Battery health state detection method and system

Abstract

The invention relates to the technical field of battery detection, and discloses a battery health state detection method and system, and the method comprises the steps: collecting voltage and current signals during the charging and discharging circulation of a battery, obtaining low-frequency and high-frequency components through wavelet transformation, and calculating the energy distribution to determine abnormal features; if the number of the abnormal features is smaller than a preset threshold value, reconstructing the signal based on low-frequency component sparse representation, otherwise, constructing a state classification tree based on the amplitude change of the high-frequency component and the preset threshold value, adjusting a traversal strategy according to the number of the abnormal features, and modifying the sampling frequency; and finally, generating a health state index. The system comprises a feature extraction module, a state analysis module and a health assessment module. According to the invention, through wavelet transformation, sparse representation and dynamic adjustment strategies, the precision and efficiency of battery health state detection are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of battery detection, and in particular to a battery health status detection method and system. Background Art

[0002] With the rapid development of new energy technologies, batteries are widely used as energy storage devices in electric vehicles, energy storage systems, and other fields. The battery's state of health (SOH) is a key indicator for evaluating battery performance and remaining life. Accurately measuring battery SOH is crucial for ensuring battery safety, optimizing battery management, and improving system reliability.

[0003] Traditional battery health status detection methods are mainly based on electrochemical models and empirical formulas. For example, they estimate the battery's health status by measuring parameters such as voltage, current, and temperature, combined with the ampere-hour integration method and equivalent circuit models. However, these methods have many limitations. First, during the charging and discharging process, the battery's internal electrochemical reactions and physical properties are complex and changeable. Traditional models have difficulty accurately depicting the battery's dynamic characteristics, resulting in low detection accuracy. Second, abnormal characteristics generated during battery aging (such as internal short circuits, plate corrosion, and active material shedding) are often nonlinear and non-stationary. Traditional methods have difficulty effectively capturing and identifying these subtle changes, which can easily lead to missed detections or false detections.

[0004] Existing detection methods typically use traditional signal processing techniques such as Fourier transforms. These methods have limited analytical capabilities for non-stationary signals and cannot meet the requirements for analyzing the time-frequency characteristics of battery signals. Furthermore, when faced with complex anomaly feature recognition and status classification problems, traditional methods lack adaptive adjustment strategies, making it difficult to optimize the detection process based on real-time changes in battery status, resulting in insufficient detection efficiency and reliability.

[0005] With the development of intelligent algorithms and signal processing techniques, technologies such as wavelet transform, sparse representation, and machine learning are gradually being applied to battery health status monitoring. Wavelet transform can perform multi-resolution analysis on signals, effectively extracting low-frequency trends and high-frequency details, making it suitable for processing non-stationary signals. Sparse representation theory can reconstruct signals using sparse dictionaries, preserving their key features. Machine learning models can automatically identify abnormal features and classify states through data training. However, the application of these technologies in battery health status monitoring still faces some challenges, such as how to rationally combine the time-frequency analysis capabilities of wavelet transform and the signal reconstruction advantages of sparse representation, how to construct efficient abnormal feature recognition models and state classification trees, and how to dynamically adjust detection strategies based on real-time data during the detection process. These issues urgently need to be addressed.

[0006] Therefore, there is an urgent need for a battery health status detection method and system that can effectively process battery non-stationary signals, accurately identify abnormal characteristics, and adaptively adjust detection strategies, so as to improve the accuracy, reliability and efficiency of battery health status detection and meet the urgent demand for battery management in the new energy field. Summary of the Invention

[0007] The purpose of the present invention is to provide a battery health status detection method and system to solve the problems raised in the above background technology.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a battery health status detection method, the method comprising: When the battery is in a charge and discharge cycle, the voltage and current signals of the battery are collected, and the collected signals are processed by wavelet transform to obtain low-frequency components and high-frequency components of the signals; the energy distribution of the low-frequency components and the energy distribution of the high-frequency components are calculated, and the abnormal characteristics of the battery are determined based on the energy distribution; If the number of abnormal features is less than the preset threshold, the original signal of the battery is reconstructed using sparse representation based on the low-frequency component; otherwise, a state classification tree is constructed based on the amplitude change of the high-frequency component and the preset threshold. If the number of levels in the classification tree is greater than 2, the traversal strategy of the state classification tree is adjusted according to the number of identified abnormal features and the total number of abnormal features. When the adjustment conditions are met, the sampling frequency of the signal acquisition is modified; The reconstructed battery signal is obtained and the corresponding health status indicator is generated in the health status assessment system.

[0009] Preferably, determining the abnormal characteristics of the battery according to the energy distribution is specifically: The energy distribution of the high-frequency component is used to normalize the energy distribution of the low-frequency component; The normalized energy distribution is input into the pre-established abnormal feature recognition model, and the number of abnormal features is output.

[0010] Preferably, the state classification tree is constructed based on the amplitude change of the high-frequency component and a preset threshold, specifically: The ratio of the maximum amplitude of the high-frequency component to the preset threshold is calculated. If the number of abnormal features is greater than or equal to 50% of the maximum amplitude, the number of branches of the state classification tree is the ratio; otherwise, the number of branches is the integer part of the ratio.

[0011] Preferably, the traversal strategy of the state classification tree is adjusted according to the number of identified abnormal features and the total number of abnormal features, specifically: If after traversing a certain level of the state classification tree, the total number of abnormal features minus the number of identified abnormal features is less than the preset threshold, the adjustment condition is triggered; Alternatively, after traversing the first 50% of the levels of the state classification tree, if the total number of abnormal features minus the number of identified abnormal features is less than a preset threshold, the adjustment condition is triggered.

[0012] Preferably, the sampling frequency of the modified signal acquisition is specifically: Adjust the sampling frequency to 1.5 times the original frequency; Alternatively, branches at adjacent levels in the state classification tree are merged, and the segmentation strategy of the sampling frequency is adjusted according to the merging result.

[0013] Preferably, the present invention further includes a battery health status detection system, the system comprising: A feature extraction module is used to collect the voltage and current signals of the battery when the battery is in a charge and discharge cycle, perform wavelet transform processing on the collected signals to obtain low-frequency and high-frequency components of the signals; calculate the energy distribution of the low-frequency components and the energy distribution of the high-frequency components, and determine the abnormal characteristics of the battery based on the energy distribution; A state analysis module is used to reconstruct the original signal of the battery using sparse representation based on the low-frequency component if the number of abnormal features is less than a preset threshold; otherwise, a state classification tree is constructed based on the amplitude change of the high-frequency component and the preset threshold. If the number of levels in the classification tree is greater than 2, the traversal strategy of the state classification tree is adjusted according to the number of identified abnormal features and the total number of abnormal features. When the adjustment conditions are met, the sampling frequency of the signal acquisition is modified; The health assessment module is used to obtain the reconstructed battery signal and generate the corresponding health status indicator in the health status assessment system.

