Double-circuit power supply method of battery management unit and electronic equipment

Abstract

The invention provides a double-circuit power supply method of a battery management unit and electronic equipment, which are applied to an energy storage system, and the system comprises a battery unit, a charging unit, the battery management unit, a double-power supply unit (a first power supply unit is connected with an external power supply, and a second power supply unit is connected with the battery unit), a voltage reduction unit and a switch unit. The core of the method comprises the steps that a battery management unit collects battery working information (temperature, voltage, current and attenuation rate) and uploads the information to a server to obtain predicted safe voltage; the second power supply unit is started to supply power when no external power supply exists; when an external power supply is connected, switching to the first power supply unit and shielding the second power supply unit; and when the external power supply is connected and the battery voltage is lower than the predicted safety voltage, the charging unit is started to charge the battery, the first power supply unit is turned off, and the second power supply unit is started to maintain BMS operation and execute shielding. And continuous power supply of the battery management unit and safe charging of the battery are ensured through intelligent switching of two paths of power supplies and dynamic judgment of a voltage threshold.

Description

Technical Field

[0001] The present application relates to the technical field of power supply, and particularly to a dual-power supply method for a battery management unit and an electronic device. Background Art

[0002] Currently, the power supply (12V or 24V) of a conventional energy storage battery BMS system is composed of a power supply converted from a battery pack (12V or 24V) + an auxiliary power supply (12V or 24V) inside the charger to form a two-way power supply. Under normal circumstances, the mains power is online for a long time for energy storage batteries. This power supply method has the following defects. The battery pack and the charger both provide 24V power without optimization. In this way, each bears about 50% of the power supply. The battery pack will be consumed by the BMS system and enter frequent charging, which is not conducive to extending the service life. Since the auxiliary power supply inside the charger has to supply power to the BMS for a long time and cannot be disconnected, the charger is in a long-term working state, reducing the service life of the charger. Summary of the Invention

[0003] The present application provides a dual-power supply method for a battery management unit and an electronic device, which is used to optimize the power supply of the BMS system.

[0004] In a first aspect, an embodiment of the present application provides a dual-power supply method for a battery management unit, which is applied to an energy storage system. The energy storage system includes: a battery unit, a charging unit, a battery management unit, a first power supply unit, a second power supply unit, a buck unit, and a switch unit. The first power supply unit and the second power supply unit are used to provide electrical energy for the battery management unit. The charging unit is respectively connected to an external power supply and the battery unit. The first power supply unit is connected to the external power supply. The second power supply unit is connected to the battery unit. The battery management unit controls the entire energy storage system through the switch unit. The method includes: The battery management unit collects the working information of the battery unit and sends the working information to the server to obtain the predicted safe voltage returned by the server. The predicted safe voltage is determined based on the working information. When the external power supply is not connected, start the second power supply unit to provide electrical energy for the battery management unit. When the external power supply is connected, start the first power supply unit to provide electrical energy for the battery management unit and shield the second power supply unit. When the external power supply is connected and the output voltage of the battery unit is less than the predicted safe voltage, start the charging unit to supply power to the battery unit, turn off the first power supply unit, start the second power supply unit to provide power for the battery management unit, and shield the second power supply unit.

[0005] In a second aspect, an embodiment of the present application provides an electronic device, which includes an energy storage system. The energy storage system includes: a battery unit, a charging unit, a battery management unit, a first power supply unit, a second power supply unit, a buck unit, and a switch unit. The first power supply unit and the second power supply unit are configured to supply power to the battery management unit. The charging unit is respectively connected to an external power supply and the battery unit. The first power supply unit is connected to the external power supply, and the second power supply unit is connected to the battery unit. The electronic device is configured to execute the dual-power supply method of the battery management unit as described in any one of the embodiments of the present application.

[0006] In a third aspect, an embodiment of the present application provides a computer-readable storage medium storing a computer program, which when executed by a processor causes the processor to implement the dual-power supply method of the battery management unit as described in any one of the embodiments of the present application.

[0007] An embodiment of the present application provides a dual-power supply method for a battery management unit, which is applied to an energy storage system. The energy storage system includes: a battery unit, a charging unit, a battery management unit, a first power supply unit, a second power supply unit, a buck unit, and a switch unit. The first power supply unit and the second power supply unit are configured to supply power to the battery management unit. The charging unit is respectively connected to an external power supply and the battery unit. The first power supply unit is connected to the external power supply, and the second power supply unit is connected to the battery unit. The battery management unit controls the entire energy storage system through the switch unit. The method includes: The battery management unit collects the working information of the battery unit and sends the working information to a server to obtain a predicted safety voltage returned by the server, where the predicted safety voltage is determined based on the working information; when the external power supply is not connected, the second power supply unit is started to supply power to the battery management unit; when the external power supply is connected, the first power supply unit is started to supply power to the battery management unit, and the second power supply unit is shielded; when the external power supply is connected and the output voltage of the battery unit is less than the predicted safety voltage, the charging unit is started to supply power to the battery unit, the first power supply unit is turned off, the second power supply unit is started to supply power to the battery management unit, and the second power supply unit is shielded. Through the above method, after determining the predicted safety voltage, the first power supply unit or the second power supply unit is respectively controlled to supply power to the battery management unit under different conditions to adapt to the optimal power supply method in the current situation, reduce the voltage loss of the battery pack, and extend the overall life of the system. Description of the Drawings

[0008] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0009] Figure 1 A schematic block diagram of an energy storage system provided by an embodiment of the present application; Figure 2 A schematic flowchart of a dual - power supply method for a battery management unit provided by an embodiment of the present application; Figure 3 A circuit schematic diagram of the energy storage system provided by an embodiment of the present application. Detailed implementation manners

[0010] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0011] The flowcharts shown in the drawings are only illustrative examples, and do not necessarily include all contents and operations / steps, nor do they necessarily need to be executed in the described order. For example, some operations / steps can also be decomposed, combined, or partially merged. Therefore, the actual execution order may change according to the actual situation.

[0012] It should also be understood that the terms used in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification of the present application and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" are intended to include the plural forms.

[0013] It should be further understood that the term "and / or" used in the specification of the present application and the appended claims refers to any combination and all possible combinations of one or more of the related listed items, and includes these combinations.

[0014] Please refer to Figure 1 , Figure 1 is a schematic block diagram of an energy storage system provided by an embodiment of the present application. As Figure 1As shown, the energy storage system 100 includes: a battery unit 11, a charging unit 12, a battery management unit 13, a first power supply unit 14, a second power supply unit 15, a buck unit 16, and a switch unit. The external power supply 200 and the energy storage system 100 are connected through an air switch Km. The first power supply unit 14 and the second power supply unit 15 are used to supply electrical energy to the battery management unit 13. The charging unit 12 is respectively connected to the external power supply 200 and the battery unit 11. The first power supply unit 14 is connected to the external power supply 200, and the second power supply unit 15 is connected to the battery unit 11. The battery management unit 13 controls the entire energy storage system 100 through the switch unit. It should be noted that the switch unit is composed of multiple switches, which are not shown in the figure and are arranged between other units as needed. The battery management unit 13 controls the energy storage system 100 through the switch unit. For example, it controls the connection between the charging unit 12 and the external power supply 200, the connection between the external power supply 200 and the first power supply unit 14, and the connection between the battery unit 11 and the load 300.

