Unmanned aerial vehicle lithium battery remaining service life monitoring method based on lithium battery multi-mode system

Abstract

The invention provides an unmanned aerial vehicle lithium battery remaining service life monitoring method based on a lithium battery multi-mode system. The method comprises the following steps: acquiring a multi-mode battery state parameter of a lithium battery of the unmanned aerial vehicle; performing health equivalent cycle evaluation processing on the state parameters of the multi-mode battery to obtain the number of cycles of the battery; and outputting the remaining service life of the lithium battery of the unmanned aerial vehicle according to the battery cycle index. The method comprises the following steps: after collecting state parameters of a multi-mode battery, determining various current running states of the lithium battery of the unmanned aerial vehicle, and then evaluating the state parameters of the multi-mode battery so as to convert data of the various running states of the lithium battery into evaluation data of corresponding battery cycle times; and finally, the remaining service life of the unmanned aerial vehicle lithium battery is determined through the evaluation data of the battery cycle index, so that prediction of the remaining service life of the unmanned aerial vehicle lithium battery is more accurate, and the prediction reliability of the remaining service life of the unmanned aerial vehicle lithium battery is effectively improved.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of unmanned aerial vehicle lithium battery, and particularly relates to a method for monitoring the remaining useful life of an unmanned aerial vehicle lithium battery based on a lithium battery multi-modal system. BACKGROUND

[0002] With the deep integration and wide application of unmanned aerial vehicle technology in the fields of logistics, agriculture, power inspection, surveying and mapping, etc., the performance and reliability of lithium batteries as the core power source have become the key elements restricting the operation efficiency, flight safety and operation cost of unmanned aerial vehicles. As the power core of unmanned aerial vehicles, the performance of lithium batteries directly determines the endurance, load capacity and flight safety of unmanned aerial vehicles. At present, the management of unmanned aerial vehicle lithium batteries is generally challenged. The common battery performance detection method is mainly focused on the evaluation of the state of health (SOH) of lithium batteries inside the battery management system (BMS), which estimates the remaining capacity (RC) of the battery by monitoring the basic parameters such as voltage, current and temperature of the battery, and combining the coulomb counting method or open circuit voltage method, and then calculates the SOH.

[0003] However, the state of health (SOH) evaluation in the above-mentioned traditional method is not accurate enough, and the remaining useful life (RUL) prediction lacks reliability, resulting in low utilization rate of battery assets and frequent unplanned failures, which seriously affects the overall performance and economic benefits of the unmanned aerial vehicle fleet. Especially in high-intensity and high-frequency operation scenarios, the rapid degradation and performance differentiation of the battery are increasingly prominent. SUMMARY

[0004] The purpose of the present disclosure is to overcome the deficiencies in the prior art and provide a method for monitoring the remaining useful life of an unmanned aerial vehicle lithium battery based on a lithium battery multi-modal system, which effectively improves the prediction reliability of the remaining useful life.

[0005] The purpose of the present disclosure is achieved by the following technical solutions:

[0006] A method for monitoring the remaining useful life of an unmanned aerial vehicle lithium battery based on a lithium battery multi-modal system, comprising: collecting and analyzing the remaining useful life of the unmanned aerial vehicle lithium battery by using a lithium battery multi-modal system, wherein the lithium battery multi-modal system comprises a battery data acquisition module, a data preprocessing module, a battery health evaluation module and a battery remaining useful life output module; the battery data acquisition module is used to acquire multi-modal battery data of the unmanned aerial vehicle lithium battery; the data preprocessing module is used to perform preprocessing operation on the multi-modal battery data; the battery health evaluation module is used to evaluate the data after the preprocessing operation; and the battery remaining useful life output module is used to output the remaining useful life according to the evaluation result.

[0007] The unmanned aerial vehicle lithium battery remaining service life monitoring method comprises:

[0008] Obtaining multi-modal battery state parameters of the unmanned aerial vehicle lithium battery;

[0009] Performing health equivalent cycle evaluation processing on the multi-modal battery state parameters to obtain a battery cycle number;

[0010] Outputting a remaining service life of the unmanned aerial vehicle lithium battery according to the battery cycle number.

[0011] In one embodiment, obtaining multi-modal battery state parameters of the unmanned aerial vehicle lithium battery comprises: obtaining electrical parameters and operating parameters of the unmanned aerial vehicle lithium battery.

[0012] In one embodiment, performing health equivalent cycle evaluation processing on the multi-modal battery state parameters to obtain a battery cycle number comprises: performing feature extraction preprocessing on the multi-modal battery state parameters to obtain a plurality of battery state feature values.

[0013] In one embodiment, performing feature extraction preprocessing on the multi-modal battery state parameters to obtain a battery state feature value further comprises: performing normalization and quantization processing on the plurality of battery state feature values respectively to obtain a plurality of battery evaluation quantization values.

[0014] In one embodiment, performing normalization preprocessing on the multi-modal battery state parameters to obtain a plurality of battery evaluation quantization values further comprises: performing weighted summation processing on the battery evaluation quantization values to obtain a battery health evaluation value.

[0015] In one embodiment, performing weighted summation processing on the battery evaluation quantization values to obtain a battery health evaluation value further comprises: performing evaluation correction processing on the battery health evaluation value to obtain a battery health correction value.