[0014] Preferably, determining the abnormal characteristics of the battery according to the energy distribution is specifically: The energy distribution of the high-frequency component is used to normalize the energy distribution of the low-frequency component; The normalized energy distribution is input into the pre-established abnormal feature recognition model, and the number of abnormal features is output.

[0015] Preferably, the state classification tree is constructed based on the amplitude change of the high-frequency component and a preset threshold, specifically: The ratio of the maximum amplitude of the high-frequency component to the preset threshold is calculated. If the number of abnormal features is greater than or equal to 50% of the maximum amplitude, the number of branches of the state classification tree is the ratio; otherwise, the number of branches is the integer part of the ratio.

[0016] Preferably, the traversal strategy of the state classification tree is adjusted according to the number of identified abnormal features and the total number of abnormal features, specifically: If after traversing a certain level of the state classification tree, the total number of abnormal features minus the number of identified abnormal features is less than the preset threshold, the adjustment condition is triggered; Alternatively, after traversing the first 50% of the levels of the state classification tree, if the total number of abnormal features minus the number of identified abnormal features is less than a preset threshold, the adjustment condition is triggered.

[0017] Preferably, the sampling frequency of the modified signal acquisition is specifically: Adjust the sampling frequency to 1.5 times the original frequency; Alternatively, branches at adjacent levels in the state classification tree are merged, and the segmentation strategy of the sampling frequency is adjusted according to the merging result.

[0018] Compared with the prior art, the present invention has the following beneficial effects: In terms of signal processing, when the battery is in the charge and discharge cycle, the collected voltage and current signals are processed by wavelet transform, decomposing the signals into low-frequency and high-frequency components. The low-frequency component reflects the overall trend and main energy distribution of the signal, while the high-frequency component contains the signal's detailed characteristics and mutation information. By calculating the energy distribution of the low-frequency and high-frequency components separately, the time-frequency characteristics of the battery signal can be more comprehensively captured, providing a rich information foundation for subsequent abnormal feature identification and signal reconstruction. Compared with traditional methods such as Fourier transform, the wavelet transform has a stronger ability to analyze non-stationary signals and can more accurately reveal the dynamic changes of the battery during the charge and discharge process.

[0019] For abnormal feature identification, the energy distribution of the high-frequency component is used to normalize the energy distribution of the low-frequency component. This normalized energy distribution is then fed into a pre-established abnormal feature identification model, which outputs the number of abnormal features. This normalization eliminates the influence of varying signal scales, making the energy distribution features more comparable and robust. By learning and training from a large amount of historical data, the abnormal feature identification model automatically extracts key features related to battery health status, enabling accurate identification of battery abnormal features. This improves the sensitivity and accuracy of abnormal feature detection and reduces missed and false detections.

[0020] In terms of signal reconstruction and state analysis, different processing strategies are adopted according to the number of abnormal features. If the number of abnormal features is less than the preset threshold, it means that the battery state is relatively stable. At this time, sparse representation is used to reconstruct the original signal of the battery based on the low-frequency component. Sparse representation can compress and reconstruct the signal through a sparse dictionary, while retaining the main features of the signal and reducing the amount of data, thereby improving the efficiency and accuracy of signal processing. If the number of abnormal features is greater than or equal to the preset threshold, it means that the battery may have more serious abnormalities. At this time, a state classification tree is constructed based on the amplitude change of the high-frequency component and the preset threshold. The number of branches of the state classification tree is determined by calculating the ratio of the maximum amplitude of the high-frequency component to the preset threshold, so that the structure of the classification tree can adapt to the characteristic changes of the battery signal, thereby improving the accuracy and flexibility of state classification.

[0021] In terms of detection strategy adjustment, when the number of levels in the classification tree is greater than 2, the traversal strategy of the state classification tree is adjusted based on the number of identified abnormal features and the total number of abnormal features. If the number of unidentified abnormal features is less than the preset threshold after traversing a certain level or the first 50% of the levels, the adjustment condition is triggered and the sampling frequency of signal acquisition is modified. This dynamic adjustment strategy can optimize the detection process based on real-time data during the detection process. When an abnormal feature is detected and is close to being identified, by adjusting the sampling frequency (such as adjusting the sampling frequency to 1.5 times the original frequency, or merging adjacent level branches and adjusting the sampling frequency segmentation strategy), the detection efficiency is improved while ensuring detection accuracy and reducing unnecessary computing resource consumption.

[0022] In terms of health status assessment, the reconstructed battery signals are acquired and corresponding health status indicators are generated in the health status assessment system. These indicators can comprehensively and accurately reflect the health status of the battery, providing a reliable basis for decision-making in the battery management system. They help achieve real-time monitoring, life prediction, and optimized management of the battery, thereby improving the safety and economic efficiency of battery use. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a working principle diagram of the battery health status detection method according to the present invention; Figure 2 A flow chart is provided for determining anomaly characteristics based on energy distribution; Figure 3 Modify the flow chart for signal acquisition sampling frequency; Figure 4 Flowchart for abnormal feature determination in feature extraction module. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0025] See also Figures 1-4 The battery health status detection method of the present invention is specifically implemented in the following steps: The battery's voltage and current signals are collected and processed using wavelet transform, which decomposes the signals into low-frequency and high-frequency components. Wavelet transform effectively extracts the different frequency components in the signal. The low-frequency components primarily reflect the overall trend and energy distribution of the signal, while the high-frequency components correspond to detailed changes and abnormal characteristics in the signal.

[0026] The energy distribution of the low-frequency and high-frequency components is calculated, and the abnormal characteristics of the battery are determined based on these energy distributions. The energy distribution can be calculated by, for example, square-integrating the signal amplitudes of the corresponding frequency components. The abnormal characteristics are determined by analyzing the energy distribution to identify characteristic points that deviate from the normal state.

[0027] Determine the relationship between the number of abnormal features and the preset threshold: If the number of abnormal features is less than a preset threshold, indicating that the battery state is relatively stable, a sparse representation method is used to reconstruct the original battery signal based on the low-frequency components. Sparse representation approximates the original signal by finding sparse representation coefficients of the low-frequency components under a specific dictionary, removing the interference of high-frequency noise and retaining the main energy characteristics.

[0028] If the number of abnormal features is greater than or equal to the preset threshold, it indicates that the battery may have a significant abnormality. In this case, a state classification tree is constructed based on the amplitude changes of the high-frequency components and the preset threshold. If the number of levels in the constructed state classification tree is greater than 2, the state classification tree traversal strategy is further adjusted based on the number of identified abnormal features and the total number of abnormal features. When the adjustment conditions are met, the sampling frequency of the signal acquisition is modified.

[0029] The reconstructed battery signal is obtained and input into the health status assessment system, which generates the corresponding health status indicator, which is used to intuitively reflect the health status of the battery.

