A method and a terminal for short - circuit identification and prediction of a battery circuit

Abstract

A method and a terminal for short - circuit identification and prediction of a battery circuit. For each battery cell in a series - connected lithium - battery system, an equivalent circuit model containing a short - circuit equivalent resistance is established. According to the circuit parameters of the equivalent circuit model, within each detection period, for each battery cell, the first SOC data set is obtained by estimating the state of charge (SOC) through the Kalman - family algorithm, and the second SOC data set is obtained by estimating the SOC through the ampere - hour integration method. According to the first SOC data set and the second SOC data set, a short - circuit resistance data set is calculated. According to the short - circuit resistance data set, the deviation between the short - circuit resistance of each battery cell and the average short - circuit resistance is calculated to determine whether there is a fault in the short - circuit resistance of the battery cell. According to the circuit parameters of the equivalent circuit model and the short - circuit resistance data set, the average heat - generation power of each battery cell is calculated, and the maximum heat - dissipation power of the battery cell is obtained to predict the time when the heat - dissipation capacity limit of each battery cell in the system is reached.

Description

Technical Field

[0001] The present invention relates to the field of lithium battery status early warning, and in particular to a method and a terminal for identifying and predicting short circuits in battery circuits. Background Art

[0002] In lithium-ion battery energy storage systems, internal short circuits are the most representative serious faults and one of the important reasons for early power failure and thermal runaway safety accidents in the system. The detection or prediction methods of internal short circuits in batteries have become a key research direction in the field. Lithium batteries are likely to have thermal runaway throughout their entire life cycle of manufacturing, use, and recycling. For the use process, such as overcharging, overdischarging, and high temperature, it is more likely to cause irreversible impedance reduction caused by internal defects or battery materials, which is reflected as an internal short circuit of the battery. Studies have shown that this process can last for a certain period of time in the early stage, manifested as a slow and abnormal decrease in short-circuit resistance; in the middle stage, it is manifested as prominent battery imbalance problems and obvious abnormal heat generation; in the late stage, it is manifested as a strong temperature rise and the action of the fire alarm device.

[0003] Current research on internal short circuit identification and warning can be divided into the following categories:

[0004] (1) The electrochemical model method is to establish equations for the solid phase, liquid phase, interface, and deintercalation reaction processes of battery materials, thereby defining a simulation of the real battery reaction process, replacing the commonly used OCV (SOC, T) curve to describe the battery working process, especially estimating the impedance used for the equivalent internal short-circuit circuit. When the actual working state of the battery deviates from the modeled working state, it is considered that an impedance reduction trend has occurred. This type of method requires the establishment of a complex mathematical model to describe the electrochemical process, which is difficult for lithium battery users outside of research institutions to implement. In addition, the accurate use of the model requires corresponding real data to undergo complex calibration to adapt to different material systems, which is a long process and difficult to operate.

[0005] (2) The essence of data-based machine learning methods is to analyze other battery data through a trained fault diagnosis model. This method relies on a large amount of test set data, especially data from the time when an internal short circuit occurs until thermal runaway occurs. This is not easy to obtain for lithium-ion batteries that emphasize safety, especially lithium iron phosphate batteries, resulting in relatively scarce data that can be used to train the model and reduced accuracy.

[0006] (3) Based on the status variables collected by the monitoring system, the inconsistency changes and fault information of abnormal mutations of the battery are analyzed through the parameter characteristic statistical calculation method. This type of method does not require the establishment of an accurate battery model and only needs to analyze the differences in battery voltage and temperature to achieve internal short circuit detection. However, the disadvantage is that the accuracy of statistical calculation depends on the sampling accuracy, and the consistency of single cells in the battery pack and the initial state deviation directly affect the detection results.

