A battery life prediction and management system and method based on cycle life modeling

Abstract

The application discloses a kind of battery life prediction and management system and method based on cycle life modeling, it is related to data processing technical field, including: the battery capacity data in battery cycle charging process is collected and pretreated;The change characteristics of battery capacity on the two sides of inflection point in battery capacity attenuation process are analyzed to construct local nonlinear degradation index;Based on the local nonlinear degradation index, the difference between inflection point and capacity regeneration phenomenon in battery capacity attenuation process is distinguished, and an asymmetric attenuation index is constructed;Based on the asymmetric attenuation index, the change characteristics of inflection point in the whole are analyzed, and an inflection point significant index is constructed;Based on the inflection point significant index, battery cycle charging data is labeled, which is used as a label in random forest algorithm, to realize the life prediction of battery.The application solves the problem that the generalization ability of the method with fixed preset empirical parameters is weak, and it cannot adapt to complex and variable actual scenes, which ultimately leads to inaccurate detection of inflection point.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a battery life prediction and management system and method based on cycle life modeling. Background Technology

[0002] With the rapid development of electric vehicles, energy storage power stations, and consumer electronics, lithium-ion batteries have been widely used as core energy storage components. However, battery performance undergoes irreversible degradation over time and use, and its cycle life is finite. Once the battery capacity has decreased to a certain level, it will be unable to meet normal usage requirements and may even pose safety hazards. Therefore, accurate prediction and management of battery health status and remaining lifespan are crucial for ensuring the safe and reliable operation of equipment, implementing predictive maintenance, and reducing total lifespan costs.

[0003] During battery capacity degradation, the formation and growth of the SEI film cause non-linear capacity decay, resulting in an inflection point. Before this inflection point, battery capacity decreases slowly, requiring less frequent maintenance. After the inflection point, however, capacity declines rapidly, necessitating frequent maintenance or timely battery replacement. Thus, battery capacity exhibits different degradation characteristics before and after the inflection point, necessitating different models for accurate battery life prediction. Existing methods typically rely on preset empirical parameters to detect the inflection point, such as degradation rate thresholds and capacity thresholds. However, battery capacity degradation data is often influenced by various factors, including temperature variations, charge-discharge cycle counts, and measurement errors. Furthermore, different batteries have different operating conditions and models; for example, a degradation rate threshold at room temperature may be inapplicable at low temperatures. This results in weak generalization ability of methods relying on fixed preset empirical parameters, failing to adapt to complex and changing real-world scenarios. Ultimately, this leads to inaccurate inflection point detection and fails to meet practical battery management needs.

[0004] Therefore, the present invention provides a battery life prediction and management system and method based on cycle life modeling. Summary of the Invention

[0005] The purpose of this invention is to provide a battery life prediction and management system and method based on cycle life modeling, so as to solve the existing problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a battery life prediction and management system based on cycle life modeling, comprising:

[0007] The data preprocessing module collects and preprocesses battery capacity data during the battery cycle charging process.

[0008] The local nonlinear degradation index calculation module analyzes the characteristics of battery capacity changes on both sides of the inflection point during battery capacity decay to construct a local nonlinear degradation index.

[0009] The asymmetric degradation index calculation module analyzes the difference between the inflection point and the capacity regeneration phenomenon during the battery capacity degradation process based on the local nonlinear degradation index, and constructs the asymmetric degradation index.

[0010] The inflection point significance index calculation module analyzes the global variation characteristics of inflection points based on the asymmetric decay index and constructs an inflection point significance index.

[0011] The battery life prediction module labels battery cycle charging data based on the inflection point significance index and uses it as a label in the random forest algorithm to predict battery life.

[0012] A further improvement of this invention is that the data preprocessing module uses a high-precision battery testing system to conduct full life cycle aging tests on multiple groups of lithium-ion batteries of the same model. This includes recording the charge-discharge curves of each cycle at a high sampling frequency, using the ampere-hour integration method to integrate the current during the discharge process over time, and constructing an original time series dataset with the number of charge-discharge cycles as the independent variable and the battery capacity as the dependent variable.