[0030] Example 1: During the battery health status detection process, the specific implementation method for determining the abnormal characteristics of the battery based on the energy distribution is as follows: When the battery is in the charge and discharge cycle, the voltage and current signals of the battery are first obtained through a signal acquisition device such as a sensor. These signals contain rich information during the battery operation process, including low-frequency components that reflect the basic working status of the battery, as well as high-frequency components that may characterize battery abnormalities. After the signal is collected, it needs to be processed by wavelet transform. As a time-frequency analysis tool, wavelet transform can decompose the original signal into components of different frequency bands, thereby realizing multi-resolution analysis of the signal. Through wavelet transform, the original voltage and current signals are decomposed into low-frequency components and high-frequency components. The low-frequency components mainly reflect the overall trend and energy distribution of the signal, while the high-frequency components focus on reflecting the detailed changes and mutation information in the signal. These mutation information are often associated with abnormal characteristics of the battery.

[0031] After completing the wavelet transform decomposition of the signal, the energy distribution of the low-frequency and high-frequency components needs to be calculated. The calculation of energy distribution is based on the amplitude of the signal. For low-frequency and high-frequency components, their energy can be obtained by integrating or summing the squares of the signal amplitudes within their respective frequency bands. The specific calculation method can be selected according to the type of signal and processing requirements. For example, for a discrete signal sequence, the energy can be approximately calculated by accumulating the squares of the amplitudes at each sampling point. The calculated low-frequency and high-frequency energy distributions can respectively reflect the energy consumption and distribution of the battery in different frequency bands, providing a data basis for subsequent abnormal feature analysis.

[0032] After obtaining the energy distribution of the low-frequency component and the high-frequency component, the energy distribution of the high-frequency component needs to be used to normalize the energy distribution of the low-frequency component. The purpose of normalization is to eliminate the influence of the energy scale differences between different components so that the two are comparable. The specific normalization method can be to divide the energy distribution of the low-frequency component by the sum of the energy distribution of the high-frequency component, or to use other appropriate normalization factors for adjustment to ensure that the normalized energy distribution value is within a reasonable range. Through normalization, the energy distribution of the low-frequency component and the high-frequency component can be converted into a relative energy distribution, highlighting the proportional relationship between the two, which makes it easier to analyze the relationship between changes in energy distribution and abnormal battery characteristics.

[0033] After normalization, the normalized energy distribution needs to be input into a pre-established abnormal feature recognition model to output the number of abnormal features. This abnormal feature recognition model is trained based on a large amount of historical data. Its construction requires collecting battery signal data under different health states and preprocessing this data, including signal acquisition, wavelet transform decomposition, energy distribution calculation, and normalization. This data is then used as the corresponding normalized energy distribution data input sample. Furthermore, the number of abnormal features corresponding to each sample must be determined manually or through other reliable methods to serve as the model's output label.

[0034] During the model training phase, machine learning algorithms, such as neural networks and support vector machines, can be used to build an abnormal feature recognition model. Taking a neural network as an example, it can consist of an input layer, a hidden layer, and an output layer. The number of nodes in the input layer corresponds to the dimensions of the normalized energy distribution data, and the number of nodes in the output layer is the predicted number of abnormal features. By inputting a large amount of normalized energy distribution data into the neural network and comparing it with the corresponding abnormal feature number labels, the backpropagation algorithm is used to adjust the neural network's weights and bias parameters to minimize the error between the model's predicted output and the actual labels, thereby achieving model training and optimization.

[0035] In actual application, when new normalized energy distribution data is input, the model performs feature extraction and pattern recognition on the input data based on the trained parameters and structure, ultimately outputting the number of abnormal features corresponding to this set of energy distribution data. It is important to note that the training data for the abnormal feature recognition model must be sufficiently diverse and representative, covering battery data of different types, ages, and operating conditions to ensure the model's generalization and accuracy in actual applications.

[0036] Throughout the entire processing process, signal acquisition accuracy and stability are crucial. The sensor must meet the requirements for battery testing, accurately capturing minute changes in voltage and current and maintaining reliable operation throughout the battery's charge and discharge cycles. Furthermore, anti-interference measures must be implemented during signal transmission to prevent external electromagnetic interference and other factors from affecting the quality of the collected signal. This ensures that subsequent wavelet transform processing and energy distribution calculations are based on authentic, valid signal data.

[0037] During the wavelet transform process, it is necessary to rationally select the wavelet basis function and the number of decomposition levels. Different wavelet basis functions have different time-frequency characteristics and are suitable for different types of signal analysis. Therefore, it is necessary to select the appropriate wavelet basis function based on the characteristics of the battery signal to ensure effective decomposition of low-frequency and high-frequency components. The number of decomposition levels also affects the effectiveness of signal decomposition. Too few decomposition levels may result in incomplete separation of low-frequency and high-frequency components, making it impossible to accurately extract the energy distribution. Too many decomposition levels may introduce excessive detail noise and increase computational complexity. Therefore, it is necessary to determine the optimal wavelet basis function and number of decomposition levels through experimentation and analysis to achieve the best decomposition of the battery signal.

[0038] When calculating energy distribution, the signal's duration and sampling frequency must be considered. Energy distribution is calculated based on the signal within a specific time window. The size of the time window should be appropriately set based on the battery's charge-discharge cycle and the timescale of abnormal characteristics. If the time window is too small, sufficient energy variation information may not be captured; if the time window is too large, some transient abnormal characteristics may be smoothed out. Furthermore, the sampling frequency affects the accuracy of energy distribution calculations. A higher sampling frequency provides more dense signal samples, but also increases the data processing workload; a lower sampling frequency may lose some critical details. Therefore, a trade-off must be made between computational accuracy and efficiency to select an appropriate time window and sampling frequency.

[0039] During the normalization process, care should be taken to prevent the denominator from being zero. When the energy distribution of the high-frequency components sums to zero, normalization cannot be performed. Therefore, it is necessary to check the energy distribution of the high-frequency components before processing. If the sum is zero, measures such as re-acquisition of the signal, adjustment of wavelet transform parameters, or use of other normalization methods can be taken to ensure the smooth progress of the normalization process.

[0040] Example 2: In the state analysis link of battery health status detection, the specific process of constructing a state classification tree based on the amplitude change of the high-frequency component and the preset threshold is as follows: When it is determined through preliminary processing that the number of abnormal features is greater than or equal to the preset threshold, it indicates that the battery may have a more significant abnormal condition, and it is necessary to further construct a state classification tree based on the amplitude characteristics of the high-frequency component to refine the analysis. First, the amplitude of the high-frequency component needs to be quantitatively analyzed. The core is to calculate the ratio of the maximum amplitude of the high-frequency component to the preset threshold. The preset threshold is a reference value pre-set based on multiple dimensional factors such as battery type, specifications, historical operating data and industry standards. It is used to measure whether the amplitude of the high-frequency component exceeds the normal fluctuation range. For example, for a certain type of lithium-ion battery, it is found through historical data statistics that the amplitude of the high-frequency component during normal operation usually does not exceed 8mV. The preset threshold can be set to 8mV. If the maximum amplitude of the high-frequency component in the actual detection is 12mV, the ratio is 12 / 8=1.5.

