Battery health state monitoring method and device, computer equipment and storage medium

Abstract

The invention relates to a battery health state monitoring method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring current operation data of a target battery; determining a target feature of the target battery on at least one target dimension according to the current operation data; determining a predicted capacity value of the target battery under at least one time step according to the target characteristics of the target battery on at least one target dimension on the basis of a trained use duration prediction model; the use duration prediction model is obtained based on long short-term memory network training; and determining the health state of the target battery according to the predicted capacity value of the target battery under the at least one time step. By adopting the method, the health state of the battery can be determined more accurately by using the duration prediction model.

Description

Technical Field

[0001] This application relates to the field of battery technology, and in particular to a method, apparatus, computer device, and storage medium for monitoring battery health status. Background Technology

[0002] In power applications, all-terrain intelligent transport equipment for distribution network operations is primarily used for safe and efficient transportation in complex terrains. In open and complex environments, the battery power system of this transport equipment is the power source ensuring reliable transport of tools between maintenance points throughout the distribution network operation. The power system uses lithium batteries for charging and discharging the equipment. Lithium batteries offer advantages such as high power density, high energy density, and long lifespan; however, their capacity typically decreases with increasing charge-discharge cycles. Prognostics and health management (PHM) methods, particularly the prediction of remaining useful life (RUL), are crucial components. RUL can be considered the length of time from the current moment until the end of the battery's total effective lifespan. Accurate RUL prediction for lithium battery-based power systems not only provides information on battery performance changes throughout its service life but also improves battery management techniques, extends battery life, and provides a more accurate understanding of battery operating parameters. Therefore, accurate prediction of lithium battery RUL plays a vital role in managing battery health and estimating battery condition.

[0003] In complex operating environments, frequent power outages or malfunctions in transport equipment can severely impact the efficiency and safety of live-line operations, potentially leading to work stoppages or equipment damage. To address this, predicting the health status of lithium batteries allows for appropriate maintenance measures to be taken before battery performance degrades, thus preventing work interruptions or other safety risks caused by battery failures. In practical applications, battery health is influenced by various factors, including temperature, humidity, charge / discharge rates, the battery's manufacturing process, and its usage history. Therefore, traditional prediction methods often fail to comprehensively consider these complex factors, resulting in significant prediction errors. To improve prediction accuracy, combining multiple data sources and applying advanced prediction models is crucial. Therefore, how to monitor the health level of the power system in real time, provide early warnings of potential failures, and accurately predict the battery's health status and remaining lifespan under various environmental and operating conditions is a pressing issue that needs to be addressed.

[0004] In traditional technologies, battery warning mechanisms typically rely on fixed thresholds or human experience for judgment. As a result, warnings are often delayed or fail, leading to incorrect battery maintenance strategies and inaccurate battery health monitoring. Summary of the Invention

[0005] Therefore, it is necessary to provide a battery health status monitoring method, device, computer equipment, and storage medium that can accurately monitor the battery health status in response to the above-mentioned technical problems.

[0006] Firstly, this application provides a method for monitoring battery health status, including:

[0007] Obtain the current operating data of the target battery;

[0008] Based on the current operating data, determine the target features of the target battery in at least one target dimension;

[0009] Based on the trained usage duration prediction model, the predicted capacity value of the target battery at at least one time step is determined according to the target features of the target battery in at least one target dimension; the usage duration prediction model is trained based on a long short-term memory network.

[0010] The health status of the target battery is determined based on the predicted capacity value of the target battery at at least one time step.

[0011] In one embodiment, determining the health status of the target battery based on the predicted capacity value of the target battery at at least one time step includes:

[0012] For each time step, the remaining usage time of the target battery at that time step is determined based on the preset battery failure time step and the difference between the time steps; and,

[0013] For each time step, the health status index of the target battery at the time step is determined based on the ratio between the predicted capacity value and the nominal capacity value of the target battery at the time step.

[0014] Based on the difference between the health status indicators of the previous time step and the time step, the rate of decline of the health status indicators of the target battery at the time step is determined.