[0016] In one embodiment, performing evaluation correction processing on the battery health evaluation value to obtain a battery health correction value further comprises: performing cycle fitting processing on the battery health correction value to obtain a battery cycle number.

[0017] In one embodiment, performing cycle fitting processing on the battery health correction value comprises: performing a double exponential health cycle number decay curve fitting operation on the battery health correction value.

[0018] In one embodiment, performing cycle fitting processing on the battery health correction value comprises: performing a polynomial health cycle number decay curve fitting operation on the battery health correction value.

[0019] In one of the embodiments, the method for outputting the remaining service life of the unmanned aerial vehicle lithium battery according to the battery cycle number comprises: calculating the difference between the battery cycle number and the total cycle number to obtain the remaining service life of the unmanned aerial vehicle lithium battery.

[0020] Compared with the prior art, the present disclosure has at least the following advantages:

[0021] After collecting the multi-modal battery state parameters, the current lithium battery operating state of the unmanned aerial vehicle lithium battery is determined, and then the multi-modal battery state parameters are evaluated to convert the lithium battery operating state data into corresponding battery cycle number evaluation data. Finally, the remaining service life of the unmanned aerial vehicle lithium battery is determined through the battery cycle number evaluation data, so that the prediction of the remaining service life of the unmanned aerial vehicle lithium battery is more accurate, and the prediction reliability of the remaining service life of the unmanned aerial vehicle lithium battery is effectively improved. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present disclosure, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0023] Figure 1 The flow chart of the method for monitoring the remaining service life of the unmanned aerial vehicle lithium battery based on the multi-modal system of the lithium battery in one embodiment;

[0024] Figure 2 The system architecture diagram of the multi-modal system of the lithium battery in one embodiment. DETAILED DESCRIPTION

[0025] In order to facilitate the understanding of the present disclosure, the following will make a more comprehensive description of the present disclosure with reference to the related drawings. The preferred embodiments of the present disclosure are shown in the drawings. However, the present disclosure can be realized in many different forms, and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present disclosure more thorough and comprehensive.

[0026] It should be noted that when an element is referred to as being "fixed" to another element, it can be directly on the other element or there can be an intervening element. When an element is referred to as being "connected" to another element, it can be directly connected to the other element or there can be an intervening element. The terms "vertical", "horizontal", "left", "right", and similar expressions used herein are for illustrative purposes only and are not intended to be the only implementation.

[0027] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of this disclosure. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0028] This disclosure relates to a method for monitoring the remaining lifespan of a drone lithium battery based on a multimodal lithium battery system. In one embodiment, the method includes acquiring multimodal battery state parameters of the drone lithium battery; performing a health equivalent cycle assessment on the multimodal battery state parameters to obtain the battery cycle count; and outputting the remaining lifespan of the drone lithium battery based on the battery cycle count. After acquiring the multimodal battery state parameters, the current multiple operating states of the drone lithium battery are determined. Then, the multimodal battery state parameters are evaluated to convert the multiple operating state data into corresponding battery cycle count evaluation data. Finally, the remaining lifespan of the drone lithium battery is determined based on the battery cycle count evaluation data, making the prediction of the remaining lifespan of the drone lithium battery more accurate and effectively improving the reliability of the prediction.

[0029] Please see Figure 1 This is a flowchart illustrating a method for monitoring the remaining lifespan of a drone lithium battery based on a lithium battery multimodal system, according to an embodiment of this disclosure. The method includes some or all of the following steps. Specifically, a lithium battery multimodal system is used to collect and analyze the remaining lifespan of the drone lithium battery. The system architecture diagram corresponding to the lithium battery multimodal system is shown below. Figure 2 As shown, the lithium battery multimodal system includes a battery data acquisition module, a data preprocessing module, a battery health assessment module, and a battery remaining life output module. The battery data acquisition module is used to acquire multimodal battery data of the UAV's lithium battery. The data preprocessing module is used to preprocess the multimodal battery data. The battery health assessment module is used to assess the preprocessed data. The battery remaining life output module is used to output the remaining life based on the assessment results.

[0030] The method for monitoring the remaining lifespan of a drone's lithium battery specifically includes the following steps:

[0031] S100: Acquire multimodal battery state parameters of the drone's lithium battery.

[0032] In this embodiment, the multimodal battery state parameters refer to various battery state data of the drone's lithium battery. Specifically, these parameters represent multiple operating condition indicators of the drone's lithium battery, corresponding to various performance characteristics of the drone's lithium battery. Collecting these multimodal battery state parameters facilitates sampling of various operating states of the drone's lithium battery, enabling comprehensive performance data collection and improving the accuracy of predicting the remaining lifespan of the drone's lithium battery.

[0033] S200: Perform a health equivalent cycle assessment on the state parameters of the multimodal battery to obtain the number of battery cycles.

[0034] In this embodiment, the multimodal battery state parameters are various battery state data of the drone's lithium battery, that is, the multimodal battery state parameters are various operating condition indicators of the drone's lithium battery, and thus correspond to multiple operating performance characteristics of the drone's lithium battery. By collecting these multimodal battery state parameters, it is convenient to sample various operating states of the drone's lithium battery, thereby facilitating comprehensive performance data collection and improving the accuracy of predicting the remaining lifespan of the drone's lithium battery. The health equivalent cycle assessment process converts the multimodal battery state parameters into corresponding assessment data. For example, the multimodal battery state parameters are first converted into battery health assessment data, and then the battery health assessment data is converted into battery cycle counts, so as to determine the operating performance of the drone's lithium battery from multiple perspectives and modes.