[0015] Please refer to Figure 2 , Figure 2 is a schematic flowchart of a dual-power supply method for a battery management unit provided by an embodiment of the present application. As Figure 2 shown, the specific steps of the dual-power supply method for the battery management unit include: S101 - S104.

[0016] S101. The battery management unit 13 collects the working information of the battery unit 11 and sends the working information to the server to obtain the predicted safety voltage returned by the server. The predicted safety voltage is determined based on the working information.

[0017] Exemplarily, the battery management unit 13 synchronously collects real-time data of the battery temperature, discharge voltage, and discharge current through a temperature sensor (such as an NTC thermistor) integrated inside the battery unit 11, a voltage sampling circuit (using a high-precision ADC chip), and a current Hall sensor with a sampling period of 10 ms. Calculate the power attenuation rate based on the Coulomb counting method, calculate the capacity attenuation value of the current cycle through current integration, and combine the historical cycle times of the battery unit 11 (stored in the EEPROM of the battery management unit 13). The specific calculation formula is: attenuation rate = current capacity initial capacity × 100% attenuation rate = initial capacity current capacity × 100% Calculate the power attenuation rate. For the collected battery temperature, discharge voltage, discharge current, and power attenuation rate data, eliminate the transmission delay between sensors through a timestamp alignment algorithm (based on the PTP clock synchronization protocol) to generate a multi-dimensional battery working data set with a unified time dimension.

[0018] The battery management unit 13 uploads the multi-dimensional battery operating data set to the cloud server via the CAN bus or the Bluetooth 5.0 protocol. The data packet encapsulation format includes a header identifier (identifying the ID of the battery unit 11), a data segment (including temperature, voltage, current, attenuation rate, and timestamp), and a CRC check code.

[0019] After the server receives the data, it performs the following operations: operating condition state classification, feature extraction and fusion, and safe voltage prediction.

[0020] Operating condition state classification: Based on the magnitude of the discharge current, the data is divided into sub-data sets corresponding to different discharge rates (such as 0.5C, 1C, 2C), and the voltage fluctuation curve and temperature drift curve at each rate are extracted.

[0021] Feature extraction and fusion: Perform kernel principal component analysis (KPCA) on the voltage fluctuation curve, and extract the first 3 principal component components as the voltage time-domain feature set; perform wavelet packet decomposition on the temperature drift curve (selecting the db4 wavelet basis) to generate the energy distribution of 8 sub-bands as the temperature frequency-domain feature set; cross-domain fuse the two types of feature sets through the tensor folding algorithm to generate a composite feature matrix.

[0022] Safe voltage prediction: Input the composite feature matrix into a pre-trained voltage safety prediction model (based on the LSTM-Transformer hybrid architecture), output the predicted safe voltage value through multi-dimensional regression analysis, and return the result to the battery management unit 13 via the MQTT protocol.

[0023] After the battery management unit 13 receives the predicted safe voltage returned by the server, it compares it with the historical safe voltage threshold stored locally. If the deviation exceeds 5%, it triggers the data retransmission mechanism and re-uploads the current working information to the server for secondary calculation.

[0024] Among them, the server can be an independent server, a server cluster, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms.

[0025] S102. When no external power supply is connected, start the second power supply unit 15 to provide electrical energy for the battery management unit 13.

[0026] Exemplarily, the second power supply unit 15 is used to transmit the electrical energy of the battery unit 11 to the battery management unit 13, and the transmission loss of the second power supply unit 15 is extremely low and will not cause waste of electrical energy.

[0027] During the power supply of the second power supply unit 15, the battery management unit 13 switches to the low power consumption mode, shuts down unnecessary peripherals (such as the Bluetooth module), and only keeps the core control circuit (MCU and voltage / temperature sampling module) running, with the power consumption reduced to less than 10 mA.

[0028] S103. When an external power supply is connected, start the first power supply unit 14 to supply power to the battery management unit 13 and shield the second power supply unit 15.

[0029] Exemplarily, the first power supply unit 14 can use the external power supply to supply power to the battery management unit 13. When the external power supply is connected and the battery unit 11 is not being charged, supplying power to the battery management unit 13 through the first power supply unit 14 can reduce the loss of the battery unit 11. After the first power supply unit 14 is started, the output voltage of the first power supply unit 14 can shield the output voltage of the second power supply unit 15, which is equivalent to shutting down the second power supply unit 15. During this period, the second power supply unit 15 will not consume the electrical energy of the battery unit 11.

[0030] S104. When the external power supply is connected and the output voltage of the battery unit 11 is less than the predicted safe voltage, start the charging unit 12 to supply power to the battery unit 11, shut down the first power supply unit 14, start the second power supply unit 15 to supply power to the battery management unit 13, and shield the second power supply unit 15.

[0031] Exemplarily, during the charging process of the battery unit 11, the second power supply unit 15 can use the output voltage of the charging unit 12 to supply power to the battery management unit 13. Since the loss of the second power supply unit 15 is less than the loss of the first power supply unit 14, thus, shut down the first power supply unit 14 and use the second power supply unit 15 to supply power to the battery management unit 13.

[0032] An embodiment of the present application provides a dual-power supply method for a battery management unit, which is applied to an energy storage system. The energy storage system includes: a battery unit, a charging unit, a battery management unit, a first power supply unit, a second power supply unit, a buck unit, and a switch unit. The first power supply unit and the second power supply unit are used to supply power to the battery management unit. The charging unit is respectively connected to an external power supply and the battery unit. The first power supply unit is connected to the external power supply, and the second power supply unit is connected to the battery unit. The battery management unit controls the entire energy storage system through the switch unit. The method includes: The battery management unit collects the working information of the battery unit and sends the working information to the server to obtain the predicted safe voltage returned by the server. The predicted safe voltage is determined based on the working information. When the external power supply is not connected, the second power supply unit is started to supply power to the battery management unit. When the external power supply is connected, the first power supply unit is started to supply power to the battery management unit, and the second power supply unit is blocked. When the external power supply is connected and the output voltage of the battery unit is less than the predicted safe voltage, the charging unit is started to supply power to the battery unit, the first power supply unit is turned off, the second power supply unit is started to supply power to the battery management unit, and the second power supply unit is blocked. Through the above method, after determining the predicted safe voltage, the first power supply unit or the second power supply unit is respectively controlled to supply power to the battery management unit under different conditions according to the predicted safe voltage to adapt to the optimal power supply method in the current situation, reduce the voltage loss of the battery pack, and extend the overall life of the system.

