation rate. Its essence is to adjust the boundary conditions of the decision in reverse according to the long-term performance of the model prediction, so that the entire early warning system can still ensure the safety of the final application by passing the dynamic safety threshold even if there are residual errors in the model prediction.

[0128] In the above implementation, the initial model is personalized for aging correction by utilizing the battery's cycle history and material property database to more accurately reflect the degradation characteristics of the battery's current life cycle. By continuously comparing the predicted values ​​of the corrected model with the actual measured values, and using this deviation as a basis, the final warning trigger threshold is dynamically adjusted, thereby constructing an intelligent early warning system that can continuously self-optimize as the battery's performance deteriorates throughout its entire life cycle.

[0129] In practical applications, this technical solution solves the industry problem that the accuracy of warnings from traditional static models gradually declines when faced with the complex aging process of batteries. It upgrades battery safety management from passive protection that relies on fixed rules to active intelligent protection with the ability to sense, learn, and adjust, thereby enhancing the reliability and safety of battery systems in long-term operation in harsh environments such as high altitudes.

[0130] This application also discloses a lithium battery altitude capacity decay early warning system based on material properties.

[0131] A lithium battery altitude-based capacity degradation early warning system based on material properties, specifically comprising: The parameter configuration module is used to configure the lithium battery material type and preset warning conditions, and to obtain historical altitude capacity decay benchmark data that matches the lithium battery material type from the material property database. The initialization module is used to initialize the communication protocol and acquisition frequency of the data acquisition device. The data acquisition module is used to synchronously collect real-time altitude data, battery capacity data, and auxiliary data through data acquisition devices, and add a unified timestamp to generate the raw dataset; The data processing and feature extraction module is used to preprocess the original dataset and extract features based on historical altitude capacity decay benchmark data to generate a dataset that integrates benchmark features. The data standardization and transformation module is used to perform standardization transformation on the merged dataset to generate a standardized analysis dataset. The model building and analysis module is used to build a ternary correlation model of lithium battery material type, altitude and battery capacity based on a standardized analysis dataset and through machine learning algorithms. It quantifies the influence weight of altitude change on battery capacity and outputs the capacity decay law of the current altitude range. The capacity degradation trend prediction module predicts the degree and type of battery capacity degradation based on the capacity degradation pattern and real-time altitude change trend. The early warning signal generation module is used to generate an early warning signal when the capacity prediction result meets the preset early warning conditions. The graded early warning module is used to classify early warning levels based on early warning signals, generate graded early warning information, and push it to terminal devices; the graded early warning information includes altitude status, capacity prediction results, and response instructions. The closed-loop feedback optimization module is used to obtain the handling feedback data after the user responds to the graded early warning information, update the historical altitude capacity attenuation benchmark data, and optimize the parameters of the ternary correlation model.

[0132] The lithium battery altitude capacity decay early warning system based on material properties according to the embodiments of this application can implement any of the above methods, and the specific working process of each module in the system can refer to the corresponding process in the above method embodiments.

[0133] In the several embodiments provided in this application, it should be understood that the provided methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for example, the division of a certain module is merely a logical functional division, and in actual implementation there may be other division methods, such as multiple modules can be combined or integrated into another system, or some features can be ignored or not executed.

[0134] This application also discloses a computer device.

[0135] A computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a lithium battery altitude capacity decay early warning method based on material properties as described above.

[0136] This application also discloses a computer-readable storage medium.

[0137] A computer-readable storage medium storing a computer program that can be loaded by a processor and executed as described above in any of the methods for early warning of lithium battery capacity degradation based on material properties at altitude.

[0138] The computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device; the program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0139] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.