[0035] S300: Outputs the remaining lifespan of the drone's lithium battery based on the number of battery cycles.

[0036] In this embodiment, the battery cycle count is obtained based on the multimodal battery state parameters, which are various battery state data of the drone lithium battery. In other words, the multimodal battery state parameters represent various operating condition indicators of the drone lithium battery, and correspond to multiple performance characteristics of the drone lithium battery. By collecting these multimodal battery state parameters, it is convenient to sample various operating states of the drone lithium battery, thereby facilitating comprehensive performance data collection and improving the accuracy of predicting the remaining lifespan of the drone lithium battery. The health equivalent cycle assessment process converts the multimodal battery state parameters into corresponding assessment data. For example, the multimodal battery state parameters are first converted into battery health assessment data, and then the battery health assessment data is converted into battery cycle counts, so as to determine the operating performance of the drone lithium battery from multiple perspectives and modes. After obtaining the battery cycle count, the number of cycles currently used by the drone lithium battery can be determined, which facilitates the calculation of the remaining cycle count of the drone lithium battery. This makes it easier to accurately calculate the remaining lifespan of the drone lithium battery and effectively improves the reliability of predicting the remaining lifespan of the drone lithium battery.

[0037] In the above embodiments, after collecting the multi-mode battery state parameters, the current multiple operating states of the drone lithium battery are determined. Then, the multi-mode battery state parameters are evaluated to convert the multiple operating state data of the lithium battery into corresponding battery cycle count evaluation data. Finally, the remaining service life of the drone lithium battery is determined by the battery cycle count evaluation data, making the prediction of the remaining service life of the drone lithium battery more accurate and effectively improving the reliability of the prediction of the remaining service life of the drone lithium battery.

[0038] In one embodiment, acquiring multimodal battery state parameters of a drone lithium battery includes acquiring the electrical parameters and operating parameters of the drone lithium battery. In this embodiment, the multimodal battery state parameters are various battery state data of the drone lithium battery, that is, the multimodal battery state parameters are various operating condition indicators of the drone lithium battery, and the multimodal battery state parameters correspond to multiple operating performance characteristics of the drone lithium battery. By collecting the multimodal battery state parameters, it is convenient to sample various operating states of the drone lithium battery, thereby facilitating the comprehensive performance data collection of the drone lithium battery and improving the accuracy of predicting the remaining service life of the drone lithium battery. The multimodal battery state parameters include the electrical parameters and operating parameters of the drone lithium battery. The electrical parameters of the drone lithium battery include voltage, current, temperature, internal resistance, capacity decay rate, internal resistance growth rate, voltage plateau change rate, and self-discharge rate. The operating parameters include historical cycle data, charge-discharge curves, usage environment data, and abnormal historical data of the drone lithium battery. Collecting multimodal data on drone batteries is appropriate for the actual situation of battery aging, because battery aging is a complex process involving multiple factors and stages, and a single indicator cannot fully reflect its health status.

[0039] In one embodiment, a health equivalent cycle assessment is performed on the multimodal battery state parameters to obtain the battery cycle count. This includes preprocessing the multimodal battery state parameters by feature extraction to obtain multiple battery state feature values. In this embodiment, the multimodal battery state parameters are various battery state data of the drone lithium battery, that is, the multimodal battery state parameters are various operating condition indicators of the drone lithium battery, and thus correspond to multiple operating performance characteristics of the drone lithium battery. By collecting the multimodal battery state parameters, it is convenient to sample various operating states of the drone lithium battery, thereby facilitating the comprehensive performance data collection of the drone lithium battery and improving the accuracy of predicting the remaining service life of the drone lithium battery. The health equivalent cycle assessment process converts the multimodal battery state parameters into corresponding assessment data. For example, the multimodal battery state parameters are first converted into battery health assessment data, and then the battery health assessment data is converted into the battery cycle count, so as to determine the operating performance of the drone lithium battery from multiple perspectives and multiple modes. The multimodal battery state parameters include the electrical parameters and operational parameters of the drone lithium battery. The electrical parameters include voltage, current, temperature, internal resistance, capacity decay rate, internal resistance growth rate, voltage plateau change rate, and self-discharge rate. The operational parameters include historical cycle data, charge / discharge curves, usage environment data, and historical anomaly data of the drone lithium battery. Collecting multimodal data from the drone battery is tailored to the actual situation of battery aging, as battery aging is a complex process involving multiple factors and stages, and a single indicator cannot fully reflect its health status.

[0040] The health equivalent cycle assessment process includes feature extraction preprocessing of multimodal battery state parameters. This feature extraction preprocessing involves preprocessing the raw collected data, such as the multimodal battery state parameters, and extracting the aforementioned key features from various dimensions. Specifically, the feature extraction preprocessing includes filtering, denoising, and outlier removal of the multimodal battery state parameters to improve the accuracy of the collected data corresponding to the battery state feature values, thereby enhancing the accuracy of predicting the remaining lifespan of the drone's lithium battery.