[0033] To more clearly introduce the technical solution of the present application, the technical solution of the present application will also be introduced through specific embodiments below. It should be noted that the specific embodiment is used to expand the description of the technical solution of the present application, rather than limiting the present application.

[0034] In some embodiments, the working information includes: battery temperature, power attenuation rate, discharge voltage, and discharge current. When the server is used to execute the determination of the predicted safe voltage of the battery unit 11 based on the working information, it is specifically used to execute: S201-S206.

[0035] S201. Align the battery temperature, power attenuation rate, discharge voltage, and discharge current in time series to generate a multi-dimensional battery working data set, and classify the working conditions of the multi-dimensional battery working data set to obtain a first voltage fluctuation data set and a temperature drift data set under different discharge rates.

[0036] Exemplarily, the timing alignment is achieved through a high-precision clock synchronization protocol (such as the IEEE 1588 PTP protocol), ensuring that the data from the temperature sensor, voltage sampling circuit, current Hall sensor, and capacity calculation module have a unified timestamp. The preprocessing module built in the battery management unit 13 performs interpolation compensation on the original data. For the transmission delay between sensors (such as a 50 ms response lag of the temperature sensor), the reverse time window alignment algorithm is adopted. Taking the current signal as a reference, the temperature and voltage data are remapped to the same time axis according to the time offset. The generated multi-dimensional battery operating data set is stored in the form of a time series, including four-dimensional data channels (temperature, attenuation rate, voltage, current) and nanosecond-level timestamps. The classification of the operating conditions is based on the numerical range of the discharge current. For example, 0 - 0.5C is defined as the low-rate operating condition, 0.5C - 1.5C is the normal operating condition, and above 1.5C is the high-rate operating condition. Each sub-condition data set independently stores the corresponding voltage fluctuation curve (continuous sampling values of voltage over time) and temperature drift curve (variation trajectory of temperature with discharge time).

[0037] The battery management unit 13 collected 2000 groups of data points during a 10-minute discharge process. During the timing alignment stage, the data of the temperature sensor is delay-compensated to be synchronized with the current signal through the PTP protocol, forming a unified data set with a timestamp interval of 300 ms. The operating condition classification module detected that the peak value of the discharge current reached 2.1C (corresponding current value of 4.2A), automatically classified the data during this period into the high-rate operating condition subset, and extracted the continuous record of the voltage fluctuating from 3.8V to 3.5V within this interval as the first voltage fluctuation data set. At the same time, the variation trajectory of the temperature rising from 25°C to 38°C was recorded as the temperature drift data set. The classified data is stored in CSV format, and each file header is marked with the discharge rate category and time range for subsequent calls by the feature extraction module.

[0038] S202. Perform kernel principal component analysis on the first voltage fluctuation data set to extract the voltage time-domain feature set, and perform wavelet packet decomposition on the temperature drift data set to generate the temperature frequency-domain feature set.

[0039] Exemplarily, Kernel Principal Component Analysis (KPCA) uses a Gaussian kernel function to perform non-linear dimensionality reduction on the time-domain features of the voltage fluctuation dataset. The kernel parameter σ is set to the standard deviation of the voltage fluctuation amplitude (e.g., 0.2V). By calculating the eigenvectors of the kernel matrix, the first 3 principal component components (contribution rate > 85%) are extracted as the voltage time-domain feature set, which characterizes the trend, fluctuation amplitude, and mutation frequency of voltage changes. For the temperature drift dataset, the db4 wavelet basis function is selected for wavelet packet decomposition, and the decomposition level is 4 layers, generating 16 sub-bands (frequency range 0 - 0.5Hz). Calculate the energy proportion of each sub-band (energy value = sum of squared wavelet coefficients), and select the top 4 frequency bands with the highest energy proportion (e.g., 0.1 - 0.2Hz, 0.3 - 0.4Hz) as the temperature frequency-domain feature set, which reflects the periodicity and abnormal fluctuation patterns of temperature changes.

[0040] For the voltage fluctuation dataset (1000 sampling points) under high-rate operating conditions, after the KPCA module calculates the Gaussian kernel matrix, the first principal component component (characterizing the overall voltage drop slope), the second principal component component (reflecting the number of voltage dips), and the third principal component component (describing the local fluctuation amplitude) are extracted to form a voltage time-domain feature set containing 3D features. After the temperature drift dataset is decomposed by wavelet packet, it is found that the energy proportion of the 0.15Hz frequency band reaches 42%, indicating that there is a significant periodic temperature rise in this frequency band (possibly related to the chemical reaction cycle inside the battery cell), while the energy of the 0.35Hz frequency band is abnormally high (proportion 28%), suggesting the existence of occasional heat dissipation abnormal events. These frequency-domain features and voltage time-domain features together form the basis for subsequent cross-domain fusion.

[0041] S203. Perform cross-domain feature fusion on the voltage time-domain feature set and the temperature frequency-domain feature set to generate a composite feature matrix.

[0042] Exemplarily, the cross-domain fusion is divided into two stages: time axis alignment and feature splicing. First, the voltage time-domain features (time granularity 300ms) and the temperature frequency-domain features (time granularity 1s) are non-linearly aligned through the Dynamic Time Warping (DTW) algorithm to eliminate the timing misalignment caused by the difference in sensor sampling frequencies. The aligned feature sequences are segmented according to a time window (e.g., 30-second window). Inside each window, the voltage time-domain features (3D) and the temperature frequency-domain features (4D) are spliced through tensors to form a 7D composite feature vector, which is arranged in chronological order as an N×7 composite feature matrix (N is the number of time windows). The rows of the feature matrix represent time slices, and the columns contain cross-domain parameters such as voltage trend, fluctuation, mutation, and temperature period, and abnormal frequency band energy.

[0043] During the battery discharge process, the voltage time-domain feature set contains 3D data (time length of 10 minutes, 200 sampling points in total), and the temperature frequency-domain feature set contains 4D data (time length of 10 minutes, 60 sampling points in total). After aligning the two to a unified time axis through the DTW algorithm, 20 time slices are generated by slicing with a 30-second window. Within each slice, the voltage features (such as [-0.3, 0.1, 0.05]) and the temperature frequency-domain features (such as [0.42, 0.28, 0.15, 0.05]) are concatenated into a 7D vector ([-0.3, 0.1, 0.05, 0.42, 0.28, 0.15, 0.05]), forming a 20×7 composite feature matrix. This matrix can be directly input into a machine learning model to associate the relationship between voltage anomalies and temperature frequency-domain changes.

[0044] S204. Perform incremental feature dimensionality expansion on the composite feature matrix to obtain an enhanced security feature set.

