Battery health state estimation method and device and computer equipment

Abstract

The invention relates to a battery health state estimation method, a battery health state estimation device and computer equipment, and provides a multi-mode health feature (HI) extraction method, which introduces a plurality of health factors with physical significance. Comprise a typical charging interval electric quantity characteristic (Qsc) of a voltage plateau stage (VPP), a charging interval electric quantity characteristic (Qes) between an energy similar point and a cut-off voltage, coulombic efficiency (CE-HI), a charging temperature rise rate (delta T-HI), equivalent impedance (Z-HI) and the like. Through fusion of multi-dimensional physical characteristics, the battery health state estimation model can more comprehensively reflect the aging behavior of the battery, stable and accurate SOH prediction can still be realized under complex operation conditions such as non-constant multiplying power and incomplete circulation, and the adaptive capacity of the model to an actual application scene is remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of battery management system, and particularly relates to a battery state of health estimation method and device and computer equipment. BACKGROUND

[0002] The characterization ability of health indicators (HI) on battery aging is crucial for accurately estimating the state of health (SOH) of the battery using data-driven methods. The current commonly used health indicator (HI) extraction methods are mostly based on complete charge-discharge curves under constant current conditions, such as available capacity, incremental capacity (IC) characteristics, temperature changes, etc. These methods have good performance in experimental environments, but have obvious limitations in actual energy storage system application scenarios.

[0003] Under actual operating conditions, the charge-discharge rate of the battery usually changes dynamically with user demand, and it is difficult to achieve complete cycles covering 100% DOD. This leads to large differences in discharge duration, heat generation, and voltage / current response between different cycles. Therefore, under non-constant rate and non-complete cycle conditions, the adaptability of traditional HI to fragmented data is poor, which seriously affects the robustness and applicability of the SOH prediction model. SUMMARY

[0004] In view of this, the present application provides a battery state of health estimation method and device and computer equipment to solve the problem of poor adaptability of existing HI to fragmented data.

[0005] In a first aspect, the present invention provides a battery health state estimation method, comprising the following steps: acquiring historical operating data of a battery system; processing the historical operating data to obtain multiple sets of training data, wherein each set of training data includes first health feature data of the battery system and an actual label of the health state of the battery system, the first health feature data including at least one of the following: a typical charging interval power characteristic (Qsc) of a voltage platform stage (VPP), a charging interval power characteristic (Qes) from an energy similarity point to a cutoff voltage, coulombic efficiency (CE-HI), a charging temperature rise rate (ΔT-HI), and an equivalent impedance (Z-HI); utilizing the multiple sets of training data to obtain a battery health state estimation method, wherein the battery health state estimation method comprises: obtaining historical operating data of a battery system; processing the historical operating data to obtain multiple sets of training data, wherein each set of training data includes first health feature data of the battery system and an actual label of the health state of the battery system, the first health feature data including at least one of the following: a typical charging interval power characteristic (Qsc) of a voltage platform stage (VPP), a charging interval power characteristic (Qes) from an energy similarity point to a cutoff voltage, a coulombic efficiency (CE-HI), a charging temperature rise rate (ΔT-HI), and an equivalent impedance (Z-HI); utilizing the multiple sets of training data to obtain a battery health state estimation method, wherein the battery health state estimation method comprises: obtaining historical operating data of a battery system; processing the historical operating data to obtain multiple sets of training data ... The method comprises the following steps: training a preset first model using training data to obtain a battery health status estimation model; obtaining actual operation data of the battery system; processing the actual operation data to obtain second health characteristic data, wherein the second health characteristic data includes at least one of the following: a typical charging interval power characteristic (Qsc) of the voltage platform stage (VPP), a charging interval power characteristic (Qes) from the energy similarity point to the cut-off voltage, coulombic efficiency (CE-HI), a charging temperature rise rate (ΔT-HI), and an equivalent impedance (Z-HI); inputting the second health characteristic data into the battery health status estimation model to obtain the actual health status of the battery system.

[0006] This paper proposes a multimodal health feature extraction method that introduces multiple health factors with physical significance, including the typical charge interval power characteristic (Qsc) in the voltage plateau phase (VPP), the charge interval power characteristic (Qes) between the energy similarity point and the cutoff voltage, the coulombic efficiency (CE-HI), the charge temperature rise rate (ΔT-HI), and the equivalent impedance (Z-HI). By integrating multi-dimensional physical features, the battery health state estimation model can more comprehensively reflect the aging behavior of the battery and still achieve stable and accurate SOH prediction under complex operating conditions such as non-constant rate and incomplete cycle, significantly improving the model's adaptability to actual application scenarios.