[0015] If the health status indicator is less than the preset indicator threshold, or the remaining usage time is less than the preset duration threshold, or the rate of decline of the health status indicator is less than the preset rate threshold, the health status of the target battery is determined to be abnormal.

[0016] In one embodiment, the method for determining the battery failure time step includes:

[0017] Obtain the preset battery capacity threshold; the battery capacity threshold is a preset multiple of the nominal capacity value;

[0018] Based on the similarity between the battery capacity threshold and the predicted capacity value at different time steps, the first time step and the second time step are selected from each time step; the first time step and the second time step are adjacent, and the second time step is later than the first time step;

[0019] The battery failure time step is determined based on the first time step, the second time step, and the battery capacity threshold.

[0020] In one embodiment, selecting a first time step and a second time step from each time step based on the similarity between a battery capacity threshold and predicted capacity values ​​at different time steps includes:

[0021] The two time steps with the highest similarity to the battery capacity threshold are selected as candidate time steps;

[0022] When two candidate time steps are adjacent, the earlier time step is taken as the first time step, and the later time step is taken as the second time step.

[0023] If two candidate time steps are not adjacent, select one of the two candidate time steps as the first time step, and take the next time step of the first time step as the second time step; wherein the similarity of the first time step is not less than that of the other candidate time step.

[0024] In one embodiment, based on current operating data, determining the target features of the target battery in at least one target dimension includes:

[0025] Obtain a preset set of benchmark features; the elements in the set of benchmark features are obtained by extracting features from the benchmark operating data of the benchmark battery.

[0026] Based on the current operating data and the benchmark feature set, determine the feature distribution difference, Spearman correlation coefficient, and feature similarity between the current operating data and the benchmark feature set in at least one initial dimension;

[0027] For each initial dimension, the evaluation value of the target battery in the initial dimension is determined based on the feature distribution difference value, Spearman correlation coefficient, and feature similarity.

[0028] The initial dimension with an evaluation value greater than the preset evaluation threshold is selected as the target dimension, and the running characteristics of the current running data on the target dimension are used as the target features.

[0029] In one embodiment, obtaining a preset set of baseline features includes:

[0030] Obtain preset baseline operating data;

[0031] Based on a pre-trained feature extraction model, the baseline features of the target battery in at least one target dimension are determined according to the benchmark running data; the feature extraction model is trained based on a delay-tolerant network.

[0032] Construct a set of benchmark features based on benchmark features in at least one target dimension.

[0033] Secondly, this application also provides a battery health status monitoring device, comprising:

[0034] The acquisition module is used to acquire the current operating data of the target battery;

[0035] The feature module is used to determine the target features of the target battery in at least one target dimension based on the current running data.

[0036] The prediction module is used to determine the predicted capacity value of the target battery at at least one time step based on the target features of the target battery in at least one target dimension, according to the trained usage duration prediction model; the usage duration prediction model is trained based on a long short-term memory network.

[0037] The detection module is used to determine the health status of the target battery based on the predicted capacity value of the target battery at at least one time step.

[0038] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0039] Obtain the current operating data of the target battery;

[0040] Based on the current operating data, determine the target features of the target battery in at least one target dimension;

[0041] Based on the trained usage duration prediction model, the predicted capacity value of the target battery at at least one time step is determined according to the target features of the target battery in at least one target dimension; the usage duration prediction model is trained based on a long short-term memory network.

[0042] The health status of the target battery is determined based on the predicted capacity value of the target battery at at least one time step.

[0043] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0044] Obtain the current operating data of the target battery;

[0045] Based on the current operating data, determine the target features of the target battery in at least one target dimension;

[0046] Based on the trained usage duration prediction model, the predicted capacity value of the target battery at at least one time step is determined according to the target features of the target battery in at least one target dimension; the usage duration prediction model is trained based on a long short-term memory network.

[0047] The health status of the target battery is determined based on the predicted capacity value of the target battery at at least one time step.

[0048] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0049] Obtain the current operating data of the target battery;

[0050] Based on the current operating data, determine the target features of the target battery in at least one target dimension;

[0051] Based on the trained usage duration prediction model, the predicted capacity value of the target battery at at least one time step is determined according to the target features of the target battery in at least one target dimension; the usage duration prediction model is trained based on a long short-term memory network.