[0045] Exemplarily, the incremental feature dimensionality expansion adopts a strategy that combines an autoencoder (Autoencoder) and the superposition of statistical features. First, use a 3-layer fully connected autoencoder (the dimension of the encoder hidden layer is 5, and the dimension of the decoder hidden layer is 7) to perform unsupervised learning on the composite feature matrix, and extract the latent features (5D) as supplementary dimensions. Secondly, calculate the statistics of the original composite feature matrix according to the time window: mean (reflecting the trend), standard deviation (fluctuation intensity), skewness (distribution asymmetry), and kurtosis (mutation concentration degree), generating 4D statistical features. Finally, concatenate the original 7D features, 5D latent features, and 4D statistical features into a 16D enhanced security feature set, significantly improving the feature expression ability.

[0046] For the 20×7 composite feature matrix obtained in S203, the autoencoder encodes each 7D vector to generate 5D latent features (such as [0.2, -0.1, 0.4, 0.05, -0.3]), and at the same time calculates the statistics such as the mean (such as [-0.2, 0.05, 0.1, 0.3, 0.2, 0.1, 0.05]) and standard deviation (such as [0.15, 0.08, 0.06, 0.12, 0.09, 0.04, 0.03]) of the 7D features within each time window. The feature vector corresponding to each time slice is expanded to 16D (7 + 5 + 4), and the dimension of the enhanced security feature set is increased to 20×16, covering triple information of the original data, abstract representation, and statistical law.

[0047] S205. Calculate the coupling coefficients of the power decay rate and the discharge current in the enhanced security feature set respectively, and perform dynamic weight allocation on the enhanced security feature set based on the coupling coefficients to generate a security assessment feature vector.

[0048] Exemplarily, the coupling coefficient is calculated by the Pearson correlation coefficient to quantify the linear correlation degree between the power attenuation rate and the discharge current. The specific process is as follows: Extract the attenuation rate sequence (such as the capacity retention rate of 100 cycles) and the corresponding average discharge current sequence from the historical data in the enhanced security feature set, and calculate the correlation coefficient r (range -1 to 1) between the two. The dynamic weight assignment adopts a piecewise function strategy: If |r| > 0.7, it is considered a strong coupling, and the weight of the attenuation rate-related features is increased by 30%; if 0.3 < |r| ≤ 0.7, the weight remains unchanged; if |r| ≤ 0.3, the weight is reduced by 20%. The adjusted feature vector is processed by L2 normalization to ensure that the sum of the weights of each dimension is 1, forming a security assessment feature vector.

[0049] In the historical data of a battery cell 11, the correlation coefficient r between the power attenuation rate and the discharge current is -0.65 (strong negative correlation, high current accelerates capacity decay), then the weight of the attenuation rate-related features (such as the capacity retention rate statistic) is increased from the initial 0.1 to 0.13, and the weight of the discharge current-related features (such as the average current) is decreased from 0.15 to 0.12. After dynamic adjustment, the dimension reflecting the attenuation-current coupling relationship in the 16-dimensional features is strengthened, and the generated security assessment feature vector focuses more on high-risk parameter combinations.

[0050] S206. Input the security assessment feature vector into a pre-trained voltage security prediction model for multi-dimensional regression analysis to obtain the predicted security voltage of the battery cell 11, where the voltage security prediction model is trained with the abnormal working data of the abnormal charging unit 12.

[0051] Exemplarily, the voltage security prediction model adopts an LSTM-Transformer hybrid architecture. The LSTM layer (128 hidden units) captures the temporal dependence, and the Transformer encoder (4-head attention mechanism) extracts the cross-time step feature correlation. The model training data comes from historical abnormal charging records, including working data sets in scenarios such as overvoltage charging (voltage > 4.3V) and undervoltage discharging (voltage < 3.0V). After being encoded by the LSTM, the input security assessment feature vector focuses on the key time slices through the attention weight matrix, and the predicted security voltage value is output by the fully connected regression layer. The model training uses the Adam optimizer, with a learning rate of 0.001, a loss function of smooth L1 loss, and 500 training epochs. Early stopping is used to prevent overfitting.

[0052] After inputting the safety evaluation feature vector (20×16 dimensions) during the discharge process into the model, the LSTM layer extracts the hidden states of each time slice, and the Transformer calculates the attention weights between different time slices (for example, the weight ratio of the 5th - 10th minute slices reaches 60%). The regression layer outputs the predicted safety voltage of 3.72V. This value will be compared with the real-time monitored voltage. If the battery voltage is lower than 3.72V and an external power source is connected, the charging protection strategy will be triggered. The MAE (Mean Absolute Error) of the model on the test set is 0.03V, meeting the engineering accuracy requirements.

[0053] In some embodiments, cross-domain feature fusion is performed on the voltage time-domain feature set and the temperature frequency-domain feature set to generate a composite feature matrix, including: S2031 - S2035.

[0054] S2031. Align the voltage time-domain feature set and the temperature frequency-domain feature set in the time dimension to generate a synchronized time-frequency feature sequence.

[0055] Exemplarily, the time dimension alignment is achieved through the Dynamic Time Warping (DTW) algorithm to solve the time scale difference problem between the voltage time-domain features (high sampling rate) and the temperature frequency-domain features (low sampling rate). Specifically, the voltage time-domain feature set is recorded with millisecond-level timestamps, while the temperature frequency-domain feature set has a time granularity of seconds due to the windowing process of wavelet packet decomposition. The DTW algorithm constructs a cost matrix to find the optimal non-linear matching path between the two feature sequences, stretching or compressing the time axis to eliminate the time delay deviation. During the alignment process, interpolation fitting is performed on the redundant sampling points of the voltage time-domain features, and cubic spline interpolation is used to fill the sparse points of the temperature frequency-domain features to generate a synchronized time-frequency feature sequence with strictly synchronized timestamps. In the synchronized time-frequency feature sequence, each time node contains both the voltage time-domain feature vector (such as the trend component, fluctuation amplitude) and the temperature frequency-domain feature vector (such as the energy value of a specific frequency band), forming multi-dimensional time series data, providing a basis for cross-domain fusion in terms of temporal consistency.

[0056] S2032. Perform kernel matrix decomposition on the voltage time-domain features in the synchronized time-frequency feature sequence to extract orthogonalized voltage principal components, and perform band energy normalization on the temperature frequency-domain features in the synchronized time-frequency feature sequence to generate a reduced-dimensional temperature frequency band energy distribution set.