[0007] Therefore, it is necessary to propose a short circuit identification and warning method that does not rely on the abnormally gradual change or fault data of the short circuit internal resistance in historical operation data, does not perform complex electrochemical principle-based modeling, and only needs to perform circuit-level modeling on the battery. Summary of the Invention

[0008] The technical problem to be solved by the present invention is: to provide a method and terminal for short circuit identification and prediction of a battery circuit, which can achieve short circuit identification and warning without relying on the abnormally gradual change or fault data of the short circuit internal resistance in historical operation data and without performing complex electrochemical principle-based modeling.

[0009] To solve the above technical problems, the technical solution adopted by the present invention is:

[0010] A method for short circuit identification and prediction of a battery circuit, comprising the steps of:

[0011] S1. For each battery cell of the series lithium battery system, establish an equivalent circuit model including a short circuit equivalent resistance;

[0012] S2. According to the circuit parameters of the equivalent circuit model, in each detection period, for each battery cell, estimate the SOC through the Kalman filter algorithm to obtain a first SOC data set, and estimate the SOC through the ampere-hour integration method to obtain a second SOC data set;

[0013] S3. Calculate a short circuit resistance data set according to the first SOC data set and the second SOC data set;

[0014] S4. Calculate the deviation between the short circuit resistance of each battery cell and the average short circuit resistance according to the short circuit resistance data set, and determine whether there is a fault in the short circuit resistance of the single cell;

[0015] S5. Calculate the average heat generation power of each battery cell according to the circuit parameters of the equivalent circuit model and the short circuit resistance data set, obtain the maximum heat dissipation power of the battery cell, and predict the time when the heat dissipation capacity limit of each battery cell in the system is reached.

[0016] To solve the above technical problems, the technical solution adopted by the present invention is:

[0017] A terminal for short - circuit identification and prediction of a battery circuit, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above - mentioned method for short - circuit identification and prediction of a battery circuit are implemented.

[0018] The beneficial effects of the present invention are as follows: The method and terminal for short - circuit identification and prediction of a battery circuit of the present invention simplify the computational complexity by using circuit - level modeling. By combining the Kalman filter and the ampere - hour integration method to estimate the SOC difference, the indirect calculation of the short - circuit resistance is realized, without relying on complex electrochemical models or historical fault data; the capacity loss of the short - circuit current is directly reflected by the SOC difference, providing a quantitative basis for short - circuit detection; at the same time, by combining heat generation prediction and heat dissipation capacity analysis, the risk of thermal runaway can be pre - warned, solving the problem of detection delay caused by insufficient data or complex models in traditional methods. Description of the Drawings

[0019] Figure 1 It is an example diagram of a second - order equivalent circuit of a method for short - circuit identification and prediction of a battery circuit according to an embodiment of the present invention;

[0020] Figure 2 It is a brief flow example diagram of a method for short - circuit identification and prediction of a battery circuit according to an embodiment of the present invention;

[0021] Figure 3 It is an example diagram of curve fitting of the average heat generation power of a battery cell in a method for short - circuit identification and prediction of a battery circuit according to an embodiment of the present invention;

[0022] Figure 4 It is a structural diagram of a terminal for short - circuit identification and prediction of a battery circuit according to an embodiment of the present invention;

[0023] Label Description:

[0024] 1. A terminal for short - circuit identification and prediction of a battery circuit; 2. Processor; 3. Memory. Detailed Embodiments

[0025] To describe the technical content, achieved objectives, and effects of the present invention in detail, the following is described in conjunction with the embodiments and with reference to the drawings.