[0041] After calculating the ratio, the number of abnormal features is compared to 50% of the maximum amplitude to determine the degree of correlation between the abnormal features and the high-frequency component amplitude change. The 50% maximum amplitude here is not a simple numerical calculation, but an empirical threshold derived from battery failure mechanism analysis to reflect the potential impact of amplitude changes on the abnormal features. For example, if the maximum amplitude is 10mV, 50% of it is 5mV. If the number of abnormal features is greater than or equal to 5 (corresponding to a 5mV ratio), the abnormal features are considered to be highly correlated with the high-frequency component amplitude change. In this case, the number of branches in the state classification tree is directly set to the calculated ratio. Assuming the ratio is 2.3, given that the number of branches in the classification tree must be an integer, the ratio can be rounded up or down based on actual conditions. If the raw value of the ratio is used as a reference for the number of branches, it may be necessary to adjust it to an integer number based on project requirements. For example, 2.3 could be adjusted to 2 or 3 branches. The specific number can be determined based on the classification tree construction rules.

[0042] If the number of abnormal features is less than 50% of the maximum amplitude, the contribution of the high-frequency component amplitude change to the abnormal features is relatively low. In this case, the number of branches in the state classification tree is the integer part of the ratio. For example, if the ratio is 2.6 and the number of abnormal features is less than 1.3 (the threshold for the number of abnormal features corresponding to 50% of the maximum amplitude), the number of branches is 2. The logic behind this method of determining the number of branches is that when the correlation between amplitude changes and abnormal features is weak, the classification tree structure is simplified to avoid overfitting and improve classification efficiency.

[0043] When constructing a state classification tree, the tree's hierarchical structure and branching logic must be clarified. Each level of the state classification tree corresponds to a judgment of the battery state, and the branches represent different state paths. For example, the first layer can be set as a first-level threshold to determine whether the amplitude of the high-frequency component exceeds a preset threshold. If it exceeds, it enters a certain branch, otherwise it enters another branch; the next level can further perform subdivision judgments based on ratios or other derived parameters (such as amplitude change rate). The dynamic adjustment mechanism of the number of branches enables the classification tree to adaptively adjust its structure according to the relationship between the number of abnormal features detected in real time and the amplitude change, thereby more accurately mapping the actual state differences of the battery.

[0044] When constructing a classification tree, the accuracy of data preprocessing must also be considered. Extracting the amplitude of the high-frequency component requires precise signal acquisition and processing, including anti-aliasing filtering and analog-to-digital conversion accuracy control. Noise interference during signal acquisition can lead to deviations in the calculated amplitude of the high-frequency component, affecting the accuracy of the number of branches in the classification tree. Therefore, during the signal acquisition phase, sensors with high signal-to-noise ratios and anti-interference circuit designs are required to ensure that the collected voltage and current signals truly reflect the actual operating status of the battery.

[0045] A dynamic update mechanism for preset thresholds is also crucial. As battery life increases and environmental conditions change, the amplitude range of the high-frequency component during normal operation may drift. Therefore, the preset thresholds need to be regularly calibrated based on the latest historical data. For example, the system can be set to automatically recalculate the preset thresholds based on the normal signal data during every 100 charge and discharge cycles to adapt to battery performance degradation or environmental changes.

[0046] In the hierarchical design of the classification tree, if the number of levels in the initially constructed classification tree is greater than two (e.g., three or more), the strategy needs to be adjusted during subsequent traversals based on abnormal feature identification. However, this falls within the scope of other embodiments; here, the focus is on the logic for building the number of branches. The algorithm for constructing the classification tree can adopt a classic decision tree algorithm (such as ID3 and C4.5), selecting the optimal partitioning attributes through indicators such as information gain and gain rate to ensure that the branches at each level can maximize the separation of battery data in different states.

[0047] It's important to note that building a state classification tree isn't a one-time process; it's a dynamic optimization process. After the initial construction, the tree structure can be pruned or expanded based on subsequent test data to improve the model's generalization capabilities. For example, if a branch is found to have too few sample data, pruning can remove that branch to avoid classification bias caused by data sparsity. If new, unusual patterns are discovered, branches can be added at the corresponding level to refine the classification.

[0048] In practical applications, the amplitude variation of high-frequency components may exhibit nonlinear characteristics. Therefore, when calculating the ratio, nonlinear transformations such as logarithmic and exponential transformations can be considered to better fit the relationship between amplitude variation and the number of abnormal features. For example, when the amplitude variation exceeds a certain range, its impact on the abnormal features may increase exponentially. In this case, nonlinear transformations can make the ratio calculation more consistent with actual physical laws.

[0049] Example 3: In battery health status monitoring, the specific method for adjusting the state classification tree traversal strategy based on the number of identified abnormal features and the total number of abnormal features is as follows: When the state classification tree is constructed and the number of levels is greater than 2, the dynamic relationship between the number of identified abnormal features and the total number of abnormal features must be monitored in real time during the traversal process to trigger corresponding adjustment conditions and optimize the traversal path. The entire process involves traversal level progress judgment, difference calculation, conditional triggering logic, and strategy adjustment mechanism, and requires a refined design based on the real-time processing requirements of battery signals.

[0050] The process of traversing the state classification tree is a layer-by-layer decision-making process, and the judgment nodes at each level correspond to different characteristic attributes or threshold conditions. For example, the first layer may make branch judgments based on the ratio of the maximum amplitude of the high-frequency component to the preset threshold, and the second layer may be subdivided based on the energy distribution differences in a certain frequency band. Subsequent layers gradually delve into more specific abnormal features. During the traversal process, the system will select the corresponding branch based on the judgment result of the current node and accumulate the number of identified abnormal features. The total number of abnormal features refers to the total number of abnormal features that may exist in the battery, which is preliminarily determined through the early energy distribution analysis and other links. This value is output by the abnormal feature recognition model at the beginning of the detection or manually preset (if it is a known scenario).

[0051] The first situation that triggers the adjustment condition is: after traversing a certain level of the state classification tree, calculate the difference between the total number of abnormal features and the number of identified abnormal features, and compare the difference with the preset threshold. The "certain level" here refers to any complete level in the classification tree, such as the 3rd or 5th level. The specific level to be traversed for judgment can be determined based on the tree structure design or real-time data processing requirements. The preset threshold is a small positive integer (such as 1, 2 or 3) that is used to characterize the acceptable range of the number of remaining unidentified abnormal features. For example, assuming the total number of abnormal features is 8 and the preset threshold is 2, when traversing the second level, 6 abnormal features have been identified, and the difference is 2 (equal to the preset threshold), it is considered that the number of remaining unidentified abnormal features is already at a low level, and the current traversal path may be redundant or inefficient, and the adjustment condition needs to be triggered.