[0057] Exemplarily, kernel matrix decomposition of voltage time-domain features uses kernel principal component analysis (KPCA). The Gaussian kernel function is selected to map to a high-dimensional space, and orthogonal principal component components are extracted through eigenvalue decomposition. During the KPCA process, the kernel function parameters are adaptively adjusted according to the statistical distribution of the voltage fluctuation amplitude to ensure effective separation of non-linear features. The extracted principal component components are sorted according to the contribution rate, and the first N components with a cumulative contribution rate exceeding 90% are retained to eliminate redundant noise and reduce the data dimension. The band energy normalization process of temperature frequency-domain features standardizes the energy values of each sub-band: first, calculate the total energy of each band within the time window, and then divide the energy value of each band by the total to obtain the normalized energy ratio. After normalization, high-frequency noise bands (such as bands with an energy ratio lower than 1%) are automatically filtered out, and the energy distribution of the main bands (such as energy ratio > 5%) is retained to form a reduced-dimension temperature band energy distribution set. This distribution set represents the frequency-domain core mode of temperature changes in vector form, retaining key information while reducing data complexity.

[0058] S2033. Construct a three-dimensional tensor of the orthogonalized voltage principal component components and the reduced-dimension temperature band energy distribution set along the time axis, and fuse the time-domain - frequency-domain - energy dimension features through the tensor folding algorithm to generate an initial composite feature tensor.

[0059] Exemplarily, the three-dimensional tensor construction takes the time axis as the first dimension, the voltage principal component components as the second dimension (time-domain features), and the temperature band energy as the third dimension (frequency-domain features). The voltage principal component vector (such as 3D) corresponding to each time slice and the temperature band energy vector (such as 4D) generate a 3×4 two-dimensional matrix through the outer product operation, representing the joint distribution of the time-domain - frequency-domain features at that moment. Stack the two-dimensional matrices at all moments along the time axis to form a three-dimensional tensor structure of time × time-domain × frequency-domain. The tensor folding algorithm unfolds the three-dimensional tensor into a two-dimensional matrix according to the mode. For example, the time-domain and frequency-domain dimensions are merged into the feature channel dimension to generate a two-dimensional representation of time × (time-domain × frequency-domain), and then low-rank approximation is performed through singular value decomposition (SVD) to reconstruct an initial composite feature tensor that retains the main features. The initial composite feature tensor fuses the time-domain fluctuation mode and the frequency-domain energy distribution through this process to establish cross-dimensional correlations.

[0060] S2034. Perform dynamic channel weight assignment driven by the self-attention mechanism on the initial composite feature tensor to obtain the cross-domain correlation degree.

[0061] Exemplarily, the self-attention mechanism dynamically allocates the importance of each channel by calculating the similarity weights between feature channels. Specifically, each feature channel (time-domain - frequency-domain combination) of the initial composite feature tensor is used as the query vector, key vector, and value vector. The dot product similarity between the query and the key is calculated, and the attention weight matrix is generated through Softmax normalization. This weight matrix reflects the cross-domain correlation strength between different time-domain features (such as voltage fluctuation trends) and frequency-domain features (such as temperature periodic performance energy). To further enhance the dynamicity, a learnable scaling factor is introduced to adjust the distribution range of the attention weights, avoiding the problem of gradient disappearance. The cross-domain correlation degree is output in the form of a weight matrix, and the matrix element value represents the contribution degree of the corresponding feature channel to the safe voltage prediction. The larger the value, the stronger the correlation.

[0062] S2035. Screen the effective feature channels according to the preset correlation degree threshold and the cross-domain correlation degree, and recombine them to generate a composite feature matrix.

[0063] Exemplarily, the preset correlation degree threshold is set based on historical data statistics. For example, the top 20% quantile of the attention weight matrix is taken as the threshold. During the screening process, all the weight matrix elements are traversed, and the feature channels with a correlation degree exceeding the threshold are retained, and the remaining channels are set to zero to suppress noise interference. The effective feature channels are recombined according to the original three-dimensional tensor structure. The non-zero channel data is extracted by slicing along the time axis and rearranged into a two-dimensional matrix form (time × number of effective features). In the recombined composite feature matrix, each row corresponds to a time slice, and each column is a high-correlation feature after cross-domain fusion, such as "the joint feature of voltage principal component 1 and temperature frequency band energy 3". This matrix eliminates redundant information and focuses on the key cross-domain correlation features, providing optimized input data for subsequent safe voltage prediction.

[0064] In some embodiments, the safety assessment feature vector is input into a pre-trained voltage safety prediction model for multi-dimensional regression analysis to obtain the predicted safe voltage of battery cell 11, including: S2061 - S2067.

[0065] S2061. Perform Gaussian kernel function mapping on the safety assessment feature vector to generate a high-dimensional non-linear feature space.

[0066] Exemplarily, the Gaussian kernel function mapping maps the low-dimensional security assessment feature vector to a high-dimensional space through a non-linear transformation to capture the complex correlations between features. Specifically, the standard deviation of each dimension in the security assessment feature vector is used as the kernel parameter to calculate the similarity weights of any two feature vectors in the high-dimensional space, generating a kernel matrix. This matrix implicitly maps the original features (such as voltage decay rate, temperature band energy) to points in the high-dimensional space, making the linearly inseparable feature patterns (such as the correlation between voltage sags and high-frequency temperature fluctuations) linearly separable in the high-dimensional space. The dimension of the generated high-dimensional non-linear feature space is usually several times that of the original dimension (such as expanding from 16 dimensions to 256 dimensions), significantly enhancing the expression ability of subsequent models while retaining the topological structure of the original data.

[0067] S2062. Perform time series slicing on the high-dimensional non-linear feature space to obtain a multi-scale feature tensor sequence.

[0068] Exemplarily, the time series slicing adopts a sliding time window algorithm to segment the continuous high-dimensional feature space according to a preset window length (such as 30 seconds) and step size (such as 10 seconds). The window length setting needs to cover the typical charge and discharge cycle (such as the relaxation effect period of lithium batteries), and the step size ensures that there is partial overlap between adjacent windows to avoid information loss. The feature points within each window are arranged in a two-dimensional matrix (time step × high-dimensional features) in chronological order, and then a multi-scale feature tensor sequence is generated through pyramid multi-scale sampling (such as using 15-second, 30-second, and 60-second windows simultaneously). This sequence contains the feature distributions at different time granularities, which can not only capture the short-term fluctuation details but also reflect the long-term trend changes, providing multi-level inputs for time series modeling.

[0069] S2063. Extract the voltage decay features in the historical charge and discharge mode from the multi-scale feature tensor sequence through a bidirectional gated recurrent network, and use a multi-head attention mechanism to calculate the correlation weights of different time steps in the voltage decay features, generating a dynamically attention-enhanced feature set.

[0070] Exemplarily, the bidirectional gated recurrent network (BiGRU) processes the multi-scale feature tensor sequence with two GRU units in the forward and backward directions respectively to capture the temporal dependence relationship of the voltage decay features (such as the impact of the discharge mode in the previous hour on the current voltage). The GRU unit dynamically selects memory information through an update gate and a reset gate, and outputs the hidden state at each time step. The multi-head attention mechanism divides the hidden state sequence output by the BiGRU into multiple sub-spaces (such as 4 heads), calculates the attention weights between different time steps respectively (such as the correlation strength between the t-th step and the t-10 step), and generates a dynamically attention-enhanced feature set after weighted fusion. This feature set strengthens the influence of key time nodes (such as voltage sag periods) and suppresses noise interference.