[0026] Please refer to Figures 1 to 3 , a method for short - circuit identification and prediction of a battery circuit, comprising the steps:

[0027] S1. For each battery cell in the series - connected lithium - battery system, establish an equivalent - circuit model including a short - circuit equivalent resistance;

[0028] S2. According to the circuit parameters of the equivalent circuit model, in each detection cycle, for each battery cell, SOC estimation is performed by using a Kalman family algorithm to obtain a first SOC data set, and SOC estimation is performed by using an ampere-hour integration method to obtain a second SOC data set;

[0029] S3, calculating a short-circuit resistance data set according to the first SOC data set and the second SOC data set;

[0030] S4. Calculate the deviation between the short-circuit resistance of each battery cell and the short-circuit resistance mean value according to the short-circuit resistance data set, and determine whether there is a fault in the short-circuit resistance of the cell;

[0031] S5. Calculate the average heat generation power of each battery cell according to the circuit parameters of the equivalent circuit model and the short-circuit resistance data set, obtain the maximum heat dissipation power of the battery cell, and predict the time to reach the heat dissipation capacity limit of the system for each battery cell.

[0032] From the above description, it can be seen that the beneficial effects of the present invention are: a method for short-circuit identification and prediction of a battery circuit of the present invention uses circuit-level modeling to simplify calculation complexity, estimates SOC differences by combining Kalman filtering and the ampere-hour integration method, and realizes indirect calculation of short-circuit resistance without relying on complex electrochemical models or historical fault data; directly reflects the capacity loss of short-circuit current through the SOC difference, providing a quantitative basis for short-circuit detection; at the same time, combined with heat generation prediction and heat dissipation capacity analysis, it can provide early warning of thermal runaway risks, solving the detection delay problem caused by insufficient data or complex models in traditional methods.

[0033] Further, step S4 comprises the steps of:

[0034] According to the short-circuit resistance data set, the short-circuit resistance of each battery cell obtained in the current detection cycle is obtained, and variance and mean calculation is performed;

[0035] Calculate the adaptive alarm threshold value based on the calculated variance and mean;

[0036] It is determined whether the short-circuit resistance of a single cell exceeds the adaptive alarm threshold range in the current detection, and if so, it is determined that the single cell is faulty.

[0037] From the above description, it can be seen that by calculating the mean and variance of the short-circuit resistance to set the adaptive alarm threshold, the interference of individual differences in the system and sampling noise can be effectively filtered out. Compared with the fixed threshold method, this adaptive mechanism can dynamically adapt to changes in battery pack consistency, reduce the probability of false alarms, and improve the accuracy of fault detection. Combined with the sliding window detection strategy, it can capture abnormal fluctuations in short-circuit resistance in real time and enhance the system's ability to respond to gradual faults.

[0038] Further, the adaptive alarm threshold includes a lower threshold;

[0039] The lower threshold is specifically calculated as follows:

[0040] ;

[0041] where and respectively represent the mean and variance of the short - circuit resistance of each battery cell obtained within the current detection period.

[0042] As can be seen from the above description, the calculation formula for the lower threshold (mean minus 3 times the standard deviation) is clarified, and a scientific fault determination criterion is established based on statistical principles. This threshold not only considers the resistance fluctuation range during normal operation but also retains a sufficient safety margin to ensure reliable alarm triggering when the short - circuit resistance drops significantly. Compared with traditional empirical thresholds, it has stronger universality and robustness, especially suitable for application scenarios with high discreteness of battery parameters.

[0043] Further, step S5 includes the steps of:

[0044] S51. For each detection period, calculate the average heat generation power of each cell according to the circuit parameters of the equivalent circuit model and the short - circuit resistance data set;

[0045] S52. Calculate the maximum heat dissipation power of each battery cell according to the total refrigeration power of the cooling system for the battery;

[0046] S53. Combine the average heat generation power calculated in several previous detection periods to construct an average heat generation power data set for each battery cell;

[0047] S54. Perform curve fitting on the average heat generation power data set, and according to the fitting result, judge the number of detection periods required to reach the maximum heat dissipation power, and predict the time to reach the heat dissipation capacity limit of each cell in the system.

[0048] As can be seen from the above description, predicting the thermal runaway time by fitting the heat generation power data breaks through the lag of traditional temperature monitoring. This method calculates the heat generation in real - time based on equivalent circuit parameters and short - circuit resistance, and combines the evaluation of the cooling system capacity to predict the trend of heat accumulation in advance. The prediction model considers the evolution law of historical heat generation data, reserves sufficient fault response time for the system, and significantly improves the safety and operation and maintenance efficiency of the energy storage system.