[0052] The health status of the target battery is determined based on the predicted capacity value of the target battery at at least one time step.

[0053] The aforementioned battery health status monitoring method, apparatus, computer equipment, and storage medium acquire the current operating data of the target battery; based on the current operating data, determine the target features of the target battery in at least one target dimension; based on a trained usage duration prediction model, determine the predicted capacity value of the target battery at at least one time step according to the target features of the target battery in at least one target dimension; the usage duration prediction model is trained based on a Long Short-Term Memory network; and based on the predicted capacity value of the target battery at at least one time step, determine the health status of the target battery. In this embodiment, real-time detection of battery health status is achieved through algorithmic feature extraction and sharing, thereby more accurately determining the battery health status by using the usage duration prediction model. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1This is an application environment diagram of a battery health status monitoring method provided in this embodiment;

[0056] Figure 2 This is a flowchart illustrating a battery health status monitoring method provided in this embodiment;

[0057] Figure 3 This is a flowchart illustrating a step for determining the battery failure time step in this embodiment;

[0058] Figure 4 This is a flowchart illustrating a step for determining target features in this embodiment;

[0059] Figure 5 This is a structural block diagram of a battery health status monitoring device provided in this embodiment;

[0060] Figure 6 This is an internal structural diagram of a computer device provided in this embodiment. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0062] The battery health status monitoring method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on the cloud or other network servers. The computer device acquires the current operating data of the target battery; based on the current operating data, it determines the target features of the target battery in at least one target dimension; based on a trained usage duration prediction model, and according to the target features of the target battery in at least one target dimension, it determines the predicted capacity value of the target battery at at least one time step; the usage duration prediction model is trained based on a Long Short-Term Memory network; based on the predicted capacity value of the target battery at at least one time step, it determines the health status of the target battery. The computer device can be either a terminal or a server. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, etc. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc. Server 104 can be implemented using a standalone server or a server cluster composed of multiple servers.

[0063] In one exemplary embodiment, such as Figure 2 As shown, a battery health status monitoring method is provided, which can be applied to... Figure 1 Taking a computer device as an example, the explanation includes the following steps S201 to S204. Wherein:

[0064] S201 acquires the current operating data of the target battery.

[0065] The current operating data includes at least one of the following: voltage, current, temperature, charge / discharge cycle, and other auxiliary data (ambient temperature, humidity, etc.).

[0066] In some embodiments, current operating data of the target battery under at least one dimension under different operating conditions is acquired for subsequent health status prediction. Various sensors and data acquisition cards are deployed at the edge nodes. The sampling frequency is set according to the dynamic characteristics of the battery; the sampling frequency for voltage, current, and temperature data is 1Hz. The data format is time-series data, containing timestamps and corresponding sensor values, such as [timestamp, voltage, current, temperature, charge / discharge cycle]. Sensor data can be read in real time through the data acquisition card, and the data is stored in the local memory and the edge computing node's memory. The data storage format is comma-separated values ​​(CSV) or a database for easy subsequent processing. The acquisition scenarios can be divided into operating conditions: Condition 1: ambient temperature charge / discharge (25℃); Condition 2: high temperature charge / discharge (40℃); Condition 3: low temperature charge / discharge (0℃). Data is collected for 1 hour under each operating condition, with a sampling frequency of 1Hz and a data volume of 3600 records. Each record occupies approximately 20 bytes, and the data volume for 1 hour is 72KB. The edge computing node storage capacity is ≥1GB to meet long-term data storage requirements.

[0067] It should be noted that, in order to further increase the accuracy of the current operating data, this embodiment can also preprocess the current operating data. Preprocessing may include at least one of signal denoising, outlier removal, and data normalization.

[0068] For example, signal denoising may include: smoothing each data point and its neighborhood using a second-order Savitzky-Golay Filter (SGF) with a polynomial fitting algorithm as shown in Equation (1-1) to obtain smoothed data.