[0013] A further improvement of this invention is that the local nonlinear degradation index is used to reflect the significance of the nonlinear degradation trend of battery capacity within a local time window, and the construction process of the local nonlinear degradation index includes:

[0014] Taking the t-th charging cycle as the center, the total length is The window is denoted as the monitoring window, and the sequence of battery capacity data within the monitoring window is denoted as the local monitoring sequence of battery capacity. This sequence is used as the input of the least squares method for linear fitting, and the output is the fitted linear equation. Based on the linear equation, the fitted values ​​of battery capacity under different charging cycles are obtained. The difference between the battery capacity and the fitted value of battery capacity is denoted as the fitting residual. The sequence of the fitting residuals is denoted as the fitting residual sequence, and then the local nonlinear degradation index is obtained.

[0015] A further improvement of this invention is that the formula for calculating the local nonlinear degradation index is expressed as:

[0016] ;

[0017] in This represents the local nonlinear degradation exponent for the t-th charging cycle. This represents the maximum absolute value of all elements in the fitted residual sequence. `ln()` represents the logarithmic function with base to the natural constant, used to smooth the data within the parentheses. This represents the standard deviation of the fitted residual sequence. This represents the mean of the absolute values ​​of all elements in the fitted residual sequence, with 1 indicating a hyperparameter used to avoid... The result after taking the logarithm is less than 0.

[0018] A further improvement of the present invention is that, after calculating the local nonlinear degradation index after each charge, the asymmetric degradation index calculation module records the sequence of the local nonlinear degradation index within the monitoring window as the local nonlinear degradation index monitoring sequence; and uses the center point of the local battery capacity monitoring sequence as the dividing point to divide the local battery capacity monitoring sequence into a first subsequence and a second subsequence to obtain the asymmetric degradation index.

[0019] A further improvement of this invention is that the method for calculating the asymmetric decay exponent is as follows:

[0020] ;

[0021] in This represents the asymmetric decay exponent during the t-th charging cycle. This represents the mean of the local nonlinear degradation index monitoring sequence. and Let represent the ranges of the first and second subsequences, respectively. This represents hyperparameters.

[0022] A further improvement of this invention is that the inflection point significance index calculation module is used to calculate the asymmetric decay index of the battery capacity data after each charge, and then record the sequence of asymmetric decay indices of all battery capacity data as the asymmetric decay index sequence. This sequence is used as the input of the AMPD peak detection algorithm, and the output is the peak point of the sequence. The total number of peak points is recorded as K. Taking the kth peak point as an example, the neighborhood with the kth peak point as the center and a width of D is recorded as the peak neighborhood, thus obtaining the inflection point significance index.

[0023] A further improvement of this invention is that the construction process of the inflection point significance index is as follows:

[0024] ;

[0025] in This represents the significance index of the inflection point at the k-th peak in an asymmetric decay index sequence. Let represent the asymmetric decay exponent at the k-th peak point, and D represent the width of the peak neighborhood. This represents the Euclidean distance between the number of charges corresponding to the s-th battery capacity data point and the number of charges corresponding to the k-th peak point within the peak neighborhood. It represents the asymmetric degradation exponent of the s-th battery capacity data within the peak neighborhood.

[0026] A further improvement of this invention is that the battery life prediction module uses the number of charging cycles with the highest inflection index during the entire battery life cycle as the inflection point of the battery life. The region before the inflection point is designated as the slow degradation region, and the region after the inflection point is designated as the accelerated degradation region. The region where the data point is located is used as the label of the data point, and the feature vector of the data point is extracted. The feature vectors and labels of all data points are used as input to the random forest algorithm to train the decision tree in the random forest. After training, for the battery capacity data to be predicted, its feature vector is extracted as input to the trained random forest algorithm, and the output is the classification result of the battery capacity data, namely the slow degradation region and the accelerated degradation region.

[0027] On the other hand, the present invention provides a battery life prediction and management method based on cycle life modeling, comprising the following steps:

[0028] Step S1: Collect battery capacity data during the battery cycle charging process and perform preprocessing;

[0029] Step S2: Analyze the characteristics of battery capacity change on both sides of the inflection point during battery capacity decay to construct a local nonlinear degradation index;

[0030] Step S3: Based on the analysis of the local nonlinear degradation index, construct an asymmetric degradation index to distinguish the difference between the inflection point and the capacity regeneration phenomenon during the battery capacity decay process;

[0031] Step S4: Construct an inflection point significance index based on the global variation characteristics of the inflection point analysis using the asymmetric decay index;

[0032] Step S5: Label the battery cycle charging data based on the inflection point significance index and use it as a label in the random forest algorithm to predict the battery life.