[0052] The second triggering condition occurs when the first 50% of the state classification tree is traversed. Similarly, the difference between the total number of abnormal features and the number of identified abnormal features is calculated and compared to a preset threshold. The calculation of the "first 50% of levels" is based on the total number of levels in the classification tree. For example, if the classification tree has 6 levels, the first 50% of levels are the first 3 levels. If the total number of levels is an odd number (such as 7), the first 50% of levels can be rounded up or down (e.g., the first 3 or first 4 levels, the specific rules need to be clearly defined during system design). For example, if the total number of levels is 8 and the first 4 levels are traversed, if the total number of abnormal features is 10, 8 abnormal features have been identified, and the difference is 2 (less than the preset threshold of 3), the adjustment condition is triggered. The design logic of this condition is that if abnormal feature identification is nearly complete by the time half of the levels are traversed, it means that the judgment of subsequent levels may not need to be carried out according to the original path. By adjusting the traversal strategy, the path can be optimized in advance to reduce computing resource consumption.

[0053] After the adjustment conditions are triggered, the system needs to adjust the traversal strategy according to the preset rules. Specific ways to adjust the strategy include but are not limited to the following: First, change the traversal order, such as switching from depth-first traversal to breadth-first traversal, or skipping certain branches that have been proven to be irrelevant to the remaining abnormal features; second, dynamically merge adjacent branches. When the abnormal features corresponding to several branches have been fully identified, merge them into one path to simplify subsequent judgments; third, adjust the threshold parameters of the judgment node, such as increasing or decreasing the judgment threshold in subsequent levels to identify the remaining abnormal features more sensitively or more leniently. The selection of these adjustment methods needs to be based on the structural characteristics of the classification tree and the real-time data characteristics. For example, if the remaining abnormal features are mainly concentrated in a specific branch, the traversal order or parameters of the branch should be adjusted first.

[0054] During the implementation process, attention should be paid to the following key technical points: First, the real-time statistics and synchronization of the number of abnormal features. The number of identified abnormal features needs to be updated in a timely manner after each node traversal, and ensure that it is consistent with the calculation benchmark of the total number of abnormal features. This requires the system to have efficient data storage and real-time computing capabilities. For example, a cache mechanism can be used to store the current traversal progress and feature quantity data to avoid judgment bias due to data delays. The second is the rationality of the preset threshold setting. The preset threshold needs to comprehensively consider the accuracy requirements and computing efficiency of battery detection. If the threshold is too small, it may cause the adjustment conditions to be triggered frequently, increasing the system burden; if the threshold is too large, the traversal path may not be optimized in time, affecting the detection speed. It is recommended to determine the initial threshold through historical data simulation or engineering experience, and provide a user-configurable interface for subsequent optimization.

[0055] Furthermore, the hierarchical division of the classification tree must match the complexity of the abnormal features. If the battery abnormality features are relatively simple, the classification tree may have fewer levels, and the judgment of the first 50% of levels may not fully reflect the feature distribution. If the abnormal features are complex and diverse, and the number of levels is large, it is necessary to ensure that the first 50% of levels can capture the main feature differences. Therefore, when constructing the classification tree, it is necessary to use feature importance analysis (such as the Gini coefficient and information gain ratio) to rationally allocate the hierarchical structure, so that the higher levels prioritize the processing of high-impact abnormal features, while the lower levels process the details.

[0056] When adjusting traversal strategies, it's also important to avoid missing abnormal features due to the adjustments. For example, when skipping a branch, logical checks must be performed to ensure that the abnormal features corresponding to that branch have been fully identified or are truly unrelated to the remaining features. This can be achieved by adding a "safety check" mechanism to the classification tree nodes. Specifically, when deciding to skip a branch, the probability of the branch containing unidentified abnormal features is first verified. If the probability exceeds a certain threshold (e.g., 5%), the traversal path for that branch is retained.

[0057] It's worth noting that adjusting the traversal strategy isn't a one-time operation; it's a dynamic, cyclical process. Each time an adjustment condition is triggered and strategy optimization is complete, the system continues traversing subsequent levels and monitors new differential changes in real time. If the adjustment condition is met again, the adjustment logic is repeated. This dynamic optimization mechanism ensures that the classification tree traversal process consistently adapts to real-time changes in battery status, improving the efficiency and accuracy of identifying abnormal features.

[0058] In engineering implementation, traversing and adjusting the state classification tree can be accomplished through software algorithms. For example, a recursive function can be used to implement a depth-first traversal, updating the number of identified abnormal features after each recursive call and modifying recursive parameters (such as skipping certain child nodes) when conditions are met. For embedded systems with limited computing resources, the algorithm's time and space complexity must also be considered to avoid system response delays caused by traversal strategy adjustments.

[0059] Example 4: During the battery health status detection process, when the adjustment conditions are met, the specific implementation of modifying the sampling frequency of signal acquisition includes the following two cases: 1. The sampling frequency is adjusted to a fixed multiple of the original frequency When the adjustment condition is triggered (such as the abnormal feature quantity difference requirement is met during the state classification tree traversal), the sampling frequency can be adjusted to 1.5 times the original frequency. It refers to the default sampling frequency set in the initial stage of detection. Its value needs to be determined according to the battery type, charge and discharge rate and signal characteristic frequency. It usually needs to satisfy the Nyquist sampling theorem, that is, (in is the highest effective frequency component in the signal). For example, if the original sampling frequency is 100Hz (corresponding to the signal component with the highest effective sampling frequency of 50Hz), it becomes At this time, signal components up to 75Hz can be collected, thereby capturing detailed changes in higher frequencies, which is suitable for scenarios that require detailed analysis of high-frequency abnormal characteristics.

[0060] The implementation of this adjustment method relies on the hardware configurability of the signal acquisition module, such as the use of an analog-to-digital converter (ADC) with a programmable sampling rate. At the software level, the system triggers a sampling rate switch instruction based on the adjustment condition detected. This instruction modifies the ADC control register through the hardware driver and reconfigures the sampling clock frequency. When adjusting the sampling frequency, it is necessary to ensure the continuity of signal acquisition to avoid data loss or acquisition misalignment due to frequency switching. To achieve this, the current sampling cycle can be processed before the switch, and a short buffer period (such as collecting signals for several cycles) can be performed after the switch to stabilize the sampling timing at the new frequency.

[0061] Furthermore, it's important to note the potential increase in data volume associated with increased sampling frequency. For example, if the original sampling frequency collects 100 data points per second, the adjusted frequency increases to 150 data points per second, a 50% increase in data volume. This places higher demands on the computing power of data storage and subsequent processing modules. Therefore, when designing the system, it's important to evaluate the hardware's processing bandwidth or employ data compression algorithms (such as differential encoding or wavelet compression) to reduce data transmission and storage pressures.

[0062] 2. Sampling frequency segmentation adjustment based on state classification tree branch merging The second way to modify the sampling frequency is to merge the branches of adjacent levels in the state classification tree and adjust the segmentation strategy of the sampling frequency based on the merge result. The specific steps are as follows: Branch merging logic: The merging of branches at adjacent levels means merging branches at adjacent levels (such as Layer and Branches with similar characteristics (layers) are merged. Similarity can be determined based on the amplitude range of the high-frequency components of the branches, the energy distribution pattern, or the type of abnormal characteristics. For example, if the amplitude change corresponding to two adjacent branches is less than 1.2 times the preset threshold and the abnormal characteristic type is "slight voltage fluctuation", they can be merged into a composite branch.