[0007] In some optional embodiments, the method for determining the typical charging interval power characteristic (Qsc) of the voltage plateau stage (VPP) includes the following steps: determining the voltage plateau stage (VPP) of each charging cycle based on the voltage and current data of each charging cycle; determining the frequency of occurrence of each voltage in each voltage plateau stage (VPP); determining the voltage with an occurrence frequency equal to the total number of charging cycles based on the frequency of occurrence of each voltage in each voltage plateau stage (VPP); determining a commonly used voltage interval based on the voltage with an occurrence frequency equal to the total number of charging cycles; calculating the charged power of each charging cycle in the commonly used voltage interval; and obtaining the typical charging interval power characteristic (Qsc) of the voltage plateau stage (VPP) based on the charged power of all charging cycles in the commonly used voltage interval.

[0008] This implementation method accurately identifies commonly used voltage ranges by counting the voltage distribution frequency of the voltage plateau phase (VPP) in each charging cycle, and extracts health features based on the charged capacity within this range. This improves the stability and representativeness of feature selection, and helps to enhance the accuracy of SOH estimation and model robustness.

[0009] In some optional embodiments, a method for determining a charge interval charge characteristic (Qes) from an energy similarity point to a cutoff voltage includes: determining a voltage platform stage (VPP) of each charging cycle based on voltage and current data of each charging cycle; dividing each voltage platform stage (VPP) into equal voltages to obtain a plurality of equal voltage interval points, and obtaining a time point at which each equal voltage interval point is first reached during the charging process; for each equal voltage interval point, obtaining a plurality of time points at which each equal voltage interval point is first reached in a plurality of charging cycles, and calculating a variance of the plurality of time points; determining an energy similarity point based on a plurality of variances corresponding to the plurality of equal voltage interval points; determining a charged charge from the energy similarity point to the upper cutoff voltage in each charging cycle; and obtaining a charge interval charge characteristic (Qes) from the energy similarity point to the cutoff voltage based on the charged charge of all charging cycles.

[0010] This implementation calculates the variance of the time points at which equal voltage intervals are first reached in each charging cycle, identifies energy similarity points, and based on this, determines the charge characteristics from the energy similarity points to the cutoff voltage. This method effectively captures the changes in the battery's energy characteristics at different charging stages, improving the accuracy of SOH estimation and the model's adaptability to different operating conditions, while also enhancing the stability and reliability of feature extraction.

[0011] In some optional embodiments, the first model is a Transformer-LSTM combined architecture.

[0012] The first model utilizes a hybrid Transformer-LSTM architecture, enhancing time series modeling capabilities for SOH prediction. The Transformer architecture effectively extracts long-term trends in battery health through a self-attention mechanism, enhancing the model's ability to capture complex aging patterns. The LSTM architecture, on the other hand, focuses on capturing short-term dynamic changes during the charge and discharge process, compensating for the Transformer's shortcomings in learning local features. This combination of the two constructs a prediction model with a long-term and short-term synergistic mechanism, improving not only the stability and accuracy of SOH prediction but also its adaptability and generalization performance in real-world scenarios such as incomplete data and complex operating conditions.

[0013] In some optional embodiments, after processing the actual operating data to obtain the second health feature data, it also includes: generating pseudo labels based on the second health feature data using Gaussian process regression to obtain pseudo label samples; and performing incremental training on the battery health status estimation model by combining a dynamic sliding window to screen pseudo label samples with a confidence level higher than a preset threshold.

[0014] This implementation improves the accuracy of SOH prediction under data-sparse conditions by introducing a pseudo-label enhancement strategy based on Gaussian process regression (GPR). When the true state of health (SOH) labels of batteries are limited, the GPR model is used to generate pseudo-label samples with high confidence, effectively expanding the training dataset, thereby improving the training effect and generalization ability of the Transformer-LSTM model. Combined with an adaptive sliding window strategy, training samples are dynamically screened and updated, further enhancing the model's online learning capabilities and adaptability to changing operating conditions in a real-time operating environment.

[0015] In some optional embodiments, after inputting the second health feature data into the battery health state estimation model to obtain the actual health state of the battery system, the method further includes: correcting the actual health state of the battery system according to Kalman filtering.

[0016] After completing SOH prediction based on the Transformer-LSTM model, an adaptive Kalman filter error correction mechanism is introduced to improve prediction accuracy and long-term stability. Kalman filtering (KF) optimizes the prediction results, effectively reducing noise interference and enhancing the robustness of the estimation. A dynamic Kalman gain adjustment mechanism is also used to adjust the gain parameters in real time based on historical error information, enabling the model to maintain stable prediction performance over long periods of time, further improving the accuracy and reliability of SOH estimation.