[0049] Further, the calculation of the average heat generation power is specifically as follows:

[0050] ;

[0051] Among them, and represent the q - order equivalent circuit parameters of battery cell i, is the rated current of the battery, is the equivalent average short - circuit current of cell i at time t in the current detection cycle, and are the sampled voltages of battery cell i at the initial time and time t in the current detection cycle respectively, represents the equivalent short - circuit resistance of battery cell i during the period from the initial time to time t in the current detection cycle.

[0052] As can be seen from the above description, the heat generation power formula comprehensively considers the equivalent circuit ohmic loss and short - circuit resistance loss, and accurately quantifies the internal energy dissipation of the battery. Compared with the method that only relies on external temperature measurement, this formula calculates through multi - parameter coupling and can more truly reflect the abnormal heat generation caused by internal short - circuit. Especially in the early stage, when the temperature has not risen significantly, potential risks can be identified through power analysis, realizing early warning of faults.

[0053] Furthermore, the calculation of the maximum heat dissipation power is specifically as follows:

[0054] ;

[0055] Among them, is the total refrigeration power of the cooling system for the battery, and m is the number of battery cells of the battery.

[0056] As can be seen from the above description, the total system refrigeration power is evenly distributed to each single - cell battery to establish a standardized heat dissipation capacity evaluation model. This method simplifies the design complexity of the thermal management system and ensures the safety evaluation of each single - cell under the same heat dissipation conditions. By directly comparing the heat generation and heat dissipation capacity, the cells with insufficient heat dissipation can be quickly located, providing a clear basis for priority ranking for thermal runaway prevention.

[0057] Furthermore, step S3 is specifically as follows:

[0058] According to the first SOC data set and the second SOC data set, calculate the short - circuit resistance:

[0059] ;

[0060] ;

[0061] Record the short - circuit resistance of the i - th battery cell in the j - th detection cycle as , and construct a short - circuit resistance data set;

[0062] Among them, Represents the equivalent average short-circuit current of battery cell i at time t during the current detection period. and respectively represent the sampled voltages of battery cell i at the initial time and time t during the current detection period. Represents the capacity of battery cell i at time t. Represents the SOC value of battery cell i at time t estimated by the ampere-hour integration method during the current detection period. Represents the SOC value of battery cell i at time t estimated by the Kalman filter algorithm during the current detection period.

[0063] As can be seen from the above description, the short-circuit resistance calculation formula converts the non-directly measurable short-circuit current into a computable parameter through the mathematical relationship between the SOC difference and the sampling time. This formula is based on the principle of capacity conservation, quantifies the short-circuit current using the SOC difference between ampere-hour integration and Kalman filtering, and avoids the technical difficulties of directly measuring high internal resistance short circuits.

[0064] Furthermore, the estimation of the first SOC dataset includes the steps of:

[0065] Construct a discretized state equation based on the circuit parameters of the q-th order equivalent circuit of the battery cell:

[0066] ;

[0067] where Q c Represents the single-cell capacity, , , and represent the equivalent circuit resistance and capacitance at temperature T and state of charge SOC, SOC t 、U1 t , Uq t and SOC t+1 、U1 t+1 and Uq t+1 respectively represent the SOC state of the battery cell at time t and t + 1 and the voltage states of the equivalent capacitances C1 and Cq, I t is the single-cell current at time t, T s is the calculation step, and W is the process noise;

[0068] Construct a measurement equation for the single cell:

[0069] ;

[0070] Among them, e t+1 represents the voltage measured at the output terminal of the monomer at time t + 1, U OCV is the open-circuit voltage of the monomer obtained, I t+1 is the monomer current at time t + 1, R 0_t+1 is the DC internal resistance of the monomer at time t + 1, and V is the measurement noise;

[0071] Use the UKF algorithm or the EKF algorithm to calculate the estimated value of the i-th of all m monomers to be evaluated at time t SOC i_t , denoted as , and form a state-of-charge set to obtain the first data set :

[0072] .