[0069] (1-1)

[0070] in, This is the output of the i-th currently running data after filtering; y i+k For the (i+k)th currently running data; c kFor the corresponding y i+k The filter coefficients are given by the coefficient matrix of the polynomial fitting; m is half the size of the window (the window size is 2m+1); the window length and polynomial order of the filter are set by experiments or optimization methods to obtain the smoothed data after denoising.

[0071] For example, outlier removal may include: based on smoothed data, using a box plot method to detect and remove outliers from the filtered data to obtain de-outliered data; then using linear interpolation to fill in the missing values ​​at the time positions of the removed outliers using the adjacent valid data points before and after those positions. Specifically, the box plot method involves: calculating the quartiles (Q1 and Q3) of the smoothed data and determining the upper and lower limits of outliers (upper limit: Q1 - 1.5 * IQR, lower limit: Q3 + 1.5 * IQR, where IQR is the interquartile range). Data points exceeding these limits are identified as outliers and removed from the dataset. Linear interpolation is then used to fill in the missing values ​​at the time positions of the removed outliers using the adjacent valid data points before and after those positions.

[0072] For example, data normalization may include: performing maximum and minimum value normalization based on the padded data, mapping the data values ​​to the [0,1] interval, thus completing the data preprocessing. Data after outlier removal. After linear interpolation, the effective data obtained is standardized using the following formula (1-2) to map the data to the [0,1] interval, resulting in the preprocessed dataset. .

[0073] (1-2)

[0074] Where x represents the data to be normalized, x min The minimum value in the dataset, x max Represents the maximum value in the dataset.

[0075] S202 determines the target characteristics of the target battery in at least one target dimension based on the current operating data.

[0076] In some embodiments, target features of the target battery in at least one target dimension are determined based on a pre-trained feature extraction model and current running data. The feature extraction model may be trained using a Deep Transfer Network (DTN).

[0077] S203, based on a trained usage duration prediction model, determines the predicted capacity value of the target battery at at least one time step according to the target features of the target battery in at least one target dimension.

[0078] The usage duration prediction model is trained based on a Long Short-Term Memory (LSTM) network. The model consists of an input layer with dimensions m (number of samples, n number of features); an LSTM layer with two LSTM units, each with 64 neurons; and a fully connected layer that outputs the capacity prediction for the next k time steps.

[0079] For example, the update formula for the LSTM cell is shown in formula (1-3) below.

[0080] (1-3)

[0081] Among them, f t i t o t These are the forgetting gate, input gate, and output gate, respectively. Ct represents the cell state, and h... t This is a hidden state. For example, the formula for a fully connected layer is shown in formula (1-4) below.

[0082] (1-4)

[0083] Among them, H (L) For the last layer of feature maps, W f For fully connected weights, b f For bias.

[0084] In some embodiments, historical features can be input into the network model for training. Mean squared error (MSE) is used as the loss function, with the objective of minimizing the prediction task loss. Model pre-training is completed when the relative error between the predicted and true values ​​is less than or equal to a preset value, and the resulting model hyperparameters are used as basic model parameters. The training strategy uses the Adam optimizer, with an initial learning rate set to 0.001. A learning rate decay strategy is used, decreasing to 0.1 times the original rate every 50 epochs. The batch size is set to 64. During training, early stopping is used to prevent overfitting; training stops when the validation set loss no longer decreases for 30 consecutive epochs. Subsequently, the learned model hyperparameters are used for real-time data. The LSTM model is used as a pre-trained model, retaining historical data model parameters. By fine-tuning the parameters of the last layer of the model, it can be used as a prediction model, inputting current real-time data to output predicted values, improving the model's generalization ability and saving training time.

[0085] For example, the loss function can be shown in the following formula (1-5).

[0086] (1-5)

[0087] in, This is the actual capacity value. To predict capacity values.

[0088] S204 determines the health status of the target battery based on the predicted capacity value of the target battery at at least one time step.

[0089] In some embodiments, for each time step, the remaining usage time of the target battery at that time step is determined based on a preset battery failure time step and the difference between time steps; and for each time step, the health status index of the target battery at that time step is determined based on the ratio between the predicted capacity value of the target battery at that time step and the nominal capacity value of the target battery; the rate of decline of the health status index of the target battery at that time step is determined based on the difference between the health status index of the previous time step and the time step; and if the health status index is less than a preset index threshold, or the remaining usage time is less than a preset duration threshold, or the rate of decline of the health status index is less than a preset rate threshold, the health status of the target battery is determined to be abnormal.