[0017] In the second aspect, the present invention also provides a battery health status estimation device, which includes a historical data acquisition module, a historical data processing module, a model training module, an actual data acquisition module, an actual data processing module and a prediction module: wherein the historical data acquisition module is used to acquire the historical operation data of the battery system; the historical data processing module is used to process the historical operation data to obtain multiple groups of training data, wherein each group of training data includes the first health feature data of the battery system and the actual label of the health status of the battery system, and the first health feature data includes at least one of the following: a typical charging interval power characteristic (Qsc) of the voltage platform stage (VPP), a charging interval power characteristic (Qes) from the energy similarity point to the cut-off voltage, and a coulombic efficiency , charging temperature rise rate, equivalent impedance; a model training module, used to train the preset first model using multiple sets of training data to obtain a battery health status estimation model; an actual data acquisition module, used to obtain the actual operation data of the battery system; an actual data processing module, used to process the actual operation data to obtain second health characteristic data, wherein the second health characteristic data includes at least one of the following: a typical charging interval power characteristic (Qsc) of the voltage platform stage (VPP), a charging interval power characteristic (Qes) from the energy similarity point to the cut-off voltage, coulomb efficiency, charging temperature rise rate, and equivalent impedance; a prediction module, used to input the second health characteristic data into the battery health status estimation model to obtain the actual health status of the battery system.

[0018] In a third aspect, the present invention also provides a computer device comprising a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the battery health status estimation method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0019] In a fourth aspect, the present invention further provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the battery health status estimation method of the first aspect or any corresponding embodiment thereof.

[0020] In a fifth aspect, the present invention further provides a computer program product, comprising computer instructions, which are used to enable a computer to execute the battery health status estimation method of the above-mentioned first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 is a flowchart of a method for estimating a battery health state according to an embodiment of the present invention;

[0023] Figure 2 is a flowchart of another battery health status estimation method according to an embodiment of the present invention;

[0024] Figure 3 is a schematic diagram of a non-constant rate operating condition provided according to an embodiment of the present invention;

[0025] Figure 4 is a schematic diagram of curves of all cycles provided according to an embodiment of the present invention;

[0026] Figure 5 Schematic diagram of voltage frequency curves and statistical curves of each cycle provided according to an embodiment of the present invention;

[0027] Figure 6 is a schematic diagram of a cluster of coordinate curves of equal voltage interval points provided in an embodiment of the present invention;

[0028] Figure 7 2 is a schematic diagram of the variance of coordinate data of each point provided by an embodiment of the present invention;

[0029] Figure 8 1. A schematic diagram of a Qsc sequence and smoothing thereof provided according to an embodiment of the present invention;

[0030] Figure 9 1. A schematic diagram of a Qes sequence and smoothing thereof provided according to an embodiment of the present invention;

[0031] Figure 10 1. It is a schematic diagram of the model training, application and update process provided by an embodiment of the present invention;

[0032] Figure 11 is a schematic diagram of test result errors provided by an embodiment of the present invention;

[0033] Figure 12 2. It is a schematic diagram showing the effect of the number of pseudo cycles on the estimation result provided by an embodiment of the present invention;

[0034] Figure 13 is a structural block diagram of a battery health status estimation device according to an embodiment of the present invention;

[0035] Figure 14 Schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0037] According to an embodiment of the present invention, an embodiment of a battery health status estimation method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0038] This embodiment provides a battery health status estimation method that can be used in computer equipment. Figure 1 FIG. 1 is a flow chart of a method for estimating a battery health state according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0039] Step S101: Acquire historical operating data of the battery system.

[0040] In this embodiment, the first model is a Transformer-LSTM combined architecture, which can improve the time series modeling capabilities in SOH prediction. The Transformer structure effectively extracts long-term trends in battery health status through a self-attention mechanism, enhancing the model's ability to characterize complex aging patterns; while the LSTM structure focuses on capturing short-term dynamic changes during the charging and discharging process, compensating for the Transformer's shortcomings in local feature learning. The combination of the two constructs a prediction model with a long-term and short-term synergistic mechanism, which not only improves the stability and accuracy of SOH prediction, but also enhances the model's adaptability and generalization performance in real-world scenarios such as incomplete data and complex operating conditions.

[0041] Step S102: Process the historical operating data to obtain multiple sets of training data, where each set of training data includes first health characteristic data of the battery system and an actual label of the health status of the battery system, where the multiple sets of training data include at least one of the following: a typical charging interval power characteristic (Qsc) of the voltage platform stage (VPP), a charging interval power characteristic (Qes) from the energy similarity point to the cut-off voltage, coulombic efficiency (CE-HI), a charging temperature rise rate (ΔT-HI), and an equivalent impedance (Z-HI).