[0071] S2064. Input the dynamically attention-enhanced feature set into the residual convolution module, and output multi-level abstract feature maps through the layer-by-layer residual connections in the residual convolution module.

[0072] Exemplarily, the residual convolution module is composed of a cascade of multiple residual blocks, and each residual block includes a convolutional layer, a batch normalization layer, and a skip connection. The convolutional layer uses convolutional kernels of different sizes (such as 3×3, 5×5) to extract local features (such as voltage fluctuation segments), the batch normalization layer stabilizes the training process, and the skip connection adds the input features to the convolutional output to avoid gradient vanishing. The dynamically attention-enhanced feature set is processed layer by layer. The shallow layer captures detailed features (such as tiny voltage fluctuations), and the deep layer extracts abstract patterns (such as periodic decay trends). The output multi-level abstract feature maps fuse semantic information at different levels, providing multi-granularity representations for abnormal pattern matching.

[0073] S2065. Based on a preset abnormal voltage pattern library, perform abnormal pattern matching on the multi-level abstract feature maps, calculate the feature similarity scores between the multi-level abstract feature maps and the preset patterns in the abnormal voltage pattern library, and dynamically adjust the weight allocation ratio of the voltage decay coefficient in the pre-trained voltage safety prediction model according to the feature similarity scores.

[0074] Exemplarily, the abnormal voltage pattern library stores feature templates under historical abnormal working conditions (such as abnormal voltage platforms caused by overcharging, and increased fluctuations caused by cell imbalance), and each template contains the statistical distribution of multi-level abstract features. Pattern matching calculates the similarity scores between the current feature map and the templates through the cosine similarity algorithm. The higher the score, the closer the current state is to a certain type of abnormality. Dynamically adjust the weight of the voltage decay coefficient according to the score: if the overcharging template is matched, increase the penalty weight of the voltage decay coefficient to suppress the overcharging risk; if the cell imbalance template is matched, reduce the weight to prioritize the balancing problem. The weight adjustment takes effect in real time through the parameters of the fully connected layer of the model to achieve adaptive prediction.

[0075] S2066. Perform multi-dimensional regression on the voltage safety prediction model with adjusted weights to generate voltage safety boundary values at different confidence levels.

[0076] Exemplarily, multi-dimensional regression uses the quantile regression algorithm, introduces multiple quantiles (such as 5%, 50%, 95%) into the loss function, and simultaneously predicts the voltage safety boundaries at different confidence levels. Input the features of the model with adjusted weights, and the regression layer outputs three channels: the lower limit value (conservative safety voltage), the median value (the most likely safety voltage), and the upper limit value (optimistic safety voltage). During training, optimize the prediction accuracy of each channel through the quantile loss function to ensure the statistical significance of the boundary values. An adaptive learning rate strategy is adopted during the regression process, and the parameter update step size is dynamically adjusted according to the prediction error to balance the convergence speed and stability.

[0077] S2067. Dynamically correct the voltage safety boundary value based on the current battery temperature and discharge rate of battery cell 11, and output the predicted safety voltage.

[0078] Exemplarily, the dynamic correction introduces a temperature-magnification compensation coefficient table, which calibrates the voltage offset under different combinations of temperature (such as 0°C - 45°C) and discharge rate (such as 0.5C - 3C) through experiments. According to the real-time collected battery temperature and discharge rate, look up the table to obtain the compensation coefficient, and linearly interpolate and correct the voltage safety boundary value obtained by regression. For example, in the scenario of low temperature and high magnification, the compensation coefficient is negative, and the safety voltage threshold is reduced to avoid over-discharge risk. The corrected boundary value generates the predicted safety voltage through weighted average (such as 60% median value weight, 20% for each of the upper and lower limits), ensuring that the output value not only conforms to the statistical law but also adapts to the real-time working conditions.

[0079] Please refer to Figure 3 , Figure 3 FIG. is a circuit schematic diagram of an energy storage system 100 provided by an embodiment of the present application.

[0080] In some embodiments, the switching unit includes: a first switching group, a second switching group, a third switching group, and a fourth switching group.

[0081] The first switching group includes: a first switch K11, a second switch K12, a third switch K13, and a fourth switch K14. Both the second switch K12 and the third switch K13 are used to connect to an external power supply and the charging unit 12. The first switch K11, the second switch K12, and the third switch K13 are all normally open switches, and the fourth switch K14 is a normally closed switch.

[0082] The second switching group includes: a fifth switch K21, and the fifth switch K21 is a normally open switch.

[0083] The third switching group includes: a sixth switch K31, and the sixth switch K31 is a normally open switch.

[0084] The fourth switching group includes: a seventh switch K41, an eighth switch K42, and a ninth switch K43. Both the seventh switch K41 and the eighth switch K42 are normally open switches, and the ninth switch K43 is a normally closed switch.

[0085] In some embodiments, the first power supply unit 14 includes: a transformer T1, a rectification component DB1, a first filter capacitor C1, and a first diode D1.

[0086] The first end of transformer T1 is used to connect to the first end of an external power supply. The second end of transformer T1 is connected to the external power supply through the fourth switch K14 and the seventh switch K41. The third end and the fourth end of transformer T1 are both connected to the input end of the rectification component DB1. The positive output of the rectification component DB1 is connected to the first end of the first filter capacitor C1 and the anode of the first diode D1. The negative output of the rectification component DB1 is connected to the first output end of the battery unit 11, the second end of the first filter capacitor C1, and the first input end of the buck unit 16. The cathode of the first diode D1 is connected to the second input end of the buck unit 16. The output end of the buck unit 16 is connected to the battery management unit 13.

[0087] In some embodiments, the second power supply unit 15 includes: a FUSE component, a tenth switch S2, an eleventh switch S3, and a second diode D2. The tenth switch S2 is a normally closed switch, and the eleventh switch S3 is a normally open switch.

[0088] The first end of the FUSE component is connected to the second output end of the battery unit 11. The second end of the FUSE component is connected to the anode of the second diode D2 through the tenth switch S2 and the eleventh switch S3. The cathode of the second diode D2 is connected to the second input end of the buck unit 16. Both ends of the eleventh switch S3 are respectively connected to both ends of the sixth switch K31.

[0089] In some embodiments, as Figure 3 shown, the energy storage system 100 further includes: an insulation detection unit 17. The first end of the insulation detection unit 17 is connected to the first output end of the battery unit 11. The second end of the insulation detection unit 17 is connected to the second output end of the battery unit 11. The third end of the insulation detection unit 17 is connected to the battery management unit 13.