[0041] The collected voltage, current, and temperature data are inevitably affected by sensor noise and electromagnetic interference during transmission and measurement. To obtain smooth and reliable data for subsequent analysis, filtering is necessary. A window-shifting average filtering method is employed. For example, if the entire window data queue contains 5 existing data points, and the limit is 5, each time a new data point is added to the queue, the oldest data point is removed. The maximum and minimum values ​​are removed from the queue, and then the average is calculated. This final average is considered the most recent and valid data.

[0042] Each physical quantity has a normal range. If the detected data is outside the range, it is considered an outlier and will be discarded. For example, the normal range for a single cell voltage is 0 to 4.5V. If a single cell voltage of 5.8V is detected, it will be considered an outlier and will be discarded.

[0043] Furthermore, feature extraction preprocessing is performed on the multimodal battery state parameters to obtain battery state feature values. This is followed by normalization and quantization processing of multiple battery state feature values ​​to obtain multiple battery evaluation quantization values. In this embodiment, the multimodal battery state parameters are various battery state data of the drone's lithium battery, i.e., the multimodal battery state parameters are various operating condition indicators of the drone's lithium battery, and thus correspond to multiple operating performance characteristics of the drone's lithium battery. By collecting the multimodal battery state parameters, it is convenient to sample various operating states of the drone's lithium battery, thereby facilitating comprehensive performance data collection and improving the accuracy of predicting the remaining lifespan of the drone's lithium battery. The health equivalent cycle evaluation process converts the multimodal battery state parameters into corresponding evaluation data. For example, the multimodal battery state parameters are first converted into battery health evaluation data, and then the battery health evaluation data is converted into battery cycle counts, so as to determine the operating performance of the drone's lithium battery from multiple perspectives and modes. The multimodal battery state parameters include the electrical parameters and operational parameters of the drone lithium battery. The electrical parameters include voltage, current, temperature, internal resistance, capacity decay rate, internal resistance growth rate, voltage plateau change rate, and self-discharge rate. The operational parameters include historical cycle data, charge / discharge curves, usage environment data, and historical anomaly data of the drone lithium battery. Collecting multimodal data from the drone battery is tailored to the actual situation of battery aging, as battery aging is a complex process involving multiple factors and stages, and a single indicator cannot fully reflect its health status.

[0044] The health equivalent cycle assessment process includes feature extraction preprocessing of multimodal battery state parameters. This feature extraction preprocessing involves preprocessing the raw collected data, such as the multimodal battery state parameters, and extracting the aforementioned key features from various dimensions. Specifically, the feature extraction preprocessing includes filtering, denoising, and outlier removal of the multimodal battery state parameters to improve the accuracy of the collected data corresponding to the battery state feature values, thereby enhancing the accuracy of predicting the remaining lifespan of the drone's lithium battery.

[0045] Following the feature extraction preprocessing, a normalization quantization process is performed. This normalizes each battery state feature value, converting it into a quantized score between 0 and 100, reflecting its contribution to battery health. The quantization function can be implemented using linear mapping, piecewise functions, or lookup tables. For example, capacity retention rate is directly used as the score; internal resistance growth rate is converted into a score using its reciprocal or decay function, with higher internal resistance resulting in a lower score; and abnormal historical records are quantified using a penalty function based on their cumulative frequency and severity, with more and more severe abnormal events resulting in a lower score.

[0046] Furthermore, the multimodal battery state parameters are preprocessed to normalize and obtain multiple battery evaluation quantification values. This is followed by a weighted summation of these quantification values ​​to obtain a battery health assessment value. In this embodiment, the multimodal battery state parameters are various battery state data of the drone's lithium battery, meaning they represent multiple operating condition indicators of the drone's lithium battery, and thus correspond to multiple performance characteristics of the drone's lithium battery. Collecting these multimodal battery state parameters facilitates sampling of various operating states of the drone's lithium battery, enabling comprehensive performance data collection and improving the accuracy of predicting the remaining lifespan of the drone's lithium battery. The health equivalent cycle assessment process converts the multimodal battery state parameters into corresponding assessment data. For example, the multimodal battery state parameters are first converted into battery health assessment data, and then the battery health assessment data is converted into battery cycle counts, allowing for multi-dimensional and multimodal determination of the drone's lithium battery's performance. The multimodal battery state parameters include the electrical parameters and operational parameters of the drone lithium battery. The electrical parameters include voltage, current, temperature, internal resistance, capacity decay rate, internal resistance growth rate, voltage plateau change rate, and self-discharge rate. The operational parameters include historical cycle data, charge / discharge curves, usage environment data, and historical anomaly data of the drone lithium battery. Collecting multimodal data from the drone battery is tailored to the actual situation of battery aging, as battery aging is a complex process involving multiple factors and stages, and a single indicator cannot fully reflect its health status.

[0047] Following the normalization quantization process, a weighted summation process is performed, summing the quantified battery assessment values ​​to obtain a preliminary SOH (State of Health) assessment value, i.e., the battery health assessment value. The weighting coefficients are preset based on the sensitivity and importance of each indicator to the battery's health status, and are statistically analyzed and optimized using a large amount of actual battery operating data. These weights are not fixed but can be dynamically adjusted according to battery type, usage scenario, and even the battery's aging stage.