[0042] Specifically, Coulombic efficiency (CE-HI) refers to the ratio of discharge capacity to charge capacity; charging temperature rise rate (ΔT-HI) refers to the rate of temperature change during VPP; equivalent impedance (Z-HI) refers to the calculation of the equivalent AC impedance under different SOCs.

[0043] In some optional embodiments, the method for determining the typical charging interval power characteristics (Qsc) of the voltage plateau stage (VPP) includes the following steps: determining the voltage plateau stage (VPP) of each charging cycle based on the voltage and current data of each charging cycle; determining the frequency of occurrence of each voltage in each voltage plateau stage (VPP); determining the voltage with an occurrence frequency equal to the total number of charging cycles based on the frequency of occurrence of each voltage in each voltage plateau stage (VPP); determining a commonly used voltage interval based on the voltage with an occurrence frequency equal to the total number of charging cycles; calculating the charged power of each charging cycle in the commonly used voltage interval; and obtaining the typical charging interval power characteristics (Qsc) of the voltage plateau stage (VPP) based on the charged power of all charging cycles in the commonly used voltage interval.

[0044] That is, obtain the voltage and current data of a charging cycle, calculate the frequency of occurrence of each voltage in the voltage plateau phase (VPP), count the voltage intervals that are commonly covered by all cycles, that is, the voltage intervals with a frequency equal to the total number of cycles, and calculate the charged capacity in this interval as Qsc. Repeat the calculation for each cycle until the Qsc of all cycles is obtained.

[0045] The above-mentioned method for determining the typical charging interval power characteristics (Qsc) of the voltage plateau stage (VPP) accurately identifies the commonly used voltage range by counting the voltage distribution frequency of the voltage plateau stage (VPP) in each charging cycle, and extracts health characteristics based on the charged power within this range. This improves the stability and representativeness of feature selection, and helps to enhance the accuracy of SOH estimation and model robustness.

[0046] In an optional embodiment, a method for determining a charge interval charge characteristic (Qes) from an energy similarity point to a cutoff voltage includes the following steps: determining a voltage platform stage (VPP) of each charging cycle based on voltage and current data of each charging cycle; dividing each voltage platform stage (VPP) into equal voltages to obtain a plurality of equal voltage interval points, and obtaining a time point at which each equal voltage interval point is first reached during the charging process; for each equal voltage interval point, obtaining a plurality of time points in a plurality of charging cycles at which each equal voltage interval point is first reached, and calculating a variance of the plurality of time points; determining an energy similarity point based on a plurality of variances corresponding to the plurality of equal voltage interval points; determining a charged charge from the energy similarity point to the upper cutoff voltage in each charging cycle; and obtaining a charge interval charge characteristic (Qes) from an energy similarity point to a cutoff voltage based on the charged charge of all charging cycles.

[0047] That is, obtain the voltage and current data of a charging cycle. For the voltage plateau phase (VPP), start from Umin and divide it into equal voltage intervals every ΔU; for each cycle, record the coordinates of the charging time when each equal voltage interval point is reached; calculate the variance of all coordinate values ​​of each equal voltage interval point; identify the voltage with the smallest variance within the VPP range and record it as the energy similarity point; calculate the charged charge Qes between Use and the upper cut-off voltage; and calculate each cycle until the Qes of all cycles is obtained.

[0048] The method for determining the charge characteristics (Qes) from the energy similarity point to the cutoff voltage calculates the variance of the time points at which the equal voltage interval is first reached in each charging cycle, identifies the energy similarity point, and based on this, determines the charge characteristics from the energy similarity point to the cutoff voltage. This method effectively captures the changes in the battery's energy characteristics at different charging stages, improving the accuracy of SOH estimation and the model's adaptability to different operating conditions, while also enhancing the stability and reliability of feature extraction.

[0049] Specifically, the voltage plateau phase (VPP) of each charging cycle can be determined based on the voltage and current data of each charging cycle. The following method can be used: extract the charging and discharging segments with stable current rates, and select the charging curve with the largest SOC span (ideally 0% to 100%) to determine the voltage plateau phase (VPP). In other words, in the battery test data, first select those charging and discharging processes with constant current, and then select the charging curve with the largest SOC variation range (preferably 0% to 100%). Then, on this curve, identify the area with slow voltage variation (i.e., the voltage plateau phase (VPP)), providing a basis for subsequent battery modeling or SOC estimation.

[0050] Furthermore, before processing the historical operation data, data cleaning, denoising and smoothing are performed on the historical operation data to make the voltage data monotonically increasing.

[0051] Step S103: using multiple sets of training data to train a preset first model to obtain a battery health status estimation model.

[0052] Step S104: Acquire actual operating data of the battery system.