[0073] As can be seen from the above description, the Kalman filter algorithm is used to process the state estimation problem of a nonlinear system, and the SOC estimation value is dynamically corrected through the state equation and the measurement equation. Compared with the traditional OCV curve method, this method can effectively suppress the influence of measurement noise and model error, and still maintain high accuracy in the case of time-varying battery parameters (such as temperature, aging). Experiments show that the SOC estimation error can be controlled within 0.05%, laying a reliable foundation for the subsequent short-circuit resistance calculation.

[0074] Furthermore, the estimation of the second SOC data set includes the steps of:

[0075] Estimate the state of charge of all battery monomers using the ampere-hour integration method:

[0076] ;

[0077] Among them, and are respectively the state of charge of battery monomer i at time t - 1 within the detection period and the state of charge at the initial moment of the detection period, is the charging or discharging efficiency of battery monomer i at time t, is the average charging or discharging current of battery monomer i from t - 1 to t, is the duration from t - 1 to t, is the rated capacity of battery monomer i;

[0078] According to the state of charge of m battery monomers, obtain the second SOC data set :

[0079] ;

[0080] Among them, at the initial moment of the current detection period, the SOC calculated by the ampere-hour integration method and the SOC calculated by using the Kalman filter algorithm have the same initial value.

[0081] As can be seen from the above description, the ampere-hour integration method synchronizes with the Kalman filter result through initial state alignment, eliminating the initial deviation of SOC estimation. This design ensures that the SOC difference between the two methods truly reflects the capacity loss of the short-circuit current, avoiding misjudgment caused by initial errors. Combining dynamic charge and discharge efficiency compensation makes the ampere-hour integration result closer to the actual operating conditions and enhances the accuracy of short-circuit resistance calculation.

[0082] Please refer to Figure 4 , a terminal for short-circuit identification and prediction of a battery circuit, including a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned method for short-circuit identification and prediction of a battery circuit are implemented.

[0083] As can be seen from the above description, the beneficial effects of the present invention are as follows: The terminal for short-circuit identification and prediction of a battery circuit of the present invention simplifies the calculation complexity by using circuit-level modeling, realizes the indirect calculation of short-circuit resistance by combining the Kalman filter and the ampere-hour integration method to estimate the SOC difference, and does not rely on complex electrochemical models or historical fault data; the capacity loss of the short-circuit current is directly reflected by the SOC difference, providing a quantitative basis for short-circuit detection; at the same time, by combining heat generation prediction and heat dissipation capacity analysis, the risk of thermal runaway can be warned in advance, solving the detection delay problem caused by insufficient data or complex models in traditional methods.

[0084] The method and terminal for short-circuit identification and prediction of a battery circuit of the present invention are applicable to the state detection and warning of lithium batteries.

[0085] Please refer to Figures 1 to 3 , the first embodiment of the present invention is:

[0086] A method for short-circuit identification and prediction of a battery circuit, including the steps of:

[0087] S1. For each battery cell in the series-connected lithium battery system, establish an equivalent circuit model including a short-circuit equivalent resistance.

[0088] In this embodiment, for a lithium battery system with multiple series-connected battery cells, a power supply network composed of m series-connected battery cells is established. For each cell, a first-order or second-order equivalent circuit model containing a short-circuit equivalent resistance is established, and the short-circuit current Is, charging or discharging current Io, and output voltage U are defined. In this embodiment, the second-order equivalent circuit is taken as an example, and reference can be made to Figure 1 .