[0033] Compared with the prior art, the beneficial effects of the present invention are:

[0034] 1. This invention first analyzes the characteristics of battery capacity change on both sides of the inflection point during battery capacity decay, and constructs a local nonlinear degradation index based on the change of residual; by analyzing the asymmetric change of decay rate before and after the inflection point, it identifies the characteristics of battery capacity regeneration phenomenon and the actual inflection point, and then constructs an asymmetric decay index to distinguish between capacity regeneration phenomenon and inflection point.

[0035] 2. By analyzing the isolation of the inflection point in the global context, the difference between the peak point of the asymmetric decay index and the asymmetric decay index of other points in the neighborhood is analyzed to construct an inflection point significance index. This assesses the significance of the isolation of candidate peak points and avoids errors in the judgment of inflection points caused by false peaks generated by strong local disturbances.

[0036] 3. Based on the inflection point significance index, identify the inflection point of battery capacity change in the dataset, label the battery capacity data, and extract feature vectors as input to the random forest algorithm to train decision trees, thereby achieving accurate classification of battery capacity changes. Based on the classification results, use different prediction models to achieve high-precision and robust classification and prediction of battery health status stages, providing a reliable decision basis for predictive maintenance and management of battery life. Attached Figure Description

[0037] Figure 1 This is a framework diagram of a battery life prediction and management system based on cycle life modeling according to the present invention.

[0038] Figure 2 This is a flowchart of a battery life prediction and management method based on cycle life modeling according to the present invention. Detailed Implementation

[0039] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0040] The term "and / or" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three cases: A exists alone, A and B exist simultaneously, and B exists alone.

[0041] Example 1

[0042] Figure 1 This embodiment illustrates a framework diagram of a battery life prediction and management system based on cycle life modeling, including:

[0043] The data preprocessing module collects and preprocesses battery capacity data during the battery cycle charging process. Data is obtained through battery cycle testing experiments, specifically: a high-precision battery testing system is used to conduct full-lifecycle cycle aging tests on multiple groups of the same model of lithium-ion batteries. The tests are conducted in a constant temperature and humidity chamber to eliminate environmental temperature interference. Standard constant current-constant voltage charging and constant current discharging modes are used. Voltage and current sensors that communicate directly with the battery management system record the charge-discharge curves of each cycle at a high sampling frequency. The ampere-hour integration method is used to integrate the current during the discharge process over time to accurately calculate the current actual capacity of the battery corresponding to each cycle. Thus, a raw time-series dataset is constructed with the number of charge-discharge cycles as the independent variable and battery capacity as the dependent variable.

[0044] After obtaining the raw dataset, necessary preprocessing is required to eliminate outliers and noise interference. First, an outlier detection method based on the Laida criterion is used to remove extreme abnormal capacity values ​​caused by accidental equipment communication interruptions or sensor jumps. Then, to smooth local capacity fluctuations caused by measurement noise and minor operating condition variations, while preserving the macroscopic trend of capacity degradation, this application employs the Savitzky-Golay filtering algorithm to smooth the capacity sequence. This method, through local polynomial least squares fitting, can effectively filter out noise while accurately maintaining key features such as the shape, width, and peak value of the curve, providing a high-quality data foundation for subsequent calculations.

[0045] The local nonlinear degradation index calculation module analyzes the characteristics of battery capacity changes on both sides of the inflection point during battery capacity decay to construct a local nonlinear degradation index.

[0046] During battery capacity degradation, before the inflection point, the degradation is dominated by relatively simple and stable electrochemical mechanisms such as SEI film growth, and the capacity decline trend exhibits a highly localized linear or quasi-linear pattern with a gradual rate of decay. After the inflection point, however, the degradation enters a faster but equally relatively stable accelerated phase, dominated by a combination of mechanisms such as lithium plating at the negative electrode. The inflection point is the point of instability transition between these two stable states. Therefore, the battery capacity changes before and after the inflection point can be well fitted by a linear trend, while the linear fitting effect for the battery capacity changes near the inflection point is poor.