[0053] Step S105: Process the actual operation data to obtain second health characteristic data, where the second health characteristic data includes at least one of the following: a typical charging interval power characteristic (Qsc) in the voltage platform stage (VPP), a charging interval power characteristic (Qes) from the energy similarity point to the cut-off voltage, coulombic efficiency (CE-HI), a charging temperature rise rate (ΔT-HI), and an equivalent impedance (Z-HI).

[0054] Step S106: inputting the second health characteristic data into a battery health state estimation model to obtain the actual health state of the battery system.

[0055] This embodiment proposes a multimodal health feature extraction method, which introduces multiple health factors with physical significance, including the charge interval characteristic (Qsc) in the voltage platform phase (VPP), the charge interval characteristic (Qes) between the energy similarity point and the cutoff voltage, the coulombic efficiency (CE-HI), the charge temperature rise rate (ΔT-HI), and the equivalent impedance (Z-HI). By integrating multi-dimensional physical features, the battery health state estimation model can more comprehensively reflect the aging behavior of the battery, and can still achieve stable and accurate SOH prediction under complex operating conditions such as non-constant rate and incomplete cycle, significantly improving the model's adaptability to actual application scenarios.

[0056] This embodiment provides a battery health status estimation method that can be used in computer equipment. Figure 2 FIG. 1 is a flow chart of another method for estimating the state of health of a battery according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0057] Step S201: Acquire historical operating data of the battery system.

[0058] Step S202: Process the historical operating data to obtain multiple sets of training data, where each set of training data includes first health characteristic data of the battery system and an actual label of the health status of the battery system, where the multiple sets of training data include at least one of the following: a typical charging interval power characteristic (Qsc) of the voltage platform stage (VPP), a charging interval power characteristic (Qes) from the energy similarity point to the cut-off voltage, coulombic efficiency (CE-HI), a charging temperature rise rate (ΔT-HI), and an equivalent impedance (Z-HI).

[0059] Step S203: using multiple sets of training data to train the preset first model to obtain a battery health status estimation model.

[0060] Step S204: Acquire actual operating data of the battery system.

[0061] Step S205: Process the actual operation data to obtain second health characteristic data, wherein the second health characteristic data includes at least one of the following: a typical charging interval power characteristic (Qsc) in the voltage platform stage (VPP), a charging interval power characteristic (Qes) from the energy similarity point to the cut-off voltage, coulombic efficiency (CE-HI), a charging temperature rise rate (ΔT-HI), and an equivalent impedance (Z-HI).

[0062] Step S206: inputting the second health characteristic data into a battery health state estimation model to obtain the actual health state of the battery system.

[0063] Step S207: Generate pseudo labels using Gaussian process regression based on the second health feature data to obtain pseudo label samples.

[0064] Step S208: Incrementally train the battery health status estimation model by combining a dynamic sliding window to screen pseudo-label samples with confidence levels higher than a preset threshold.

[0065] By introducing a pseudo-label enhancement strategy based on Gaussian process regression (GPR), the accuracy of SOH prediction is improved under data-sparse conditions. When the true state-of-health (SOH) labels are limited, the GPR model is used to generate pseudo-label samples with high confidence, effectively expanding the training dataset, thereby improving the training effect and generalization ability of the Transformer-LSTM model. Combined with an adaptive sliding window strategy, training samples are dynamically filtered and updated, further enhancing the model's online learning capabilities and adaptability to changing operating conditions in real-time operating environments.

[0066] Step S209: Correcting the actual health status of the battery system according to Kalman filtering.

[0067] Specifically, the actual state of health of the battery system is corrected according to Kalman filtering, including the following steps: obtaining the actual state of health of the battery system, and correcting errors by using Kalman filtering, wherein a gain weight K in the Kalman filtering is adaptively adjusted according to historical errors.

[0068] For example, the gain weight K in the Kalman filtering is adaptively adjusted according to historical errors by using a formula:

[0069] SOH KF = K x SOH LSTM + (1-K) x SOH GPR

[0070] wherein SOH KF represents a corrected value of the actual state of health of the battery to be measured, SOH LSTM represents the actual state of health of the battery to be measured, and SOH GPR represents a pseudo label of a Gaussian regression process.

[0071] After the SOH prediction based on the Transformer-LSTM model is completed, an adaptive Kalman filtering error correction mechanism is introduced to improve the prediction accuracy and long-term stability. The prediction result is optimized by Kalman filtering (KF), which effectively reduces the noise interference and enhances the robustness of the estimation. At the same time, a dynamic Kalman gain adjustment mechanism is adopted to adjust the gain parameter in real time according to the historical error information, so that the model can still maintain stable prediction performance in long-term operation, further improving the accuracy and reliability of SOH estimation.

[0072] To make the battery state of health estimation method of the application clearer, a specific example is given.