[0048] In another embodiment, the weighted summation process satisfies the following formula: SOH 初步 =Σω i *T i , where ω i For the corresponding indicator weights, and Σω i =1,T i These are the quantitative values ​​for various battery evaluations. Specifically, SOH... 初步 =ω1*CRR score +ω2*IRIR score +ω3*VPD score +ω4*SDRC score +ω5*CDEC score +ω6*Thermal sc ore +ω7*AHR score CRR score The capacity retention rate, with a value of 0.4, directly reflects the available energy; IRIR score The internal resistance growth rate, with a value of 0.2, reflects power performance and heat generation, and has a significant impact on UAV flight; VPD score The voltage plateau decay rate, with a value of 0.1, reflects the internal chemical changes of the battery; SDRC score The self-discharge rate, with a value of 0.05, reflects the battery's internal self-discharge. CDEC score The charge / discharge efficiency is represented by a value of 0.05, reflecting the energy conversion efficiency; Thermal score The value is 0.1, representing temperature and reflecting thermal management and safety risks; AHR score This represents an abnormal historical record, with a value of 0.1. It has a significant impact on safety and may cause irreversible damage. The above weights will be dynamically adjusted according to the subsequent adaptive optimization process to adapt to different individual batteries and usage environments.

[0049] In another embodiment, the weighted summation of each battery evaluation quantification value is performed to obtain a battery health evaluation value. This is followed by an evaluation correction process to obtain a corrected battery health value. In this embodiment, the multimodal battery state parameters are various battery state data of the drone's lithium battery, meaning they represent various operating condition indicators of the drone's lithium battery, and thus correspond to multiple performance characteristics of the drone's lithium battery. Collecting these multimodal battery state parameters facilitates sampling of various operating states of the drone's lithium battery, enabling comprehensive performance data collection and improving the accuracy of predicting the remaining lifespan of the drone's lithium battery. The health equivalent cycle evaluation process converts the multimodal battery state parameters into corresponding evaluation data. For example, the multimodal battery state parameters are first converted into battery health evaluation data, and then the battery health evaluation data is converted into battery cycle counts, allowing for multi-dimensional and multimodal determination of the drone's lithium battery's performance. The multimodal battery state parameters include the electrical parameters and operational parameters of the drone lithium battery. The electrical parameters include voltage, current, temperature, internal resistance, capacity decay rate, internal resistance growth rate, voltage plateau change rate, and self-discharge rate. The operational parameters include historical cycle data, charge / discharge curves, usage environment data, and historical anomaly data of the drone lithium battery. Collecting multimodal data from the drone battery is tailored to the actual situation of battery aging, as battery aging is a complex process involving multiple factors and stages, and a single indicator cannot fully reflect its health status.

[0050] Following the weighted summation process, an evaluation and correction process is performed. This is because the initial SOH assessment value may have some deviation, and an empirical correction factor and adaptive optimization mechanism are introduced to correct the initial SOH. The correction factor can be calculated by looking up tables or empirical formulas based on the battery's cycle count, cumulative discharge capacity, and performance under specific operating conditions. Simultaneously, the system periodically compares the SOH estimated by the BMS with the actual battery performance, which is obtained through periodic capacity calibration tests. An error feedback mechanism is used to adaptively adjust the weights or correction factors, thereby continuously optimizing the accuracy of the SOH assessment.

[0051] Specifically, the following methods are included:

[0052] (1) Establishing a correction factor lookup table: By analyzing a large amount of battery lifecycle data, a multidimensional lookup table is established. The inputs to this table are the battery cycle count, cumulative discharge capacity, average operating temperature, etc., and the output is a correction factor f. 修正For example, when a battery operates for a long time in a high-temperature environment, its SOH (State of Health) will decrease more rapidly. At this time, the correction factor obtained from the table will be less than 1, thus reducing the initial SOH.

[0053] (2) Adaptive weight optimization: The system periodically (e.g., every 50 cycles) compares the SOH estimated by the model with the actual SOH obtained through standard capacity testing and calculates the error. Based on the magnitude and direction of the error, optimization algorithms such as gradient descent or least squares are used to adjust the weight coefficients ω in the weighted fusion model. i Fine-tuning can be performed. For example, if the internal resistance is found to have a greater impact on SOH than the preset value, the system will automatically increase the weight of ω2.

[0054] Battery Health Correction Value (SOH) 最终 Satisfy the following formula:

[0055] SOH 最终 =SOH 初步 *f 修正

[0056] Among them, f 修正 This represents an empirical correction function or a lookup table result.

[0057] In another embodiment, gradient descent is chosen as the adaptive optimization weight algorithm. Gradient descent is an iterative optimization algorithm that finds the parameter combination that minimizes the objective function by gradually adjusting the parameters (i.e., weights) along the opposite direction of the gradient of the objective function (i.e., the error function).

[0058] The loss function L is defined as the mean square error between the predicted SOH value and the "true" SOH, which makes larger errors more significantly corrected, and its mathematical properties are easy to differentiate.

[0059]

[0060] Where: W=[ω1,ω2,...,ω n [ ] is the weight vector of each evaluation indicator; SOH 真实 The reference SOH value is obtained through offline capacity testing; this is the "true value" learned by the model. 预测 The current model is based on the quantified scores of various indicators X = [X1, X2, ..., X...]. n The SOH value calculated using the current weight W is the battery health assessment value.