Claims

1. A method for early warning of capacity degradation of lithium batteries based on material properties at altitude, characterized in that, The method includes: Configure lithium battery material types and preset warning conditions, and obtain historical altitude capacity decay benchmark data that matches the lithium battery material type from the material property database; Initialize the communication protocol and acquisition frequency of the data acquisition device, and synchronously acquire real-time altitude data, battery capacity data and auxiliary data through the data acquisition device, add a unified timestamp, and generate the original dataset; Preprocessing is performed on the original dataset, and feature extraction is performed based on the historical elevation capacity decay benchmark data to generate a dataset that integrates benchmark features; Perform a standardization transformation on the merged dataset to generate a standardized analysis dataset; Based on the standardized analysis dataset, a ternary correlation model of lithium battery material type, altitude and battery capacity is constructed through machine learning algorithms to quantify the influence weight of altitude change on battery capacity and output the capacity decay law of the current altitude range. Based on the capacity decay law and the real-time altitude change trend, the degree and type of battery capacity decay are predicted, and a warning signal is generated when the capacity prediction result meets the preset warning conditions. Based on the warning signals, warning levels are divided, graded warning information is generated and pushed to terminal devices; wherein, the graded warning information includes altitude status, capacity prediction results and response instructions; Obtain user response feedback data after receiving the graded early warning information, update the historical altitude capacity attenuation benchmark data, and optimize the parameters of the ternary correlation model.

2. The method for early warning of lithium battery capacity degradation based on material properties according to claim 1, characterized in that, The steps of preprocessing the original dataset and extracting features based on the historical elevation capacity attenuation benchmark data to generate a dataset that fuses the benchmark features include: Obtain the raw dataset, including timestamped altitude data, battery capacity data, and auxiliary data; Perform timestamp alignment processing on the original dataset to generate a time-synchronized dataset; Based on preset anomaly detection rules, outliers are removed from the time synchronization dataset to generate a cleaned dataset. Material sensitivity characteristics and degradation trend characteristics are extracted from the historical altitude capacity degradation benchmark data; The material sensitivity features and attenuation trend features are fused into the cleaned dataset to generate a dataset with fused baseline features.

3. The method for early warning of lithium battery capacity degradation based on material properties according to claim 2, characterized in that, Based on the standardized analysis dataset, the steps of constructing a ternary correlation model between lithium battery material type, altitude, and battery capacity using machine learning algorithms, quantifying the impact weight of altitude changes on battery capacity, and outputting the capacity decay pattern within the current altitude range include: Obtain standardized analysis datasets, including lithium battery material types, altitude data, battery capacity data, and auxiliary data; Based on the standardized analysis dataset, a ternary correlation model of lithium battery material type, altitude data, and battery capacity data is trained using machine learning algorithms. The machine learning algorithm is either a gradient boosting tree or a support vector machine. The training includes fitting the nonlinear relationship between material type, altitude, and battery capacity through an iterative optimization algorithm, and using auxiliary data as input feature variables to optimize the fitting process. The ternary correlation model is used to analyze the capacity decay rate corresponding to the unit of altitude change, and to calculate the weight of the impact of altitude change on battery capacity; wherein, the unit of altitude change is preset to a fixed value; Based on the aforementioned influence weights, the altitude sensitivity thresholds for lithium batteries of different material types are determined, and the capacity decay pattern within the current altitude range is output.

4. The method for early warning of lithium battery capacity degradation based on material properties according to claim 1, characterized in that, Based on the capacity decay law and real-time altitude change trend, the steps for predicting the degree and type of battery capacity decay and generating a warning signal when the capacity prediction result meets the preset warning conditions include: The capacity decay pattern and real-time altitude change trend are obtained; wherein, the real-time altitude change trend is obtained by performing time series analysis on real-time altitude data, including the trend types of continuous increase, maintenance at a high level, or decrease in altitude; Based on the real-time altitude change trend and combined with the capacity decay law, the degree and type of battery capacity decay are predicted, and a capacity prediction result is generated; wherein, the degree of battery decay includes the future battery capacity value, and the decay type includes reversible decay and irreversible decay. Determine whether the capacity prediction result meets the preset warning conditions; if so, generate a warning signal; wherein, the preset warning conditions include the future battery capacity value being lower than the capacity safety threshold or the real-time capacity decay rate exceeding the decay rate threshold.