[0047] Based on the above analysis, this application constructs a local nonlinear degradation index to reflect the significance of nonlinear degradation trend in battery capacity within a local time window. The construction process of the local nonlinear degradation index is as follows:

[0048] Taking the t-th charging cycle as an example, with the t-th charging cycle as the center, the total length is... The window is denoted as the monitoring window, and the size of L is set to 10 in this application, but can be selected according to the situation. The sequence of battery capacity data within the monitoring window is denoted as the local battery capacity monitoring sequence. This sequence is used as the input of the least squares method for linear fitting, and the output is the fitted linear equation. Based on the linear equation, the fitted battery capacity values ​​under different charging cycles are obtained. The difference between the battery capacity and the fitted value is denoted as the fitting residual, and the sequence of fitting residuals is denoted as the fitting residual sequence. The least squares method is a well-known technique and will not be described in detail here.

[0049] Based on the above processing steps, the calculation method of the local nonlinear degradation index in this application is as follows:

[0050] ;

[0051] in This represents the local nonlinear degradation exponent for the t-th charging cycle. This represents the maximum absolute value of all elements in the fitted residual sequence. `ln()` represents the logarithmic function with base to the natural constant, used to smooth the data within the parentheses. This represents the standard deviation of the fitted residual sequence. This represents the mean of the absolute values ​​of all elements in the fitted residual sequence, with 1 indicating a hyperparameter used to avoid... If the result after taking the logarithm is less than 0, you can choose to do so depending on the situation.

[0052] During the degradation of battery capacity, if the t-th charge cycle marks the inflection point of battery capacity change, then the rate of change of battery capacity on both sides of the inflection point differs significantly. This results in a worse fit of the fitted linear equation to the local monitoring sequence of battery capacity, leading to a larger residual term. The value is relatively large. When using the least squares method for linear fitting, the fitted straight line will pass through the middle part of the data. If the inflection point feature at time t is more obvious, the change feature of the concave function is more significant. When the fitted straight line passes through the battery capacity change curve, it produces a small residual near the intersection point and a large residual at other points. The standard deviation of the fitted residual sequence and the mean of the absolute values ​​of all elements in the fitted residual sequence are relatively large, which makes the final calculated local nonlinear degradation index larger.

[0053] The asymmetric degradation index calculation module analyzes the difference between the inflection point and the capacity regeneration phenomenon during the battery capacity degradation process based on the local nonlinear degradation index, and constructs the asymmetric degradation index.

[0054] During battery capacity degradation, a physical rebalancing process occurs internally during periods of rest or inactivity, leading to capacity regeneration. This includes the redistribution and diffusion of active lithium ions within the electrode materials, allowing some temporarily deactivated lithium ions to re-participate in electrochemical reactions. Macroscopically, this manifests as a temporary, small-scale rebound in capacity within a long-term degradation trend, potentially exhibiting inflection point characteristics within a local monitoring window. However, the inflection point is an irreversible state adjustment, marking an irreversible turning point where the dominant degradation mechanism within the battery undergoes a qualitative change. It typically transitions from linear degradation, primarily driven by slow SEI film growth, to nonlinear, malignant degradation with a self-accelerating effect, triggered by factors such as lithium plating at the negative electrode. In other words, capacity regeneration is a reversible, localized physical state adjustment that does not alter the long-term degradation rate. The inflection point, on the other hand, is an irreversible, global initiation of an electrochemical failure mode that permanently and significantly increases the battery's degradation rate.

[0055] The local nonlinear degradation index obtained from the above steps reflects the linear degradation law of battery capacity within a local monitoring window. Further differentiation is needed between battery capacity regeneration phenomena and the true inflection point of battery capacity. Therefore, this application constructs an asymmetric degradation index based on the local nonlinear degradation index. The construction process is as follows:

[0056] Calculate the local nonlinear degradation index after each charge according to the above steps, and record the sequence of local nonlinear degradation indices within the monitoring window as the local nonlinear degradation index monitoring sequence; use the center point of the local battery capacity monitoring sequence as the dividing point to divide the local battery capacity monitoring sequence into the first subsequence and the second subsequence.