[0090] For example, the difference between the preset battery failure time step and the time step is used as the remaining usage time of the target battery at the time step, and a confidence interval is introduced for evaluation using the following formula (1-6).

[0091] (1-6)

[0092] in, RUL represents the standard deviation of the prediction error. range This represents the remaining usage time.

[0093] For example, for each time step, the State of Health (SOH) index of the target battery at the time step is determined by the following formula (1-7) based on the ratio between the predicted capacity value of the target battery at the time step and the nominal capacity value of the target battery.

[0094] (1-7)

[0095] Among them, C pred To predict the capacity value, C nominal This is the nominal capacity value.

[0096] For example, battery health levels are classified according to SOH (State of Health): normal range SOH ≥ 80% (preset threshold); warning range 60% ≤ SOH < 80%; fault range SOH < 60%; edge computing nodes are deployed to collect the target battery's operating data in real time and calculate SOH and RUL (Right Limit Up). A warning mechanism is set up to trigger a warning when the SOH decrease rate exceeds a preset rate threshold (> 5%) or the RUL is less than a preset duration threshold (< 100 cycles).

[0097] The aforementioned battery health status monitoring method, apparatus, computer equipment, and storage medium acquire the current operating data of the target battery; based on the current operating data, determine the target features of the target battery in at least one target dimension; based on a trained usage duration prediction model, determine the predicted capacity value of the target battery at at least one time step according to the target features of the target battery in at least one target dimension; the usage duration prediction model is trained based on a Long Short-Term Memory network; and based on the predicted capacity value of the target battery at at least one time step, determine the health status of the target battery. In this embodiment, real-time detection of battery health status is achieved through algorithmic feature extraction and sharing, thereby more accurately determining the battery health status by using the usage duration prediction model.

[0098] Figure 3 This is a flowchart illustrating the steps for determining the battery failure time step in one embodiment. This embodiment provides a method for determining the battery failure time step, including the following steps:

[0099] S301 obtains the preset battery capacity threshold.

[0100] Among them, the battery capacity threshold is a preset multiple of the nominal capacity value (such as 0.8).

[0101] For example, the battery capacity threshold is determined by the following formula (1-8).

[0102] (1-8)

[0103] Among them, C nomial Nominal capacity value; C threshold This represents the battery capacity threshold.

[0104] S302 selects the first time step and the second time step from each time step based on the similarity between the battery capacity threshold and the predicted capacity value at different time steps.

[0105] In this case, the first time step and the second time step are adjacent, and the second time step is later than the first time step.

[0106] In some embodiments, the two time steps with the highest similarity to the battery capacity threshold are selected as candidate time steps; when the two candidate time steps are adjacent, the earlier time step is selected as the first time step, and the later time step is selected as the second time step; when the two candidate time steps are not adjacent, one of the two candidate time steps is selected as the first time step, and the next time step of the first time step is selected as the second time step; wherein, the similarity corresponding to the first time step is not less than that of the other candidate time step.

[0107] For example, in this embodiment, the intersection of the capacity prediction curve and the battery capacity threshold can be determined by linear interpolation, and the time point closest to the battery capacity threshold is selected as the first time step and the second time step.

[0108] S303 determines the battery failure time step based on the first time step, the second time step, and the battery capacity threshold.

[0109] For example, the battery failure time step is determined based on the first time step, the second time step, and the battery capacity threshold using the following formula (1-9).

[0110] (1-9)

[0111] Among them, t i and t i+1 The capacity prediction value is closest to C threshold Two time points; t EoL This refers to the point in time when the battery capacity fails.

[0112] In this embodiment, a preset battery capacity threshold is obtained; the battery capacity threshold is a preset multiple of the nominal capacity value; based on the similarity between the battery capacity threshold and the predicted capacity value at different time steps, a first time step and a second time step are selected from each time step; the first time step and the second time step are adjacent, and the second time step is later than the first time step; based on the first time step, the second time step, and the battery capacity threshold, the battery failure time step is determined. This example can more accurately and reasonably determine the battery failure time step.