[0090] In some embodiments, the second switch group further includes: a twelfth switch S4. The first end of the twelfth switch S4 is connected to a preset voltage of 24V+. The second end of the twelfth switch S4 is connected to the control end of the fifth switch K21. The control end of the fifth switch K21 is the control signal input end of a relay. The control end of the fifth switch K21 is also connected to the first end of the ninth switch K43. The second end of the ninth switch K43 is connected to the battery management unit 13. The first end of the fifth switch K21 is connected to the second end of the FUSE component. The second end of the fifth switch K21 is connected to one end of a load. The other end of the load is connected to the first end of the battery unit 11. Under normal discharge conditions, the battery management unit 13 controls the fifth switch K21 to close to control the battery unit 11 to supply electrical energy to the load. In the case of low battery power, the battery management unit 13 is in a sleep state, and the electrical energy guarantee in the emergency state is achieved by continuously pressing the twelfth switch S4.

[0091] Press the eleventh switch S3, and the battery pack voltage passes through the tenth switch S2, the eleventh switch S3, and the second diode D2 to the buck unit 16, and the drive voltage (12V or 24V) of the battery management unit 13 is output through the buck unit 16. The battery management unit 13 obtains the electrical energy provided by the second power supply unit 15 and enters the monitoring startup state. When the output voltage of the battery unit 11 is above the set sleep voltage, the insulation detection unit 17 detects whether the battery unit 11 is leaking electricity. If no leakage is detected, the battery management unit 13 sends a drive signal to the relay of the sixth switch K31 to control the sixth switch K31 to close, and the energy storage system 100 enters the normal working state; at the same time, a drive signal is sent to the relay of the fifth switch K21 to control the fifth switch K21 to close, and power can be supplied to the load.

[0092] When an external power supply (220V AC) is connected, the air switch Km switches to the closed state. The seventh switch K41 does not switch to the closed state, the first power supply unit 14 of the energy storage system 100 will not work, the eighth switch K42 does not switch to the closed state, the relay of the fourth switch K14 will not work, and the battery unit 11 will not enter automatic charging. When the output voltage of the battery unit 11 drops to the first output warning voltage, the battery management unit 13 will issue a low voltage prompt. After the battery unit 11 discharges for a certain period of time, the battery management unit 13 controls the fifth switch K21 to open, so that the battery unit 11 stops discharging. If the battery unit 11 needs to discharge emergently in a low battery state, press and hold the twelfth switch, so that the battery unit 11 can discharge emergently. When the output voltage of the battery management unit 13 drops again to the second output warning voltage, the battery management unit 13 drives the sixth switch K31 to open. At this time, the battery management unit 13 loses power supply, and the battery unit 11 will no longer be discharged.

[0093] If 220V is externally connected, the air switch switches to closed, the charging switch S1 switches to closed, S1-1 closes, the first power supply unit 14 enters the working state, the battery management unit 13 obtains the electrical energy provided by the first power supply unit 14, and the voltage of the cathode of the first diode D1 is 10V to 20V higher than the voltage of the cathode of the second diode D2, and the second diode D2 will be cut off (the second power supply unit 15 stops supplying power and no longer consumes the power of the battery unit 11), and the input to the buck unit 16 is all provided by the first power supply unit 14. When it is detected that the output voltage of the battery unit 11 drops below the first output warning voltage, the battery management unit 13 controls the first switch K11, the second switch K12, and the third switch K13 to close, the charging unit 12 enters the working state to charge the battery unit 11, and at the same time the fourth switch K14 opens and the first power supply unit 14 disconnects, and it is switched to be powered by the second power supply unit 15 to the battery management unit 13.

[0094] After the battery unit 11 is fully charged, the battery management unit 13 controls the first switch K11, the second switch K12, and the third switch K13 to open, and the charging unit 12 stops working (to prevent the charging unit 12 from affecting its lifespan due to long-term online operation). At the same time, the fourth switch K14 closes, the first power supply unit 14 resumes power supply, and the second power supply unit 15 stops power supply, no longer consuming the power of the battery unit 11. In the standby state, only the self-power consumption of the battery cells in the battery unit 11 exists, and the power drops slowly. In the no-load discharge state of the battery unit 11, it will not enter frequent charging due to the self-power consumption of the battery cells in the battery unit 11, which can extend the service life of the battery unit 11.

[0095] An embodiment of the present application provides an electronic device, which includes a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement the dual-path power supply method of the battery management unit as described in any one of the embodiments of the present application when executing the computer program.

[0096] An embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the processor is caused to implement the dual-path power supply method of the battery management unit as described in any one of the embodiments of the present application.

[0097] As mentioned above, the above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A dual-power supply method for a battery management unit, characterized in that, Applied to an energy storage system, the energy storage system includes: a battery unit, a charging unit, a battery management unit, a first power supply unit, a second power supply unit, a buck unit, and a switch unit. The first power supply unit and the second power supply unit are used to supply power to the battery management unit. The charging unit is respectively connected to an external power supply and the battery unit. The first power supply unit is connected to the external power supply. The second power supply unit is connected to the battery unit. The battery management unit controls the entire energy storage system through the switch unit. The method includes: The battery management unit collects the working information of the battery unit and sends the working information to the server to obtain the predicted safety voltage returned by the server, where the predicted safety voltage is determined based on the working information; When the external power supply is not connected, start the second power supply unit to supply power to the battery management unit; When the external power supply is connected, start the first power supply unit to supply power to the battery management unit and shield the second power supply unit; When the external power supply is connected and the output voltage of the battery unit is less than the predicted safety voltage, start the charging unit to supply power to the battery unit, turn off the first power supply unit, start the second power supply unit to supply power to the battery management unit, and shield the second power supply unit.

2. The dual-power supply method of the battery management unit according to claim 1, characterized in that The working information includes: battery temperature, charge decay rate, discharge voltage, and discharge current. When the server is used to execute the determination of the predicted safety voltage of the battery unit based on the working information, it is specifically used to execute: Perform time series alignment on the battery temperature, the charge decay rate, the discharge voltage, and the discharge current to generate a multi-dimensional battery working data set, and perform working condition state classification on the multi-dimensional battery working data set to obtain a first voltage fluctuation data set and a temperature drift data set under different discharge rates; Perform kernel principal component analysis on the first voltage fluctuation data set to extract a voltage time domain feature set, and perform wavelet packet decomposition on the temperature drift data set to generate a temperature frequency domain feature set; Perform cross-domain feature fusion on the voltage time domain feature set and the temperature frequency domain feature set to generate a composite feature matrix; Perform incremental feature dimension expansion on the composite feature matrix to obtain an enhanced safety feature set; Calculate the coupling coefficients of the charge decay rate and the discharge current in the enhanced safety feature set respectively, and perform dynamic weight assignment on the enhanced safety feature set based on the coupling coefficients to generate a safety evaluation feature vector; Input the safety evaluation feature vector into a pre-trained voltage safety prediction model for multi-dimensional regression analysis to obtain the predicted safety voltage of the battery unit, where the voltage safety prediction model is trained by abnormal working data of an abnormal charging unit.