[0057] Based on the above processing steps, the method for calculating the asymmetric attenuation index in this application is as follows:

[0058] ;

[0059] in This represents the asymmetric decay exponent during the t-th charging cycle. This represents the mean of the local nonlinear degradation index monitoring sequence. and Let represent the ranges of the first and second subsequences, respectively. This represents a hyperparameter used to avoid the denominator being zero, which would prevent calculation. In this application, the value is 0.1, but it can be selected according to the situation.

[0060] When battery capacity regeneration occurs, the change in battery capacity, although exhibiting a concave function shape, results in a relatively large calculated local nonlinear degradation exponent. However, this increase is only a small one within the long-term degradation trend, and it gradually decreases after the initial rise. This differs from the rapid decline after the inflection point. In other words, when battery capacity regeneration occurs, the difference between the ranges of the first and second subsequences is relatively small. The smaller value results in a smaller calculated asymmetric degradation exponent. If the battery capacity reaches the true inflection point, the capacity decreases slowly before the inflection point and rapidly after it, causing the battery capacity data near the inflection point to not satisfy the linear degradation characteristics. This corresponds to a larger local nonlinear degradation exponent, and the range of the second subsequence will be significantly larger than the range of the first subsequence. The larger value results in a larger asymmetric decay exponent in the final calculation.

[0061] The inflection point significance index calculation module analyzes the global variation characteristics of inflection points based on the asymmetric decay index and constructs an inflection point significance index.

[0062] Throughout the battery's lifespan, the inflection point of battery capacity change is an isolated and significant point in time. The asymmetric degradation index obtained in the above steps reflects the similarity between the local change characteristics of battery capacity and the inflection point change characteristics. Therefore, for global battery capacity data, the asymmetric degradation index near the inflection point will exhibit a significant isolated peak. In actual charging and discharging processes, noise factors may interfere with some battery capacity data, causing spurious peaks in the asymmetric degradation index. Directly selecting the charging count corresponding to the maximum value of the asymmetric degradation index as the inflection point may lead to misjudgment. Therefore, this application constructs an inflection point significance index based on the asymmetric degradation index to identify the true inflection point of battery capacity change. The construction process of the inflection point significance index is as follows:

[0063] Following the same steps as described above, calculate the asymmetric decay index of the battery capacity data after each charge. The sequence of asymmetric decay indices for all battery capacity data is denoted as the asymmetric decay index sequence. This sequence is used as the input to the AMPD peak detection algorithm, and the output is the peak point of this sequence. The total number of peak points is denoted as K. Taking the k-th peak point as an example, the neighborhood centered at the k-th peak point with a width of D is denoted as the peak neighborhood. The AMPD peak detection algorithm is a well-known technique and will not be elaborated upon here.

[0064] Based on the above processing steps, the construction process of the inflection point significance index in this application is as follows:

[0065] ;

[0066] in This represents the significance index of the inflection point at the k-th peak in an asymmetric decay index sequence. Let represent the asymmetric decay exponent at the k-th peak point, and D represent the width of the peak neighborhood, which is set to 10 in this application, but can be selected according to the situation. This represents the Euclidean distance between the number of charges corresponding to the s-th battery capacity data point and the number of charges corresponding to the k-th peak point within the peak neighborhood. It represents the asymmetric degradation exponent of the s-th battery capacity data within the peak neighborhood.

[0067] If the battery capacity reaches an inflection point, the asymmetric degradation exponent at the peak point will be significantly greater than that at other points in the peak neighborhood, i.e. Significantly greater than This results in a larger calculated inflection point significance index; however, if the peak value of the obtained asymmetric decay index is noise or random disturbance, the asymmetric decay index at the peak point is less significant compared to other points in the peak's neighborhood. It will not be significantly greater than This results in a smaller calculated inflection point significance index.

[0068] The battery life prediction module labels battery cycle charging data based on the inflection point significance index and uses it as a label in the random forest algorithm to predict battery life.