[0113] Figure 4 This is a flowchart illustrating the steps for determining target features in one embodiment. This embodiment provides an optional method for determining target features, including the following steps:

[0114] S401 acquires a preset set of baseline features.

[0115] The elements in the benchmark feature set are obtained by extracting features from the benchmark operating data of the benchmark battery.

[0116] In some embodiments, preset benchmark operating data is acquired; based on a pre-trained feature extraction model, benchmark features of the target battery in at least one target dimension are determined according to the benchmark operating data; the feature extraction model is trained based on a delay-tolerant network; and a benchmark feature set is constructed based on the benchmark features in at least one target dimension. The delay-tolerant network structure includes: a feature extraction layer, a domain adaptation layer, an output layer, a fully connected network, and convolutional layers. Specifically, the feature extraction layer uses a convolutional neural network (CNN) with shared weights to extract high-order features; the domain adaptation layer incorporates a gradient inversion layer (GRL) to reduce inter-domain differences through adversarial training.

[0117] For example, the formula for a convolutional layer is as follows (1-10):

[0118] (1-10)

[0119] Among them, H (l) For the feature map of layer l, W (l) For the convolution kernel, b (l) For bias, This is the activation function.

[0120] For example, the GRL formula is as follows (1-11):

[0121] (1-11)

[0122] in, This is the gradient reversal coefficient.

[0123] For example, the expression for a fully connected layer is as follows (1-12):

[0124] (1-12)

[0125] Among them, H (L) For the last layer of feature maps, W f For fully connected weights, b f For bias.

[0126] For example, the total loss function design, as shown in formula (1-13), consists of three parts:

[0127] (1-13)

[0128] Among them, L task To predict task loss, mean squared error (MSE) is used; L domain For domain classification loss, cross-entropy loss is used; d i For the domain label (0 represents the source domain, 1 represents the target domain). L reg This is a regularization term to prevent overfitting.

[0129] For example, the loss function uses the Huber loss as shown in formula (1-14):

[0130] (1-14)

[0131] in, This is a hyperparameter with a value of 0.1.

[0132] S402 determines the feature distribution difference, Spearman correlation coefficient, and feature similarity between the current operating data and the benchmark feature set in at least one initial dimension.

[0133] In some embodiments, based on the current running data and the benchmark feature set, the feature distribution difference value, Spearman correlation coefficient, and feature similarity between the current running data and the benchmark feature set in at least one initial dimension are determined by the following formula (1-15).

[0134] (1-15)

[0135] in, For Gaussian kernel function space mapping; n s For the source domain, n t The sample size for the target domain; This represents the difference value of the characteristic distribution.

[0136] S403 determines the evaluation value of the target battery in each initial dimension based on the feature distribution difference value, Spearman correlation coefficient, and feature similarity.

[0137] In some embodiments, for each initial dimension, the evaluation value of the target battery in the initial dimension is determined by the following formula (1-16) based on the feature distribution difference value, Spearman correlation coefficient and feature similarity.

[0138] (1-16)

[0139] CSI is the evaluation value. It is the first weight corresponding to the feature distribution difference value; It is the second weight corresponding to the Spearman correlation coefficient; This is the third weight corresponding to feature similarity; Spearman's coefficient is the Spearman correlation coefficient for two-domain feature ranking; DTW is the feature sequence similarity calculated by the Dynamic Time Warping algorithm. MMD is the feature distribution difference value. Among these, .

[0140] S404 selects the initial dimension whose evaluation value is greater than the preset evaluation threshold as the target dimension, and uses the running characteristics of the current running data on the target dimension as the target features.

[0141] In some embodiments, a preset evaluation threshold can be set in this embodiment. The target feature F is selected using the following formula (1-17). shared .

[0142] (1-17)

[0143] Among them, F shared These are the selected target features; CSI(f) k ) is the evaluation value of the target battery in the k-th initial dimension; is the preset evaluation threshold; m is the initial total number of dimensions.