[0089] S2. According to the circuit parameters of the equivalent circuit model, within each detection period, for each battery cell, the first SOC data set is obtained through SOC estimation using the Kalman filter algorithm, and the second SOC data set is obtained through SOC estimation using the ampere-hour integration method;

[0090] The estimation of the first SOC data set includes the steps:

[0091] Based on the circuit parameters of the q - order equivalent circuit of the battery cell, construct a discretized state equation:

[0092] ;

[0093] Where, Q c represents the cell capacity, 、 、 and represent the equivalent circuit resistance and capacitance at temperature T and state of charge SOC, SOC t 、U1 t 、 Uq t and SOC t+1 、U1 t+1 and Uq t+1 respectively represent the SOC state of the battery cell at time t and t + 1, and the voltage states of the equivalent capacitances C1 and Cq, I t is the cell current at time t, T s is the calculation step, and W is the process noise;

[0094] Construct a measurement equation for the cell:

[0095] ;

[0096] Where, e t+1 represents the voltage measured at the output terminal of the cell at time t + 1, U OCV is the open - circuit voltage of the cell obtained, I t+1 is the cell current at time t + 1, R 0_t+1 is the DC internal resistance of the cell at time t + 1, and V is the measurement noise;

[0097] Use the UKF algorithm or the EKF algorithm to calculate the SOC i_t SOC i_t at time t for the i - th of all m cells to be evaluatedThe estimated value is denoted as and forms a set of state of charge, obtaining the first data set :

[0098] .

[0099] In this embodiment, the extended Kalman filter (EKF) or unscented Kalman filter (UKF) method is used to estimate the state of charge set of all cells as follows:

[0100] Estimate the circuit parameters of each cell offline at different temperatures. For the first-order equivalent circuit, it includes R0, R1, C1, and for the second-order equivalent circuit, it includes R0, R1, C1, R2, C2. The specific estimation method can be found in "Parameter Identification Method for 1st and 2nd Order Thevenin Equivalent Circuits".

[0101] Obtain the correspondence function between the open-circuit voltage and the state of charge SOC of the cell at different temperatures U OCV =f T (SOC) The curve can be obtained in Section 2 of "Battery SOC Estimation Method and Energy Storage System".

[0102] Establish the state equation and measurement equation of the battery cell:

[0103] Establish a first-order, second-order or other-order equivalent circuit for the battery cell; construct a discretized state equation according to the initial values of the cell parameters R 1_(T,SOC) 、C 1_(T,SOC)、 R 2_(T,SOC) 、C 2_(T,SOC ) For example, with the second-order equivalent circuit, the discretized state equation is as follows: C 1_(T,SOC) The voltage across U1 is C 2_(T,SOC) The voltage across U2 is (T,SOC) The subscript of the parameter indicates that its value is related to the cell temperature T and SOC.

[0104] Taking the second-order equivalent circuit as an example, the discretized state equation is as follows:

[0105] ;

[0106] where Q c is the cell capacity, R 1_(T,SOC) 、C 1_(T,SOC) , R 2_(T,SOC)、C 2_(T,SOC) are the equivalent circuit resistance and capacitance at a certain temperature T and SOC. SOC t 、U1 t and U2 t and SOC t+1 、U1 t+1 、U2 t+1 are the state of charge (SOC) of the single cell and the voltage states of the equivalent capacitances C1 and C2 at times t and t + 1 respectively. I t is the current of the single cell at time t. T s is the calculation step size, and W is the process noise.

[0107] The measurable state variables of the single cell are voltage, current, and temperature. The measurement equation for the single cell is constructed as follows:

[0108]

[0109] where e t+1 represents the voltage measured at the output terminal of the single cell at time t + 1. U OCV is the open-circuit voltage of the single cell obtained. ，I t+1 is the current of the single cell at time t + 1. R 0_t+1 is the DC internal resistance of the single cell at time t + 1, and V is the measurement noise.

[0110] Use the UKF algorithm or the EKF algorithm to calculate the estimated value of the i-th of all m single cells to be evaluated at time t, denoted as SOC i_t , and form the state of charge set : :

[0111] .