[0061]

[0062] The core of gradient descent lies in updating the weights based on the partial derivative of the loss function with respect to each weight (i.e., the gradient), with each update moving in the direction that reduces the error.

[0063] First, calculate the loss function L for each weight ω. k The partial derivative function,

[0064]

[0065] Then, update the weights based on the gradient:

[0066]

[0067] in: It is the weight of the k-th indicator after the update; This represents the weight of the current k-th metric; α is the learning rate, a small positive number that controls the step size for each weight update; (SOH) 真实 -SOH 预测 ) represents the error in the current SOH assessment; X k It is the quantitative score of the kth indicator.

[0068] In another embodiment, the battery health assessment value is evaluated and corrected to obtain a corrected battery health value. This is followed by a cyclic fitting process on the corrected battery health value to obtain the battery cycle count. In this embodiment, the multimodal battery state parameters are various battery state data of the drone lithium battery, i.e., the multimodal battery state parameters are various operating condition indicators of the drone lithium battery, and thus correspond to multiple operating performance characteristics of the drone lithium battery. By collecting the multimodal battery state parameters, it is convenient to sample various operating states of the drone lithium battery, thereby facilitating comprehensive performance data collection and improving the accuracy of predicting the remaining lifespan of the drone lithium battery. The health equivalent cycle evaluation process converts the multimodal battery state parameters into corresponding evaluation data. For example, the multimodal battery state parameters are first converted into battery health assessment data, and then the battery health assessment data is converted into the battery cycle count, so as to determine the operating performance of the drone lithium battery from multiple perspectives and modes. The multimodal battery state parameters include the electrical parameters and operational parameters of the drone lithium battery. The electrical parameters include voltage, current, temperature, internal resistance, capacity decay rate, internal resistance growth rate, voltage plateau change rate, and self-discharge rate. The operational parameters include historical cycle data, charge / discharge curves, usage environment data, and historical anomaly data of the drone lithium battery. Collecting multimodal data from the drone battery is tailored to the actual situation of battery aging, as battery aging is a complex process involving multiple factors and stages, and a single indicator cannot fully reflect its health status.

[0069] After evaluation and correction, a cyclic fitting process is performed. The degradation of drone lithium batteries is a gradual process, and there is a certain regularity between the decrease in SOH and the number of cycles or cumulative discharge capacity. By analyzing a large amount of historical degradation data of similar batteries, an empirical curve or mathematical model between SOH and RUL is established. Combined with the current real-time SOH of the battery, the remaining number of cycles or time required to reach the end-of-life (EOL) threshold (e.g., SOH below 80%) is predicted.

[0070] We collected extensive lifecycle data of lithium batteries for similar drones, from new batteries to end-of-life, and plotted the SOH (State of Health) decay curves as a function of cycle number. We then mathematically fitted these curves using polynomial, exponential, or logarithmic functions to establish an empirical model between SOH and cycle number. A decay model of the following form can be fitted:

[0071] (1) Double exponential model: Considering that the rate of battery degradation is different in the early and late stages, the double exponential model can better fit the SOH degradation curve, that is, the double exponential healthy cycle number degradation curve fitting operation, which satisfies the following formula:

[0072] SOH (N) =A*e -aN +B*e -bN +C

[0073] Where N is the number of iterations, and A, a, B, b, and C are the model parameters obtained by fitting using the least squares method.

[0074] (2) Polynomial Model: For certain types of batteries, higher-order polynomial models can also achieve better fitting results, namely, polynomial healthy cycle decay curve fitting operation, which satisfies the following formula:

[0075] SOH (N) =C0+C1N+C2N 2 +C3N 3 +...+C k N k

[0076] Where N is the number of iterations, C i These are the fitting parameters.

[0077] In one embodiment, the remaining lifespan of the drone's lithium battery is output based on the battery cycle count, including: calculating the difference between the battery cycle count and the total cycle count to obtain the remaining lifespan of the drone's lithium battery. In this embodiment, the battery cycle count is obtained based on the multimodal battery state parameters, which are various battery state data of the drone's lithium battery, i.e., various operating condition indicators of the drone's lithium battery, and corresponding to multiple operating performance characteristics of the drone's lithium battery. By collecting the multimodal battery state parameters, it is convenient to sample various operating states of the drone's lithium battery, thereby facilitating the comprehensive collection of performance data of the drone's lithium battery and improving the accuracy of predicting the remaining lifespan of the drone's lithium battery. The health equivalent cycle assessment process converts the multimodal battery state parameters into corresponding assessment data. For example, the multimodal battery state parameters are first converted into battery health assessment data, and then the battery health assessment data is converted into battery cycle count, so as to determine the operating performance of the drone's lithium battery from multiple perspectives and modes. After obtaining the battery cycle count, the current number of cycles used by the drone's lithium battery can be determined, facilitating the calculation of the remaining cycle count and thus enabling an accurate calculation of the remaining lifespan of the drone's lithium battery. This effectively improves the reliability of predicting the remaining lifespan of the drone's lithium battery. The remaining cycle count of the drone's lithium battery is determined by calculating the difference between the battery cycle count and the total cycle count, making it easier to calculate the remaining lifespan accordingly.