[0069] In the dataset, the number of charging cycles with the highest significance index during the entire battery lifespan is defined as the inflection point. The region before the inflection point is designated as the slow degradation region, and the region after the inflection point is designated as the accelerated degradation region. The region containing a data point is used as its label, and its feature vector is extracted. All data point feature vectors and labels are used as input to a random forest algorithm to train the decision tree. After training, for the battery capacity data to be predicted, its feature vector is extracted and used as input to the trained random forest algorithm. The output is the classification result of the battery capacity data, i.e., the slow degradation region or the accelerated degradation region. Here, the data point represents the battery capacity after a certain number of charging cycles, and the feature vector of the data point is a vector composed of a local nonlinear degradation index, an asymmetric decay index, and the battery capacity value.

[0070] If the classification result of the battery capacity data to be predicted is a slow degradation zone, the capacity decay in this zone is gradual, the trend is close to linear, and the noise impact is relatively controllable. Therefore, a simple and stable autoregressive model can be used to predict the battery life. If the classification result of the battery capacity data to be predicted is an accelerated degradation zone, the battery capacity decay rate in this zone is faster and the nonlinear characteristics are significant. Therefore, this application uses a Gaussian process regression model with strong nonlinear mapping ability to predict the battery life. At the same time, since the battery capacity has already experienced accelerated decay at this time, it may randomly affect the use of the equipment. Therefore, a targeted alarm is issued at this time to remind relevant personnel to replace and handle the problem in a timely manner.

[0071] By performing the above steps, the degradation characteristics of battery capacity at different stages are segmented and predicted, enabling high-precision prediction and management of battery life.

[0072] The threshold and weight settings can be set by default according to the present invention, or they can be set by those skilled in the art.

[0073] Example 2

[0074] Figure 2 This invention illustrates a flowchart of a battery life prediction and management method based on cycle life modeling. Based on the same inventive concept as Embodiment 1, this invention provides a battery life prediction and management method based on cycle life modeling, comprising:

[0075] Step S1: Collect battery capacity data during the battery cycle charging process and perform preprocessing.

[0076] Step S2: Analyze the characteristics of battery capacity change on both sides of the inflection point during battery capacity decay to construct a local nonlinear degradation index.

[0077] Step S3: Based on the analysis of the local nonlinear degradation index, an asymmetric degradation index is constructed to distinguish the difference between the inflection point and the capacity regeneration phenomenon during the battery capacity decay process.

[0078] Step S4: Construct a significant inflection point index based on the global variation characteristics of the inflection point by analyzing the asymmetric decay index.

[0079] Step S5: Label the battery cycle charging data based on the inflection point significance index and use it as a label in the random forest algorithm to predict the battery life.

[0080] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0081] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0082] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0083] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0084] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A battery life prediction and management system based on cycle life modeling, characterized in that: include: The data preprocessing module collects and preprocesses battery capacity data during the battery cycle charging process. The local nonlinear degradation index calculation module analyzes the characteristics of battery capacity changes on both sides of the inflection point during battery capacity decay to construct a local nonlinear degradation index. The asymmetric degradation index calculation module analyzes the difference between the inflection point and the capacity regeneration phenomenon during the battery capacity degradation process based on the local nonlinear degradation index, and constructs the asymmetric degradation index. The inflection point significance index calculation module analyzes the global variation characteristics of inflection points based on the asymmetric decay index and constructs an inflection point significance index. The battery life prediction module labels battery cycle charging data based on the inflection point significance index and uses it as a label in the random forest algorithm to predict battery life. After calculating the local nonlinear degradation index after each charge, the asymmetric degradation index calculation module records the sequence of the local nonlinear degradation index within the monitoring window as the local nonlinear degradation index monitoring sequence; taking the center point of the local battery capacity monitoring sequence as the dividing point, the local battery capacity monitoring sequence is divided into a first subsequence and a second subsequence to obtain the asymmetric degradation index. The asymmetric decay exponent is calculated as follows: ； in This represents the asymmetric decay exponent during the t-th charging cycle. This represents the mean of the local nonlinear degradation index monitoring sequence. and Let represent the ranges of the first and second subsequences, respectively. This represents hyperparameters.

2. The battery life prediction and management system based on cycle life modeling according to claim 1, characterized in that: The data preprocessing module uses a high-precision battery testing system to conduct full life cycle aging tests on multiple groups of lithium-ion batteries of the same model. This includes recording the charge-discharge curves of each cycle at a high sampling frequency, using the ampere-hour integration method to integrate the current during the discharge process over time, and constructing an original time series dataset with the number of charge-discharge cycles as the independent variable and the battery capacity as the dependent variable.