5. The method for early warning of lithium battery capacity degradation based on material properties according to claim 4, characterized in that, The steps of classifying warning levels based on the warning signals, generating graded warning information, and pushing it to terminal devices include: Obtain early warning signals and associated battery material types; Based on the warning signal and battery material type, a preset classification rule is used to determine the warning level; Based on the warning level, a graded warning information is generated, which includes altitude status, capacity prediction results and response instructions. If the attenuation type is reversible attenuation, a response instruction to reduce equipment power consumption or adjust altitude is generated. If the attenuation type is irreversible attenuation, a response instruction to suspend operation or replace the battery is generated. The graded early warning information is pushed to the corresponding terminal device according to the communication protocol type of the terminal device.

6. The method for early warning of lithium battery capacity degradation based on material properties according to claim 5, characterized in that, The steps of obtaining user response feedback data after receiving the graded early warning information, updating the historical altitude capacity attenuation baseline data, and optimizing the parameters of the ternary correlation model include: After obtaining user response to graded early warning information, the feedback data includes altitude change information, battery capacity recovery status, and capacity decay type verification results; Based on the lithium battery material type in the disposal feedback data, match the corresponding entry in the historical altitude capacity decay benchmark data; Based on the altitude change information and battery capacity recovery status in the disposal feedback data, the historical altitude capacity decay benchmark data is updated based on the capacity decay type verification results. Based on the updated historical altitude capacity decay baseline data, the optimization operation of the ternary correlation model parameters is triggered.

7. A method for early warning of lithium battery capacity degradation based on material properties according to any one of claims 1 to 6, characterized in that, The method further includes: Obtain the charge / discharge cycle count and individual identifier of the lithium battery to generate an aging parameter dataset; Based on the individual identifiers and lithium battery material types in the aging parameter dataset, the corresponding aging sensitivity parameter values ​​are retrieved from the material property database. Based on the charge-discharge cycle number and aging sensitivity parameter value, the aging compensation factor is calculated and generated by calling the preset attenuation rate calculation formula. The aging compensation factor is applied to the output of the ternary correlation model to generate an adjusted capacity decay law. Based on the adjusted capacity decay law, the deviation rate between the predicted capacity decay value and the actual measured capacity value is calculated. If the deviation rate exceeds the preset tolerance threshold, then the warning trigger threshold in the preset warning conditions is updated.

8. A lithium battery altitude-based capacity decay early warning system based on material properties, characterized in that, The system includes: The parameter configuration module is used to configure the lithium battery material type and preset warning conditions, and to obtain historical altitude capacity decay benchmark data that matches the lithium battery material type from the material property database. The initialization module is used to initialize the communication protocol and acquisition frequency of the data acquisition device. The data acquisition module is used to synchronously acquire real-time altitude data, battery capacity data, and auxiliary data through the data acquisition device, and add a unified timestamp to generate the raw dataset; The data processing and feature extraction module is used to preprocess the original dataset and extract features based on the historical altitude capacity decay benchmark data to generate a dataset that integrates benchmark features. The data standardization and transformation module is used to perform standardization transformation on the merged dataset to generate a standardized analysis dataset. The model building and analysis module is used to build a ternary correlation model of lithium battery material type, altitude and battery capacity based on the standardized analysis dataset and through machine learning algorithms, quantify the influence weight of altitude change on battery capacity, and output the capacity decay law of the current altitude range. The capacity degradation trend prediction module predicts the degree and type of battery capacity degradation based on the capacity degradation law and the real-time altitude change trend. The early warning signal generation module is used to generate an early warning signal when the capacity prediction result meets the preset early warning conditions. The graded early warning module is used to classify early warning levels based on the early warning signal, generate graded early warning information, and push it to the terminal device; wherein, the graded early warning information includes altitude status, capacity prediction results, and response instructions; The closed-loop feedback optimization module is used to obtain the handling feedback data after the user responds to the graded early warning information, update the historical altitude capacity attenuation benchmark data, and optimize the parameters of the ternary correlation model.

9. A computer device, characterized in that: The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer program is stored that can be loaded by a processor and executed as described in any one of claims 1 to 7.

Images