[0078] In another embodiment, once the SOH decay model is established, the equivalent number of cycles N that the battery has completed can be deduced from the current real-time SOH value of the battery. current. Then, based on the lifespan termination threshold SOH EOL Calculate the total number of cycles N required to reach the end of life. total The remaining lifetime (RUL) then satisfies the following formula:

[0079] RUL=N total -N current

[0080] In another embodiment, the battery degradation rate is influenced by various factors, such as charge / discharge rate, operating temperature, and depth of discharge. A correction factor is introduced to adjust the predicted Remaining Life (RUL). These correction factors can be obtained through statistical analysis (such as multiple regression) or table lookup methods, reflecting the impact of different operating conditions on battery life. This correction function is constructed based on the Arrhenius and Peukert equations and optimized using empirical data, resulting in a corrected RUL. 修正Satisfy the following formula:

[0081] RUL 修正 =RUL*f 环境 (T avg DOD avg C rate,avg )

[0082] Where: f 环境 As an environmental correction factor, it comprehensively considers the average operating temperature (T). avg ), mean depth of discharge (DOD) avg ), average charge / discharge rate (C) rate,avg The influence of factors such as temperature, depth of discharge, and charge / discharge rate on RUL (Rear Usage Limit) can be considered. This correction factor can be obtained through a multidimensional lookup table or by fitting a multivariate function. Temperature correction: Based on the Arrhenius equation, higher temperatures lead to faster degradation. Depth of discharge correction: Greater depth of discharge results in faster battery degradation. Charge / discharge rate correction: Higher charge / discharge rates lead to faster battery degradation.

[0083] In another embodiment, the remaining lifespan of the drone's lithium battery is output based on the number of battery cycles, followed by grading the drone's lithium battery according to the remaining lifespan. In this embodiment, the drone's lithium battery is divided into multiple health levels, for example, five levels: A, B, C, D, and E. Each level corresponds to different SOH, RUL, power performance, and internal resistance ranges, and is recommended for different application scenarios. This grading standard is dynamically adjustable and can be customized according to actual application needs and battery type. Power performance and internal resistance indicators can be evaluated in real-time or near real-time using BMS data, as detailed below.

[0084]

[0085]

[0086] In another embodiment, the intelligent task allocation system intelligently matches the most suitable battery for the task based on the specific requirements of the UAV mission (such as required flight time, payload, flight distance, environmental conditions, etc.) and the real-time health level of the battery, using an optimization algorithm. Its core principles include:

[0087] (1) Safety assurance: Prioritize ensuring that the selected battery can complete the task safely and reliably, and avoid safety hazards caused by insufficient battery performance.

[0088] (2) Precise performance matching: Based on the strict requirements of the task for battery capacity, power, range and other performance, the battery grade with the highest performance matching degree is selected from the available battery pool.

[0089] (3) Maximize asset utilization: Under the premise of meeting the task requirements, prioritize scheduling batteries with lower health but still capable of performing the task, thereby extending the service life of high-level batteries and maximizing the overall lifespan and tiered utilization of battery assets.

[0090] (4) Operational cost optimization: Taking into account the residual value of the battery, maintenance costs and task benefits, the battery usage cost is minimized through intelligent scheduling.

[0091] In another embodiment, the cloud-based or local intelligent analysis and management platform is the core brain of the invention, responsible for the aggregation, storage, in-depth analysis, model training and optimization of massive battery data, and providing advanced functions such as battery digital twins, visual management, fault warning and decision support, in order to build an intelligent management ecosystem for the entire battery life cycle.

[0092] I. Core Functional Modules

[0093] (1) Data Lake and Big Data Storage: Aggregates all battery-related data from BMS, charging and discharging equipment, and historical maintenance records. Utilizing distributed storage technologies (such as Hadoop HDFS and NoSQL databases), it supports the storage and efficient querying of massive amounts of heterogeneous data. The platform allows users to choose between cloud or local storage based on deployment requirements.

[0094] (2) Data cleaning and preprocessing: The original data is cleaned, deduplicated, formatted, missing values ​​are filled and outliers are handled to ensure data quality and provide a reliable foundation for subsequent analysis.

[0095] (3) Battery Digital Twin Construction and Management: A real-time updated digital twin is established for each UAV lithium battery. This twin is a virtual, dynamic battery model that accurately maps the current state, historical behavior, and future trends of the physical battery. The digital twin includes basic battery information, real-time SOH / RUL, historical charge / discharge curves, abnormal event records, maintenance records, predictive analysis results, etc. Through the digital twin, users can intuitively and comprehensively understand the "life cycle" of a single battery and perform personalized management.

[0096] (4) In-depth analysis and model optimization:

[0097] 1) Attenuation pattern identification: Using statistical analysis methods such as cluster analysis and principal component analysis, the attenuation patterns and rules of different types of batteries are identified from massive amounts of data.

[0098] 2) Analysis of influencing factors: Through regression analysis, correlation analysis and other methods, the influence of environmental and usage factors such as temperature, humidity, charge and discharge rate, charging strategy and depth of discharge on battery degradation is quantified.

[0099] 3) SOH / RUL model optimization: Based on the continuous accumulation of real battery life cycle data, we continuously optimize the parameters and make empirical corrections to the multi-index fusion SOH evaluation model and RUL prediction model to ensure that the accuracy and generalization ability of the model are always at their best.