[0073] The data used in this example is a certain photovoltaic park energy storage system, such as Figure 3 , the working conditions are photovoltaic consumption (A~B), grid support (C~F), and the F~H stage is a certain device constant power consumption. Figure 3 The c and d in the formula are constant-volume working conditions interspersed in the running working conditions, which are used to correct the actual value of the SOH value.

[0074] Figure 4 All data after a period of operation are shown in the following table: Figure 4 It can be seen that the voltage is reduced and the charging time is shortened.

[0075] Figure 5 The voltage frequency curve and statistical curve diagram of each cycle provided by the embodiment of the application is shown in the following figure: Figure 5 In the figure, the curve cluster represents the voltage-frequency curve of each cycle, and a single curve represents the number of times the voltage point appears in all cycles. For this battery system, Umin＝ 3.362, U max =3.428.

[0076] Figure 6 Schematic diagram of a cluster of coordinate curves of equal voltage interval points provided according to an embodiment of the present invention. Figure 6 As shown in Figure 1, there is a cluster of equal voltage coordinates in the VPP stage. The equal voltage intervals between 3.44 and 3.364 V (ΔU = 0.005) are calculated, and the coordinates (Index) of each equal voltage point reached in each cycle are counted. It can be seen that: (1) the coordinates reaching higher voltages are larger, that is, the required charging time is longer; (2) as the number of cycles increases and the battery ages, the coordinates required to reach 3.44 V become larger, but the coordinates reaching 3.364 V become smaller, that is, the time spent on this curve becomes shorter, and less electricity is charged at the same rate.

[0077] For cycles without a complete voltage plateau period, the voltage points in the plateau period in multiple cycles are counted, and the sampling point curve of the voltage is calculated. The voltage point with the smallest standard deviation is the "relatively stable energy similarity point". Figure 7 As shown in the figure, 3.386V is the energy similarity point voltage, which means that no matter what the working conditions are, a certain amount of charging power will be required to reach this voltage point.

[0078] The typical charge interval power characteristics (Qsc) Qsc of the voltage platform phase (VPP) and the charge interval power characteristics (Qes) Qes of the energy similarity point to the cut-off voltage obtained by the method of the present invention are respectively as follows: Figure 8 and Figure 9 shown.

[0079] Based on the above data, the M1 model was trained. After the battery system was tested in the laboratory, it took a long time for it to start operation in the park. The model (M1) established based on the experimental data lost its effectiveness. For this reason, a small amount of data was used as the test set 1 during the trial operation, and the GPR model was used to form a pseudo test set. A small amount of data from the experimental data was selected as the validation set, and together with the pseudo test set, the training of M2 was completed. The process of training, application and update is as follows: Figure 10 .

[0080] During operation, the error is evaluated using the test set 2, as shown in Figure 11 As shown in the figure, compared with the traditional hcLSTM, the evaluation accuracy of this method is significantly improved.

[0081] Adjust the number of pseudo-cycles, such as Figure 12 It can be seen that at least 6 pseudo cycles need to be predicted using GPR to ensure that the upper and lower errors do not exceed 0.9%, and 10 pseudo cycles are required to achieve the optimal estimate.

[0082] This embodiment also provides a battery health status estimation device, which is used to implement the above-mentioned embodiments and preferred embodiments. Details that have already been described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0083] This embodiment provides a battery health status estimation device, such as Figure 13 Shown, including:

[0084] The historical data acquisition module 1301 is used to acquire historical operating data of the battery system.

[0085] The historical data processing module 1302 is used to process the historical operating data to obtain multiple sets of training data, where each set of training data includes the first health characteristic data of the battery system and the actual label of the health status of the battery system, where the multiple sets of training data include at least one of the following: a typical charging interval power characteristic (Qsc) of the voltage platform stage (VPP), a charging interval power characteristic (Qes) from the energy similarity point to the cut-off voltage, coulombic efficiency, charging temperature rise rate, and equivalent impedance.

[0086] The model training module 1303 is used to train a preset first model using multiple sets of training data to obtain a battery health status estimation model.

[0087] The actual data acquisition module 1304 is configured to acquire actual operating data of the battery system.

[0088] The actual data processing module 1305 is used to process the actual operation data to obtain second health characteristic data, where the second health characteristic data includes at least one of the following: a typical charging interval power characteristic (Qsc) in the voltage platform stage (VPP), a charging interval power characteristic (Qes) from the energy similarity point to the cut-off voltage, coulombic efficiency, and equivalent impedance.

[0089] The prediction module 1306 is configured to input the second health characteristic data into a battery health state estimation model to obtain the actual health state of the battery system.