[0063] Segmentation strategy design: After merging branches, the sampling frequency is divided into several segmentation intervals according to the number and feature distribution of the remaining branches. Assume that after merging, There are effective branches, which can divide the sampling frequency into Each segment corresponds to a different sampling frequency range. For example, if there are three branches left after merging, corresponding to "normal state", "slight abnormality" and "serious abnormality", three sampling frequency segments can be set: Segment 1 (normal state): using the original frequency , meet basic detection accuracy requirements and reduce data processing volume; Segment 2 (slightly abnormal): uses medium frequency , used to capture feature changes corresponding to slight anomalies; Segment 3 (Severe Abnormality): Using high frequency , high-resolution acquisition of rapidly changing features of severe anomalies.

[0064] Dynamic switching mechanism: The system automatically selects the appropriate sampling frequency segment based on the current branch state. For example, when the classification tree traversal results enter the "Severe Abnormal" branch, high-frequency sampling is triggered, and the signal is collected at 1.8 times the original frequency. If the system returns to the "Normal State" branch, it switches back to the original frequency. This segmentation strategy ensures detection accuracy while avoiding the resource waste caused by using high-frequency sampling throughout the process.

[0065] When implementing branch merging, you need to establish clear merging rules. For example, define the merge threshold , when the difference in characteristic parameters (such as amplitude mean, energy distribution variance, etc.) of two adjacent branches is less than The selection of characteristic parameters should be related to the battery abnormality mechanism. For example, for lithium-ion batteries, the amplitude standard deviation of the high-frequency component and energy entropy can be selected as similarity measurement indicators.

[0066] The segmented adjustment of the sampling frequency needs to be linked in real time with the traversal process of the state classification tree. To this end, the sampling frequency configuration information can be embedded in each node of the classification tree. When traversing to a certain node, the system reads the segmentation parameters corresponding to the node and sends a frequency switching instruction to the signal acquisition module. At the same time, an anti-shake mechanism for frequency switching needs to be designed to avoid frequent switching of the sampling frequency due to short-term branch fluctuations, which affects the stability of the signal. For example, setting the minimum hold time , after entering a branch, the sampling frequency must be maintained at least The duration of the process, and then decide whether to switch based on the subsequent traversal results.

[0067] Key technical points and implementation details ① Hardware compatibility: Whether it is fixed-multiple adjustment or segmented adjustment, the signal acquisition hardware must support programmable sampling rate. Common implementation methods include: Use an acquisition module with a digital signal processor (DSP) or field programmable gate array (FPGA) to dynamically modify the sampling clock through software programming; A multi-rate sampling circuit is used to switch different clock sources (such as crystal oscillators) through analog switches to achieve frequency switching.

[0068] ② Anti-aliasing: When the sampling frequency increases, the cutoff frequency of the anti-aliasing filter needs to be adjusted accordingly to prevent high-frequency noise from entering the acquisition channel. For example, when the original frequency is 100Hz, the anti-aliasing filter cutoff frequency is set to 50Hz. After adjusting to 150Hz, the cutoff frequency needs to be increased to 75Hz to meet the requirements of the sampling theorem.

[0069] ③ Data synchronization and calibration: After the sampling frequency is modified, the collected data needs to be timestamped to ensure that the signal data at different frequencies are aligned on the time axis. This can be achieved through a global clock synchronization mechanism or by embedding frequency identification information in the data frame.

[0070] ④ Computing resource allocation: Data processing at high sampling frequencies requires higher computing power. Pipeline processing, parallel computing (such as GPU acceleration) or edge computing architecture can be used to complete some signal preprocessing (such as wavelet transform) at the acquisition end, reducing the amount of data transmitted to the main processor.

[0071] Applicable scenarios and selection logic of the two adjustment methods Fixed multiple adjustment is suitable for scenarios where abnormal features are evenly distributed and detection accuracy needs to be improved globally. For example, when the number of abnormal features detected for the first time approaches a threshold, a comprehensive investigation can be performed by uniformly increasing the sampling frequency.

[0072] Segment adjustment after branch merging is suitable for scenarios where abnormal features have a clear hierarchical structure. For example, the status classification tree has clearly distinguished abnormal types of different severities. In this case, dynamically adjusting the sampling frequency according to the branch status can achieve "on-demand sampling" and optimize resource utilization efficiency.

[0073] In practical applications, the two approaches can be combined. For example, the overall sampling frequency can be increased by a fixed multiple, and then the sampling segments can be further subdivided based on the results of the classification tree branch merging, forming a multi-level optimized sampling strategy. Regardless of the approach used, the sampling frequency modification must not affect the normal charging and discharging process of the battery to ensure the reliability and safety of the detection system.

[0074] Example 5: During the battery health status detection process, the specific implementation of modifying the sampling frequency of signal acquisition when the adjustment conditions are met includes the following two cases. The technical principles, implementation process and key points are described in detail below: 1. The sampling frequency is adjusted to 1.5 times the original frequency In the battery health monitoring system, the original sampling frequency is a pre-set default frequency based on the basic characteristics of the battery, signal bandwidth, and detection accuracy requirements. When an adjustment condition is triggered during the traversal of the status classification tree (for example, after traversing a certain level or the first 50% of the levels, the number of remaining unidentified abnormal features is less than a preset threshold), the system activates a sampling frequency increase mechanism, adjusting the current sampling frequency to 1.5 times the original frequency.

[0075] The technical logic behind this adjustment is that when a potential battery anomaly is detected or when more detailed signal details need to be captured, increasing the sampling frequency can increase the number of signal sampling points per unit time, thereby more accurately restoring the true form of the signal, especially abnormal characteristics in the high-frequency band. For example, if the original sampling frequency is 200Hz and it is adjusted to 300Hz, the system can now collect 300 voltage and current signal sample points per second. Compared to the 200 sample points at the original frequency, it can capture transient changes in the signal more densely, helping to identify small abnormal fluctuations.

[0076] At the hardware implementation level, this adjustment relies on the programmable features of the signal acquisition module. Specifically, the analog-to-digital converter (ADC), as the core acquisition component, needs to support dynamic modification of the sampling clock frequency through software instructions. After detecting the adjustment conditions, the system software layer sends a frequency switching instruction to the ADC driver. The driver adjusts the sampling clock frequency to 1.5 times the original frequency by configuring the ADC's control register or external clock source. In this process, it is necessary to ensure that the switching of the sampling clock is seamless to avoid signal acquisition interruption or data misalignment due to clock jumps. This can usually be achieved in the following way: before switching the sampling frequency, wait for the current sampling cycle to complete, then enable the new sampling frequency at the beginning of the new clock cycle, and calibrate the first few sampling points to eliminate the timing deviation caused by the clock switching.