[0100] 4) Fault diagnosis and tracing: Combining historical data and real-time abnormal events, using methods such as fault tree analysis and cause-effect graphs, we conduct in-depth diagnosis and tracing of battery faults.

[0101] (5) Visualization and decision support

[0102] 1) Battery Health Dashboard: Provides an intuitive overview of battery health status, including SOH distribution, RUL prediction, and a list of abnormal batteries.

[0103] 2) Individual Battery Details Page: Displays digital twin data for each battery, including real-time parameters, historical curves, forecast reports, and maintenance recommendations.

[0104] 3) Fleet Management View: Provides a macro view of the entire drone fleet's battery assets, supporting filtering and management by level, status, location, etc.

[0105] 4) Intelligent early warning system: Based on the SOH / RUL prediction results and anomaly detection results, it issues early warning information (such as SMS, email, and App notification) in advance to remind users to perform maintenance or replacement.

[0106] 5) Task scheduling optimization suggestions: Based on battery health status and task requirements, provide intelligent battery matching suggestions for drone scheduling.

[0107] II. Platform Advantages

[0108] (1) Data-driven: Make full use of massive BMS data to achieve intelligent and refined battery management.

[0109] (2) Strong predictive power: The practical model provides accurate SOH and RUL predictions, supporting preventive maintenance.

[0110] (3) Visualization: Intuitive interface and digital twin improve user management efficiency.

[0111] (4) Scalability: Modular design, easy to integrate new features and support more battery types.

[0112] (5) Security: Strict data encryption and access control ensure data security and privacy.

[0113] The embodiments described above are merely illustrative of several implementations of this disclosure, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this disclosure, and these all fall within the protection scope of this disclosure. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A method for monitoring the remaining service life of a UAV lithium battery based on a multi-modal lithium battery system, characterized in that, include: A lithium battery multimodal system is used to collect and analyze the remaining lifespan of the UAV's lithium battery. The lithium battery multimodal system includes: A battery data acquisition module, which is used to acquire multimodal battery data of the UAV's lithium battery; A data preprocessing module is used to perform preprocessing operations on multimodal battery data. A battery health assessment module, which is used to assess the data after preprocessing. A battery remaining life output module is used to output the remaining life based on the evaluation result. The method for monitoring the remaining lifespan of a drone's lithium battery includes: Obtain multimodal battery state parameters of drone lithium batteries; The state parameters of the multimodal battery are subjected to a health equivalent cycle assessment to obtain the number of battery cycles; The remaining lifespan of the drone's lithium battery is output based on the number of battery cycles.

2. The method for monitoring the remaining service life of a UAV lithium battery based on a multi-modal lithium battery system according to claim 1, characterized in that, Obtain multimodal battery state parameters of the drone's lithium battery, including: Obtain the electrical and operational parameters of the drone's lithium battery.

3. The method for monitoring the remaining service life of a UAV lithium battery based on a multi-modal lithium battery system according to claim 1, characterized in that, The multimodal battery state parameters are subjected to a health equivalent cycle assessment to obtain the battery cycle count, including: Feature extraction preprocessing is performed on the state parameters of multimodal batteries to obtain multiple battery state feature values.

4. The method for monitoring the remaining service life of a UAV lithium battery based on a multi-modal lithium battery system according to claim 3, characterized in that, Feature extraction preprocessing is performed on the multimodal battery state parameters to obtain battery state feature values, followed by: Multiple battery state characteristic values ​​are normalized and quantized separately to obtain multiple battery evaluation quantization values.

5. The method for monitoring the remaining service life of a UAV lithium battery based on a multi-modal lithium battery system according to claim 4, characterized in that, The state parameters of the multimodal battery are normalized and preprocessed to obtain multiple battery evaluation quantification values, followed by: The battery health assessment values ​​are weighted and summed to obtain the battery health assessment value.

6. The method for monitoring the remaining service life of a UAV lithium battery based on a multi-modal lithium battery system according to claim 5, characterized in that, The weighted summation of each battery assessment quantification value yields the battery health assessment value, which then includes: The battery health assessment value is evaluated and corrected to obtain the corrected battery health value.

7. The method for monitoring the remaining service life of a UAV lithium battery based on a multi-modal lithium battery system according to claim 6, characterized in that, The battery health assessment value is evaluated and corrected to obtain a corrected battery health value, which then includes: The battery health correction value is subjected to cyclic fitting to obtain the battery cycle count.

8. The method for monitoring the remaining service life of a UAV lithium battery based on a multi-modal lithium battery system according to claim 7, characterized in that, The battery health correction values ​​are subjected to iterative fitting, including: A bi-exponential health cycle decay curve fitting operation was performed on the battery health correction value.

9. The method for monitoring the remaining service life of a UAV lithium battery based on a multi-modal lithium battery system according to claim 7, characterized in that, The battery health correction values ​​are subjected to iterative fitting, including: A polynomial health cycle decay curve fitting operation is performed on the battery health correction value.

10. The method for monitoring the remaining service life of a UAV lithium battery based on a multi-modal lithium battery system according to claim 1, characterized in that, The remaining lifespan of the drone's lithium battery is output based on the stated number of battery cycles, including: The difference between the number of battery cycles and the total number of cycles is used to determine the remaining lifespan of the drone's lithium battery.

Images