[0144] In some embodiments, a preset set of benchmark features is obtained; based on the current operating data and the set of benchmark features, the feature distribution difference value, Spearman correlation coefficient, and feature similarity between the current operating data and the set of benchmark features in at least one initial dimension are determined; for each initial dimension, the evaluation value of the target battery in the initial dimension is determined based on the feature distribution difference value, Spearman correlation coefficient, and feature similarity; the initial dimension with an evaluation value greater than a preset evaluation threshold is selected as the target dimension, and the operating features of the current operating data in the target dimension are taken as the target features. This embodiment can more accurately determine the target features.

[0145] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0146] Based on the same inventive concept, this application also provides a battery health status monitoring device for implementing the battery health status monitoring method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more battery health status monitoring device embodiments provided below can be found in the limitations of the battery health status monitoring method described above, and will not be repeated here.

[0147] In one exemplary embodiment, such as Figure 5 As shown, a battery health status monitoring device is provided, including: an acquisition module 10, a feature module 11, a prediction module 12, and a detection module 13, wherein:

[0148] The acquisition module 10 is used to acquire the current operating data of the target battery;

[0149] Feature module 11 is used to determine the target features of the target battery in at least one target dimension based on the current running data;

[0150] Prediction module 12 is used to determine the predicted capacity value of the target battery at at least one time step based on the target features of the target battery in at least one target dimension, according to the trained usage duration prediction model; the usage duration prediction model is trained based on a long short-term memory network.

[0151] The detection module 13 is used to determine the health status of the target battery based on the predicted capacity value of the target battery at at least one time step.

[0152] In some embodiments, the detection module 13 is further configured to, for each time step, determine the remaining usage time of the target battery at the time step based on the difference between a preset battery failure time step and the time step; and, for each time step, determine the health status index of the target battery at the time step based on the ratio between the predicted capacity value of the target battery at the time step and the nominal capacity value of the target battery; determine the rate of decline of the health status index of the target battery at the time step based on the difference between the health status index of the previous time step and the time step; and determine the health status of the target battery as abnormal if the health status index is less than a preset index threshold, or the remaining usage time is less than a preset duration threshold, or the rate of decline of the health status index is less than a preset rate threshold.

[0153] In some embodiments, the detection module 13 is further configured to obtain a preset battery capacity threshold; the battery capacity threshold is a preset multiple of the nominal capacity value; select a first time step and a second time step from each time step based on the similarity between the battery capacity threshold and the predicted capacity value at different time steps; the first time step and the second time step are adjacent, and the second time step is later than the first time step; determine the battery failure time step based on the first time step, the second time step and the battery capacity threshold.

[0154] In some embodiments, the detection module 13 is further configured to select the two time steps with the highest similarity to the battery capacity threshold as candidate time steps; when the two candidate time steps are adjacent, the earlier time step is selected as the first time step, and the later time step is selected as the second time step; when the two candidate time steps are not adjacent, one of the two candidate time steps is selected as the first time step, and the next time step of the first time step is selected as the second time step; wherein the similarity corresponding to the first time step is not less than that of the other candidate time step.

[0155] In some embodiments, the feature module 11 is further configured to obtain a preset benchmark feature set; the elements in the benchmark feature set are obtained by feature extraction from the benchmark operating data of the benchmark battery; based on the current operating data and the benchmark feature set, the feature distribution difference value, Spearman correlation coefficient, and feature similarity between the current operating data and the benchmark feature set in at least one initial dimension are determined; for each initial dimension, the evaluation value of the target battery in the initial dimension is determined based on the feature distribution difference value, Spearman correlation coefficient, and feature similarity; the initial dimension with an evaluation value greater than a preset evaluation threshold is selected as the target dimension, and the operating features of the current operating data in the target dimension are used as the target features.

[0156] In some embodiments, the feature module 11 is further configured to acquire preset benchmark operating data; determine the benchmark features of the target battery in at least one target dimension based on the benchmark operating data according to the pre-trained feature extraction model; the feature extraction model is trained based on a delay-tolerant network; and construct a benchmark feature set according to the benchmark features in at least one target dimension.

[0157] Each module in the aforementioned battery health status monitoring device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0158] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 6 As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When executed by the processor, the computer program implements a battery health status monitoring method.