[0090] In some optional embodiments, the method for determining the typical charging interval power characteristic (Qsc) of the voltage plateau stage (VPP) includes the following steps: determining the voltage plateau stage (VPP) of each charging cycle based on the voltage and current data of each charging cycle; determining the frequency of occurrence of each voltage in each voltage plateau stage (VPP); determining the voltage with an occurrence frequency equal to the total number of charging cycles based on the frequency of occurrence of each voltage in each voltage plateau stage (VPP); determining a commonly used voltage interval based on the voltage with an occurrence frequency equal to the total number of charging cycles; calculating the charged power of each charging cycle in the commonly used voltage interval; and obtaining the typical charging interval power characteristic (Qsc) of the voltage plateau stage (VPP) based on the charged power of all charging cycles in the commonly used voltage interval.

[0091] In some optional embodiments, a method for determining a charge interval charge characteristic (Qes) from an energy similarity point to a cutoff voltage includes: determining a voltage platform stage (VPP) of each charging cycle based on voltage and current data of each charging cycle; dividing each voltage platform stage (VPP) into equal voltages to obtain a plurality of equal voltage interval points, and obtaining a time point at which each equal voltage interval point is first reached during the charging process; for each equal voltage interval point, obtaining a plurality of time points at which each equal voltage interval point is first reached in a plurality of charging cycles, and calculating a variance of the plurality of time points; determining an energy similarity point based on a plurality of variances corresponding to the plurality of equal voltage interval points; determining a charged charge from the energy similarity point to the upper cutoff voltage in each charging cycle; and obtaining a charge interval charge characteristic (Qes) from the energy similarity point to the cutoff voltage based on the charged charge of all charging cycles.

[0092] In some optional embodiments, the first model is a Transformer-LSTM combined architecture.

[0093] In some optional embodiments, the battery health state estimation apparatus further includes an incremental training module. After processing actual operating data to obtain second health characteristic data, the incremental training module is configured to: generate pseudo-labels based on the second health characteristic data using Gaussian process regression to obtain pseudo-label samples; and perform incremental training on the battery health state estimation model by filtering pseudo-label samples with confidence levels above a preset threshold using a dynamic sliding window.

[0094] In some optional embodiments, the battery health state estimation apparatus further includes a correction module. After inputting the second health characteristic data into the battery health state estimation model to obtain the actual health state of the battery system, the correction module is configured to correct the actual health state of the battery system based on a Kalman filter.

[0095] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0096] The battery health status estimation device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0097] The embodiment of the present invention also provides a computer device having the above Figure 13 The battery health status estimation device shown.

[0098] See also Figure 14 , Figure 14 is a structural diagram of a computer device provided by an optional embodiment of the present invention, such as Figure 14 As shown, the computer device includes: one or more processors 10, memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components utilize different buses to communicate with each other and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in the memory or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Equally, multiple computer devices can be connected, and each device provides part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 14 A processor 10 is taken as an example.

[0099] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0100] The memory 20 stores instructions that can be executed by at least one processor 10, so as to enable at least one processor 10 to execute the method shown in the above embodiment.

[0101] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and application programs required for at least one function; the data storage area may store data created based on the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0102] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0103] The computer device further includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 14 The bus connection is taken as an example.

[0104] The input device 30 can receive input digital or character information and generate key signal input related to user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a trackpad, a touch pad, an indicator stick, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 can include a display device, an auxiliary lighting device (e.g., an LED), and a tactile feedback device (e.g., a vibration motor). The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display, and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0105] The embodiments of the present application further provide a computer readable storage medium, and the method according to the embodiments of the present application can be implemented in hardware, firmware, or recorded in a storage medium, or stored in a remote storage medium or a non-transitory machine readable storage medium and downloaded to a local storage medium through network, so that the method described herein can be processed by such software on a storage medium using a general purpose computer, a special purpose processor, or programmable or special hardware. The storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid state disk, etc. Further, the storage medium can also include a combination of the above-mentioned memories. It can be understood that the computer, the processor, the microprocessor controller, or the programmable hardware includes a storage component that can store or receive software or computer code, when the software or computer code is accessed and executed by the computer, the processor, or the hardware, the method shown in the above embodiments is implemented.

[0106] Part of the present application can be applied as a computer program product, for example, computer program instructions, when executed by a computer, the operation of the computer can invoke or provide the method and / or technical solutions according to the present application. Those skilled in the art should understand that the form of computer program instructions in computer readable medium includes but is not limited to source file, executable file, installation package file, etc. Correspondingly, the way of computer program instructions executed by computer includes but is not limited to: the computer directly executes the instructions, or the computer compiles the instructions and then executes the corresponding compiled program, or the computer reads and executes the instructions, or the computer reads and installs the instructions and then executes the corresponding installed program. Here, the computer readable medium can be any available computer readable storage medium or communication medium accessible to the computer.