[0077] Increasing the sampling frequency will lead to a significant increase in the amount of data. Taking the original frequency of 200Hz as an example, the adjusted data volume increases from 200 sample points per second to 300, a 50% increase in data volume. This places higher demands on data storage, transmission, and processing modules. To meet this challenge, the system can use data compression technologies such as differential encoding and run-length encoding to compress the collected original signal in real time, reducing the data transmission bandwidth and storage capacity requirements. At the same time, in the design of signal processing algorithms, it is necessary to optimize the calculation process, such as using pipeline processing or parallel computing architecture to ensure that processing steps such as wavelet transform and energy distribution calculation can be completed in a timely manner at high data rates.

[0078] Increasing the sampling frequency also requires consideration of anti-aliasing filter matching. According to the Nyquist sampling theorem, the sampling frequency must be at least twice the highest frequency component of the signal to avoid aliasing distortion. Therefore, when the sampling frequency is adjusted to 1.5 times the original frequency, the anti-aliasing filter's cutoff frequency must also be increased accordingly to ensure that noise signals above half the new sampling frequency are effectively filtered out. For example, if the original frequency of 200Hz corresponds to an anti-aliasing filter cutoff frequency of 100Hz, the cutoff frequency for the adjusted sampling frequency of 300Hz should be increased to 150Hz to ensure the integrity and accuracy of the sampled signal.

[0079] 2. Merge adjacent level branches of the state classification tree and adjust the sampling frequency segmentation strategy When the number of levels in the state classification tree is greater than 2 and the adjustment condition is triggered, another way to modify the sampling frequency is to merge branches of adjacent levels and dynamically adjust the sampling frequency segmentation strategy based on the merging results. This process involves the structural optimization of the state classification tree and the coordinated design of the sampling strategy. The specific steps are as follows: ① Merging of adjacent level branches: The merging of adjacent level branches of the state classification tree is based on the similarity of branch features. The system will analyze the adjacent levels (such as Layer and The high-frequency component amplitude range, energy distribution pattern, or abnormal feature type corresponding to two adjacent branches (layers) are combined into a single composite branch if the difference in these feature parameters is less than a preset merging threshold. For example, if two adjacent branches have high-frequency component amplitudes in the ranges [5mV, 8mV] and [6mV, 9mV], respectively, and both exhibit "periodic voltage fluctuation" abnormality, the two branches are considered highly similar and are merged into a single branch with the amplitude range [5mV, 9mV] and the same abnormal feature type.

[0080] The core goal of branch merging is to simplify the classification tree structure and reduce unnecessary hierarchical judgments, while retaining the distinguishability of key abnormal features. The merging threshold should be set based on the specific battery type and historical data. For example, for lithium iron phosphate batteries, the amplitude difference threshold can be set to 2mV, and the energy distribution variance difference threshold can be set to 0.1J. When the difference in the characteristic parameters of adjacent branches is less than these thresholds, the merging operation is triggered.

[0081] ② Sampling frequency segmentation based on merging results: After the branches are merged, the system divides the sampling frequency into several segmented intervals based on the number and characteristics of the remaining branches. Each segment corresponds to a specific battery state or abnormality level and matches the corresponding sampling frequency. For example, if there are three branches remaining after the merger, corresponding to "normal state", "moderate abnormality" and "severe abnormality", the following three-segment sampling frequency strategy can be designed: Normal state segmentation: using the original sampling frequency , meet basic detection needs and balance data volume and detection accuracy; Moderate anomaly segmentation: using medium sampling frequency (e.g. 1.2 times the original frequency) to capture signal detail changes corresponding to moderate anomalies; Severe abnormal segmentation: use high sampling frequency (e.g. 1.8 times the original frequency) to perform high-frequency acquisition of rapidly changing features of severe anomalies.

[0082] The number and specific values ​​of the sampling frequency segments must be dynamically adjusted based on the characteristics of the merged branches. For example, if two branches remain after the merge ("normal" and "abnormal"), a two-segment strategy can be used: the normal state maintains the original frequency, while the abnormal state switches to 1.5 times the original frequency. The segmentation strategy design must adhere to the "sampling on demand" principle. That is, while ensuring that abnormal characteristics can be identified, the proportion of time spent on high-frequency sampling should be minimized to reduce system resource consumption.

[0083] ③ Dynamic sampling frequency switching mechanism: When traversing the state classification tree, the system switches the sampling frequency segment in real time based on the current branch node. For example, when the classification tree traversal results enter the "severe anomaly" branch, the system immediately triggers the high-frequency sampling segment, acquiring signals at 1.8 times the original frequency; when the traversal results return to the "normal state" branch, it automatically switches back to the original frequency. To avoid frequent fluctuations in the sampling frequency due to rapid branch switching, a switching anti-shake mechanism can be set. For example, after entering a certain branch, the sampling frequency must be maintained for at least a period of time (such as five charge and discharge cycles), and then the decision to adjust is made based on the subsequent traversal results.

[0084] To implement dynamic switching, a mapping relationship between branch nodes and sampling frequency segments must be established. This mapping can be achieved by embedding metadata in the classification tree nodes, with each node storing its corresponding sampling frequency parameters (such as multiplication factors or absolute frequency values). When traversing to a node, the system reads the sampling frequency parameters and updates the configuration of the signal acquisition module through the hardware control interface, achieving real-time response to frequency switching.

[0085] 3. Key Technology Implementation and System Collaboration a. Hardware and software collaborative design: Whether it is fixed-multiple adjustment or segmented adjustment, the signal acquisition hardware must have flexible frequency configuration capabilities. Common implementation methods include: Use an ADC chip with an integrated programmable clock manager to dynamically modify the sampling clock division factor through the SPI or I2C interface; A field programmable gate array (FPGA) is used as the acquisition controller, and sampling clocks of different frequencies are generated in real time by reconfiguring the FPGA logic; Combined with multiple crystal oscillator clock sources, clock signals of different frequencies are switched through analog switches and input to the ADC.

[0086] At the software level, the system must seamlessly integrate the sampling frequency adjustment logic with the state analysis module. Upon detecting an adjustment condition, the state analysis module sends a frequency adjustment instruction to the acquisition control module through an event-driven mechanism. The acquisition control module interprets the instruction, executes the hardware configuration, and feeds back the adjustment results to the state analysis module, forming a closed-loop control system.

[0087] b. Data consistency and time synchronization: Sampling frequency modification may lead to inconsistent data rates collected at different time periods, so a data time synchronization mechanism needs to be established. Specific methods include: Embed a timestamp in each sampling data frame to record the absolute time or relative time offset of data acquisition; Global clock synchronization technology is used to ensure the clock reference consistency of the signal acquisition module, data processing module and health assessment module; Insert synchronization markers at the frequency switching points to facilitate identification of frequency change boundaries during subsequent data splicing and analysis.

[0088] c. Exception handling and fault tolerance: If a hardware failure occurs during the sampling frequency adjustment process (such as clock source failure or ADC configuration error), the system must be able to capture and recover from the exception. For example, if the sampling frequency cannot be adjusted as expected, it will automatically fall back to the original frequency, trigger a fault alarm, and record the fault log for subsequent troubleshooting. In addition, it is necessary to avoid the possibility of battery charge and discharge control being out of sync with signal acquisition due to frequency adjustment. For example, the charge and discharge control loop can be paused before the frequency is adjusted and resumed after the sampling frequency stabilizes to ensure that the battery's operating status is not affected by the adjustment of the detection system.