3. The dual-power supply method of the battery management unit according to claim 2, characterized in that, The performing cross-domain feature fusion on the voltage time domain feature set and the temperature frequency domain feature set to generate a composite feature matrix includes: Perform time dimension alignment on the voltage time domain feature set and the temperature frequency domain feature set to generate a synchronized time-frequency feature sequence; Perform kernel matrix decomposition on the voltage time-domain features in the synchronized time-frequency feature sequence, extract the orthogonalized voltage principal component, perform frequency-band energy normalization on the temperature frequency-domain features in the synchronized time-frequency feature sequence, and generate a reduced-dimension temperature frequency-band energy distribution set; Construct a three-dimensional tensor from the orthogonalized voltage principal component and the reduced-dimension temperature frequency-band energy distribution set along the time axis, and fuse the time-domain - frequency-domain - energy dimension features through the tensor folding algorithm to generate an initial composite feature tensor; Perform dynamic channel weight assignment driven by the self-attention mechanism on the initial composite feature tensor to obtain the cross-domain correlation degree; Screen the effective feature channels according to a preset correlation degree threshold and the cross-domain correlation degree, and reconstruct and generate a composite feature matrix.

4. The dual-power supply method of the battery management unit according to claim 2, characterized in that The inputting the safety assessment feature vector into a pre-trained voltage safety prediction model for multi-dimensional regression analysis to obtain the predicted safety voltage of the battery cell includes: Perform Gaussian kernel function mapping on the safety assessment feature vector to generate a high-dimensional non-linear feature space; Perform time-series slicing on the high-dimensional non-linear feature space to obtain a multi-scale feature tensor sequence; Extract the voltage decay features under the historical charge and discharge modes from the multi-scale feature tensor sequence through a bidirectional gated recurrent network, and calculate the correlation weights of different time steps in the voltage decay features by using a multi-head attention mechanism to generate a dynamically attention-enhanced feature set; Input the dynamically attention-enhanced feature set into a residual convolution module, and output a multi-level abstract feature map through the layer-by-layer residual connection in the residual convolution module; Based on a preset abnormal voltage pattern library, perform abnormal pattern matching on the multi-level abstract feature map, calculate the feature similarity score between the multi-level abstract feature map and the preset patterns in the abnormal voltage pattern library, and dynamically adjust the weight allocation ratio of the voltage decay coefficient in the pre-trained voltage safety prediction model according to the feature similarity score; Perform multi-dimensional regression on the voltage safety prediction model with adjusted weights to generate voltage safety boundary values at different confidence levels; Dynamically correct the voltage safety boundary values based on the current battery temperature and discharge rate of the battery cell, and output the predicted safety voltage.

5. The dual-power supply method of the battery management unit according to claim 1, characterized in that, The switching unit includes: a first switch group, a second switch group, a third switch group, and a fourth switch group; The first switch group includes: a first switch (K11), a second switch (K12), a third switch (K13), and a fourth switch (K14). Both the second switch (K12) and the third switch (K13) are used to connect the external power supply and the charging unit. The first switch (K11), the second switch (K12), and the third switch (K13) are all normally open switches, and the fourth switch (K14) is a normally closed switch; The second switch group includes: a fifth switch (K21), and the fifth switch (K21) is a normally open switch; The third switch group includes: a sixth switch (K31), and the sixth switch (K31) is a normally open switch; The fourth switch group includes: a seventh switch (K41), an eighth switch (K42), and a ninth switch (K43). The seventh switch (K41) and the eighth switch (K42) are both normally open switches, and the ninth switch (K43) is a normally closed switch.

6. The dual-power supply method of the battery management unit according to claim 5, characterized in that, The first power supply unit includes: a transformer (T1), a rectification component (DB1), a first filter capacitor (C1), and a first diode (D1); The first end of the transformer (T1) is used to be connected to the first end of the external power supply. The second end of the transformer (T1) is connected to the external power supply through the fourth switch (K14) and the seventh switch (K41). The third end and the fourth end of the transformer (T1) are both connected to the input end of the rectification component (DB1). The positive output of the rectification component (DB1) is connected to the first end of the first filter capacitor (C1) and the anode of the first diode (D1). The negative output of the rectification component (DB1) is connected to the first output end of the battery unit, the second end of the first filter capacitor (C1), and the first input end of the buck unit. The cathode of the first diode (D1) is connected to the second input end of the buck unit. The output end of the buck unit is connected to the battery management unit.

7. The dual-power supply method of the battery management unit according to claim 6, characterized in that, The second power supply unit includes: a FUSE component, a tenth switch (S2), an eleventh switch (S3), and a second diode (D2). The tenth switch (S2) is a normally closed switch, and the eleventh switch (S3) is a normally open switch; The first end of the FUSE component is connected to the second output end of the battery unit. The second end of the FUSE component is connected to the anode of the second diode (D2) through the tenth switch (S2) and the eleventh switch (S3). The cathode of the second diode (D2) is connected to the second input end of the buck unit. The two ends of the eleventh switch (S3) are respectively connected to the two ends of the sixth switch (K31).

8. The dual-power supply method of the battery management unit according to claim 7, characterized in that, The second switch group further includes: a twelfth switch (S4). The first end of the twelfth switch (S4) is connected to a preset voltage. The second end of the twelfth switch (S4) is connected to the control end of the fifth switch (K21). The control end of the fifth switch (K21) is the control signal input end of a relay. The control end of the fifth switch (K21) is further connected to the first end of the ninth switch (K43). The second end of the ninth switch (K43) is connected to the battery management unit. The first end of the fifth switch (K21) is connected to the second end of the FUSE component. The second end of the fifth switch (K21) is connected to one end of a load. The other end of the load is connected to the first end of the battery unit.

9. The dual-power supply method of the battery management unit according to claim 7, characterized in that, The energy storage system further includes: An insulation detection unit, a first end of the insulation detection unit is connected to a first output end of the battery unit, a second end of the insulation detection unit is connected to a second output end of the battery unit, and a third end of the insulation detection unit is connected to the battery management unit.

10. An electronic device, characterized in that, The electronic device includes an energy storage system, and the energy storage system includes: a battery unit, a charging unit, a battery management unit, a first power supply unit, a second power supply unit, a buck unit, and a switching unit. The first power supply unit and the second power supply unit are used to supply electrical energy to the battery management unit. The charging unit is respectively connected to an external power supply and the battery unit. The first power supply unit is connected to the external power supply, and the second power supply unit is connected to the battery unit. The electronic device is configured to execute the dual-path power supply method of the battery management unit according to any one of claims 1 to 9.

Images