[0159] Those skilled in the art will understand that Figure 6The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0160] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0161] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0162] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0163] It should be noted that the data involved in this application (including but not limited to data used for analysis, data stored, data displayed, etc.) are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0164] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0165] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0166] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for monitoring battery health status, characterized in that, The method includes: Obtain the current operating data of the target battery; Based on the current operating data, determine the target features of the target battery in at least one target dimension; Based on the trained usage duration prediction model, the predicted capacity value of the target battery at at least one time step is determined according to the target features of the target battery in at least one target dimension; the usage duration prediction model is trained based on a long short-term memory network. The health status of the target battery is determined based on the predicted capacity value of the target battery at at least one time step.

2. The method according to claim 1, characterized in that, Determining the health status of the target battery based on its predicted capacity value at at least one time step includes: For each time step, the remaining usage time of the target battery at that time step is determined based on the difference between a preset battery failure time step and the time step itself; and, For each time step, the health status index of the target battery at that time step is determined based on the ratio between the predicted capacity value of the target battery at that time step and the nominal capacity value of the target battery. Based on the difference between the health status index of the previous time step and the current time step, the rate of decline of the health status index of the target battery at the current time step is determined. If the health status indicator is less than a preset indicator threshold, or the remaining usage time is less than a preset duration threshold, or the rate of decline of the health status indicator is less than a preset rate threshold, the health status of the target battery is determined to be abnormal.

3. The method according to claim 2, characterized in that, The method for determining the battery failure time step includes: Obtain a preset battery capacity threshold; the battery capacity threshold is a preset multiple of the nominal capacity value; Based on the similarity between the battery capacity threshold and the predicted capacity value at different time steps, a first time step and a second time step are selected from each of the time steps; the first time step and the second time step are adjacent, and the second time step is later than the first time step; The battery failure time step is determined based on the first time step, the second time step, and the battery capacity threshold.

4. The method according to claim 3, characterized in that, The step of selecting a first time step and a second time step from each time step based on the similarity between the battery capacity threshold and the predicted capacity value at different time steps includes: The two time steps with the highest similarity to the battery capacity threshold are selected as candidate time steps; When two candidate time steps are adjacent, the earlier time step is taken as the first time step, and the later time step is taken as the second time step. If two candidate time steps are not adjacent, one of the two candidate time steps is selected as the first time step, and the next time step of the first time step is taken as the second time step; wherein, the similarity corresponding to the first time step is not less than that of the other candidate time step.

5. The method according to claim 1, characterized in that, The step of determining the target features of the target battery in at least one target dimension based on the current operating data includes: Obtain a preset set of benchmark features; the elements in the set of benchmark features are obtained by extracting features from the benchmark operating data of the benchmark battery. Based on the current running data and the benchmark feature set, determine the feature distribution difference value, Spearman correlation coefficient, and feature similarity between the current running data and the benchmark feature set in at least one initial dimension; For each initial dimension, the evaluation value of the target battery in that initial dimension is determined based on the feature distribution difference value, Spearman correlation coefficient, and feature similarity. An initial dimension with an evaluation value greater than a preset evaluation threshold is selected as the target dimension, and the operational characteristics of the current operational data on the target dimension are taken as the target features.

6. The method according to claim 5, characterized in that, The acquisition of the preset benchmark feature set includes: Obtain preset baseline operating data; Based on a pre-trained feature extraction model, and according to the benchmark running data, the baseline features of the target battery in at least one target dimension are determined; the feature extraction model is trained based on a delay-tolerant network. Construct a set of benchmark features based on benchmark features in at least one target dimension.

7. A battery health status monitoring device, characterized in that, The device includes: The acquisition module is used to acquire the current operating data of the target battery; The feature module is used to determine the target features of the target battery in at least one target dimension based on the current operating data. The prediction module is used to determine the predicted capacity value of the target battery at at least one time step based on the target features of the target battery in at least one target dimension, according to a trained usage duration prediction model; the usage duration prediction model is trained based on a long short-term memory network. The detection module is used to determine the health status of the target battery based on the predicted capacity value of the target battery at at least one time step.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images