[0089] IV. Applicable Scenarios and Selection Logic of the Two Adjustment Methods Fixed-multiple adjustment is applicable to the following scenarios: When the number of abnormal features detected for the first time approaches the preset threshold, the sampling accuracy needs to be globally improved for comprehensive investigation; when the battery operating environment undergoes significant changes (such as a sudden temperature rise), the frequency range of the signal features may expand, requiring rapid adaptation through fixed-multiple adjustment; the state classification tree structure is relatively simple, with fewer levels, and no complex segmentation strategy is required.

[0090] Branch merging and segmentation adjustment is applicable to the following scenarios: the state classification tree has many levels (for example, more than 5 levels), and the abnormal characteristics have a clear hierarchical distribution (for example, abnormalities of different severities correspond to different frequency ranges); the battery abnormal characteristics show a dynamic evolution process, and the sampling accuracy needs to be dynamically matched according to the current abnormality level; the system resources are limited, and the "on-demand sampling" strategy needs to be used to optimize the allocation of computing and storage resources.

[0091] In practical applications, the two approaches can be combined. For example, when the adjustment condition is first triggered, the sampling frequency is initially increased by a fixed multiple to obtain richer signal data. As the state classification tree traversal progresses, the sampling frequency segmentation is further refined based on the branch merging results, forming a multi-level adjustment strategy of "global improvement + local optimization," thereby achieving a dynamic balance between sampling efficiency and accuracy at different detection stages.

[0092] By modifying the sampling frequency using these two methods, the battery health status monitoring system can dynamically adjust its signal acquisition strategy based on real-time analysis results, ensuring high-resolution capture of abnormal features while avoiding inefficient resource consumption. The core of this mechanism lies in the deep coupling of the structural characteristics of the state classification tree with signal acquisition parameters. This intelligent adjustment logic enables adaptive optimization of the detection process, providing critical technical support for accurate battery health assessment.

[0093] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0094] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A battery health status detection method, characterized in that: The method comprises: When the battery is in a charge and discharge cycle, the voltage and current signals of the battery are collected, and the collected signals are processed by wavelet transform to obtain low-frequency components and high-frequency components of the signals; the energy distribution of the low-frequency components and the energy distribution of the high-frequency components are calculated, and the abnormal characteristics of the battery are determined based on the energy distribution; If the number of abnormal features is less than the preset threshold, the original signal of the battery is reconstructed using sparse representation based on the low-frequency component; otherwise, a state classification tree is constructed based on the amplitude change of the high-frequency component and the preset threshold. If the number of levels in the classification tree is greater than 2, the traversal strategy of the state classification tree is adjusted according to the number of identified abnormal features and the total number of abnormal features. When the adjustment conditions are met, the sampling frequency of the signal acquisition is modified; The reconstructed battery signal is obtained and the corresponding health status indicator is generated in the health status assessment system.

2. The method according to claim 1, wherein The determining of abnormal characteristics of the battery according to the energy distribution is specifically as follows: The energy distribution of the high-frequency component is used to normalize the energy distribution of the low-frequency component; The normalized energy distribution is input into the pre-established abnormal feature recognition model, and the number of abnormal features is output.

3. The method according to claim 1, wherein The state classification tree is constructed based on the amplitude change of the high-frequency component and the preset threshold, specifically: The ratio of the maximum amplitude of the high-frequency component to the preset threshold is calculated. If the number of abnormal features is greater than or equal to 50% of the maximum amplitude, the number of branches of the state classification tree is the ratio; otherwise, the number of branches is the integer part of the ratio.

4. The method according to claim 1, wherein The traversal strategy of the state classification tree is adjusted according to the number of identified abnormal features and the total number of abnormal features, specifically: If after traversing a certain level of the state classification tree, the total number of abnormal features minus the number of identified abnormal features is less than the preset threshold, the adjustment condition is triggered; Alternatively, after traversing the first 50% of the levels of the state classification tree, if the total number of abnormal features minus the number of identified abnormal features is less than a preset threshold, the adjustment condition is triggered.

5. The method according to claim 1, wherein The sampling frequency of the modified signal acquisition is specifically: Adjust the sampling frequency to 1.5 times the original frequency; Alternatively, branches at adjacent levels in the state classification tree are merged, and the segmentation strategy of the sampling frequency is adjusted according to the merging result.

6. Battery health status detection system, characterized in that, The system comprises: A feature extraction module is used to collect the voltage and current signals of the battery when the battery is in a charge and discharge cycle, perform wavelet transform processing on the collected signals to obtain low-frequency and high-frequency components of the signals; calculate the energy distribution of the low-frequency components and the energy distribution of the high-frequency components, and determine the abnormal characteristics of the battery based on the energy distribution; A state analysis module is used to reconstruct the original signal of the battery using sparse representation based on the low-frequency component if the number of abnormal features is less than a preset threshold; otherwise, a state classification tree is constructed based on the amplitude change of the high-frequency component and the preset threshold. If the number of levels in the classification tree is greater than 2, the traversal strategy of the state classification tree is adjusted according to the number of identified abnormal features and the total number of abnormal features. When the adjustment conditions are met, the sampling frequency of the signal acquisition is modified; The health assessment module is used to obtain the reconstructed battery signal and generate the corresponding health status indicator in the health status assessment system.

7. The system according to claim 6, wherein: The determining of abnormal characteristics of the battery according to the energy distribution is specifically as follows: The energy distribution of the high-frequency component is used to normalize the energy distribution of the low-frequency component; The normalized energy distribution is input into the pre-established abnormal feature recognition model, and the number of abnormal features is output.

8. The system according to claim 6, wherein: The state classification tree is constructed based on the amplitude change of the high-frequency component and the preset threshold, specifically: The ratio of the maximum amplitude of the high-frequency component to the preset threshold is calculated. If the number of abnormal features is greater than or equal to 50% of the maximum amplitude, the number of branches of the state classification tree is the ratio; otherwise, the number of branches is the integer part of the ratio.

9. The system according to claim 6, wherein: The traversal strategy of the state classification tree is adjusted according to the number of identified abnormal features and the total number of abnormal features, specifically: If after traversing a certain level of the state classification tree, the total number of abnormal features minus the number of identified abnormal features is less than the preset threshold, the adjustment condition is triggered; Alternatively, after traversing the first 50% of the levels of the state classification tree, if the total number of abnormal features minus the number of identified abnormal features is less than a preset threshold, the adjustment condition is triggered.

10. The system according to claim 6, wherein: The sampling frequency of the modified signal acquisition is specifically: Adjust the sampling frequency to 1.5 times the original frequency; Alternatively, branches at adjacent levels in the state classification tree are merged, and the segmentation strategy of the sampling frequency is adjusted according to the merging result.

Images