3. The battery life prediction and management system based on cycle life modeling according to claim 2, characterized in that: The local nonlinear degradation index is used to reflect the significance of the nonlinear degradation trend of battery capacity within a local time window. The construction process of the local nonlinear degradation index includes: Taking the t-th charging cycle as the center, the total length is The window is denoted as the monitoring window, and the sequence of battery capacity data within the monitoring window is denoted as the local monitoring sequence of battery capacity. This sequence is used as the input of the least squares method for linear fitting, and the output is the fitted linear equation. Based on the linear equation, the fitted values ​​of battery capacity under different charging cycles are obtained. The difference between the battery capacity and the fitted value of battery capacity is denoted as the fitting residual. The sequence of the fitting residuals is denoted as the fitting residual sequence, and then the local nonlinear degradation index is obtained.

4. A battery life prediction and management system based on cycle life modeling according to claim 3, characterized in that: The formula for calculating the local nonlinear degradation index is expressed as follows: ； in This represents the local nonlinear degradation exponent for the t-th charging cycle. This represents the maximum absolute value of all elements in the fitted residual sequence. `ln()` represents the logarithmic function with base to the natural constant, used to smooth the data within the parentheses. This represents the standard deviation of the fitted residual sequence. This represents the mean of the absolute values ​​of all elements in the fitted residual sequence, with 1 indicating a hyperparameter used to avoid... The result after taking the logarithm is less than 0.

5. A battery life prediction and management system based on cycle life modeling according to claim 4, characterized in that: The inflection point significance index calculation module is used to calculate the asymmetric decay index of battery capacity data after each charge, and then record the sequence of asymmetric decay indices of all battery capacity data as the asymmetric decay index sequence. This sequence is used as the input of the AMPD peak detection algorithm, and the output is the peak point of the sequence. The total number of peak points is recorded as K. Taking the k-th peak point as an example, the neighborhood with the k-th peak point as the center and a width of D is recorded as the peak neighborhood, thus obtaining the inflection point significance index.

6. A battery life prediction and management system based on cycle life modeling according to claim 5, characterized in that: The process of constructing the inflection point significance index is as follows: ； in This represents the significance index of the inflection point at the k-th peak in an asymmetric decay index sequence. Let represent the asymmetric decay exponent at the k-th peak point, and D represent the width of the peak neighborhood. This represents the Euclidean distance between the number of charges corresponding to the s-th battery capacity data point and the number of charges corresponding to the k-th peak point within the peak neighborhood. It represents the asymmetric degradation exponent of the s-th battery capacity data within the peak neighborhood.

7. A battery life prediction and management system based on cycle life modeling according to claim 6, characterized in that: The battery life prediction module uses the number of charging cycles with the highest significance index during the entire battery lifespan as the inflection point. The region before the inflection point is designated as the slow degradation zone, and the region after the inflection point as the accelerated degradation zone. The region containing a data point is used as its label, and the feature vector of the data point is extracted. All feature vectors and labels of all data points are used as input to a random forest algorithm to train the decision tree in the random forest. After training, for the battery capacity data to be predicted, its feature vector is extracted. The input to the Random Forest algorithm after training is the output of the battery capacity data classification results, namely the slow degradation region and the accelerated degradation region.

8. A battery life prediction and management method based on cycle life modeling, applied to a battery life prediction and management system based on cycle life modeling as described in any one of claims 1-7, characterized in that: Includes the following steps: Step S1: Collect battery capacity data during the battery cycle charging process and perform preprocessing; Step S2: Analyze the characteristics of battery capacity change on both sides of the inflection point during battery capacity decay to construct a local nonlinear degradation index; Step S3: Based on the analysis of the local nonlinear degradation index, construct an asymmetric degradation index to distinguish the difference between the inflection point and the capacity regeneration phenomenon during the battery capacity decay process; Step S4: Construct an inflection point significance index based on the global variation characteristics of the inflection point analysis using the asymmetric decay index; Step S5: Label the battery cycle charging data based on the inflection point significance index and use it as a label in the random forest algorithm to predict the battery life.

Images