[0107] Although the embodiments of the present application are described in conjunction with the accompanying drawings, various modifications and changes can be made by those skilled in the art without departing from the spirit and scope of the present application, and such modifications and changes fall within the scope defined by the appended claims.

Claims

1. A method for estimating a battery health state, characterized in that: include: Obtain historical operating data of the battery system; Processing the historical operating data to obtain multiple sets of training data, wherein each set of training data includes first health characteristic data of the battery system and an actual label of the health state of the battery system, the first health characteristic data including at least one of the following: a typical charging interval power characteristic in a voltage plateau phase, a charging interval power characteristic from an energy similarity point to a cutoff voltage, coulombic efficiency, a charging temperature rise rate, and an equivalent impedance; Using the multiple sets of training data to train a preset first model to obtain a battery health state estimation model; Acquiring actual operating data of the battery system; Processing the actual operation data to obtain second health characteristic data, wherein the second health characteristic data includes at least one of the following: a typical charging interval power characteristic in a voltage plateau phase, a charging interval power characteristic from an energy similarity point to a cutoff voltage, coulombic efficiency, a charging temperature rise rate, and an equivalent impedance; The second health characteristic data is input into the battery health state estimation model to obtain the actual health state of the battery system.

2. The method according to claim 1, characterized in that The method for determining the typical charging interval power characteristics in the voltage plateau stage includes the following steps: Determining a voltage plateau phase of each charging cycle based on voltage and current data of each charging cycle; respectively determining the occurrence frequency of each voltage in each voltage platform stage; Determining, based on the frequency of occurrence of each voltage in each voltage plateau stage, a voltage having an occurrence frequency equal to the total number of charging cycles; Determine the commonly used voltage range based on the voltage whose frequency is equal to the total number of charging cycles; Calculating the amount of charge in each charging cycle within the commonly used voltage range; According to the charged quantity in the common voltage range of all charging cycles, the typical charging interval quantity characteristic of the voltage platform stage is obtained.

3. The method according to claim 1, characterized in that The method for determining the charge characteristics of the charging interval from the energy similarity point to the cut-off voltage includes: Determining a voltage plateau phase of each charging cycle based on voltage and current data of each charging cycle; Dividing each voltage platform stage into equal voltages to obtain a plurality of equal voltage interval points, and obtaining a time point at which each equal voltage interval point is first reached during the charging process; For each of the equal voltage interval points, obtaining multiple time points at which each of the equal voltage interval points is first reached in multiple charging cycles, and calculating the variance of the multiple time points; determining energy similarity points based on a plurality of variances corresponding to a plurality of the equal voltage interval points; respectively determining the amount of charge from the energy similarity point to the upper cut-off voltage in each charging cycle; The charge characteristics of the charging interval from the energy similarity point to the cut-off voltage are obtained based on the charged charge of all charging cycles.

4. The method according to claim 1, wherein The first model is a Transformer-LSTM combined architecture.

5. The method according to claim 1, wherein After processing the actual operation data to obtain the second health characteristic data, the method further includes: Generating pseudo labels using Gaussian process regression based on the second health feature data to obtain pseudo label samples; The battery health state estimation model is incrementally trained by combining a dynamic sliding window to screen pseudo-label samples with confidence levels higher than a preset threshold.

6. The method according to claim 1, characterized in that After inputting the second health characteristic data into the battery health state estimation model to obtain the actual health state of the battery system, the method further includes: The actual health state of the battery system is corrected according to the Kalman filter.

7. A battery health status estimation device, characterized in that: The device comprises: A historical data acquisition module is used to obtain historical operating data of the battery system; a historical data processing module, configured to process the historical operating data to obtain multiple sets of training data, wherein each set of training data includes first health characteristic data of the battery system and an actual label of the health status of the battery system, wherein the first health characteristic data includes at least one of the following: a typical charging interval power characteristic during a voltage plateau phase, a charging interval power characteristic from an energy similarity point to a cutoff voltage, coulombic efficiency, a charging temperature rise rate, and an equivalent impedance; A model training module, configured to train a preset first model using the multiple sets of training data to obtain a battery health state estimation model; An actual data acquisition module, used to acquire actual operating data of the battery system; an actual data processing module, configured to process the actual operation data to obtain second health characteristic data, wherein the second health characteristic data includes at least one of the following: a typical charging interval power characteristic in a voltage plateau phase, a charging interval power characteristic from an energy similarity point to a cutoff voltage, coulombic efficiency, a charging temperature rise rate, and an equivalent impedance; A prediction module is used to input the second health feature data into the battery health state estimation model to obtain the actual health state of the battery system.

8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the battery health status estimation method according to any one of claims 1 to 6 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the battery health state estimation method according to any one of claims 1 to 6.

10. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the battery health state estimation method according to any one of claims 1 to 6.

Images