[0112] The estimation of the second SOC dataset includes the steps of:

[0113] Estimate the state of charge of all battery single cells using the ampere-hour integration method:

[0114] ;

[0115] where and are the state of charge of battery single cell i at time t - 1 within the detection period and the state of charge at the initial moment of the detection period respectively. is the charging or discharging efficiency of battery cell i at time t, is the average charging or discharging current of battery cell i from time t-1 to t, is the duration from time t-1 to t, is the rated capacity of battery cell i;

[0116] According to the state of charge of m battery cells, a second SOC dataset is obtained :

[0117] ;

[0118] Among them, at the initial moment of the current detection period, the SOC calculated by the ampere-hour integration method and the SOC calculated by the Kalman family algorithm have the same initial value.

[0119] In this embodiment, for the in-use lithium battery system, the ampere-hour integration method is used to estimate the state of charge set of all battery cells :

[0120] Calculate the at time t of battery cell i within a detection period according to the following method :

[0121] ;

[0122] Among them and are the state of charge of battery cell i estimated by the ampere-hour integration method at time t-1 and the state of charge at the initial moment of the detection period respectively; is the charging or discharging efficiency of battery cell i at time t; is the average charging or discharging current of battery cell i from time t-1 to t (positive for charging and negative for discharging); is the duration from time t-1 to t; is the rated capacity of battery cell i.

[0123] .

[0124] At the initial moment of the detection period, align the values of each battery cell within and , that is, make , so that the SOC calculated by the ampere-hour integration method and the SOC calculated by the KF family method have the same initial value.

[0125] Among them, is the SOC of battery cell i at the initial moment of the detection period by the ampere-hour integration method, is the SOC of battery cell i at the initial moment of the detection period by the Kalman method.

[0126] S3. Calculate the short - circuit resistance dataset according to the first SOC dataset and the second SOC dataset;

[0127] Step S3 is specifically as follows:

[0128] Calculate the short - circuit resistance according to the first SOC dataset and the second SOC dataset:

[0129] ;

[0130] ;

[0131] Denote the short - circuit resistance of the \(i\) - th battery cell in the \(j\) - th detection period as , and construct the short - circuit resistance dataset;

[0132] Wherein, represents the equivalent average short - circuit current of the \(i\) - th battery cell at time \(t\) within the current detection period, and respectively represent the sampled voltages of the \(i\) - th battery cell at the initial time and time \(t\) within the current detection period, represents the capacity of the \(i\) - th battery cell at time \(t\).

[0133] Calculate the equivalent short - circuit resistance of the \(i\) - th battery cell during the period from the initial time to time \(t\) in a certain detection period according to the above formula. For the \(j\) - th detection period, \(j\in(1,n)\), where \(n\) is the number of detection periods, the equivalent short - circuit resistance is denoted as .

[0134] The meaning of the formula is that within the sampling period from 0 to \(t\) in a certain detection period, and The difference in measurement represents the capacity loss accumulated by the short - circuit current during this period, and the result is the average value of the equivalent short - circuit current.

[0135] In this embodiment, calculate the calculated at time \(t\) in each of the consecutive \(n\) detection periods, :

[0136] .

[0137] S4. Calculate the deviation between the short - circuit resistance of each battery cell and the average short - circuit resistance according to the short - circuit resistance dataset, and determine whether there is a fault in the short - circuit resistance of the battery cell;

[0138] Step S4 includes the steps:

[0139] According to the short-circuit resistance data set, obtain the short-circuit resistance of each battery cell obtained during the current detection period, and perform variance and mean calculations;

[0140] Calculate the adaptive alarm threshold according to the calculated variance and mean;

[0141] The adaptive alarm threshold includes a lower threshold;

[0142] The lower threshold is specifically calculated as:

[0143] ;

[0144] ;

[0145] ;

[0146] wherein, and respectively represent the mean and variance of the short-circuit resistance of each battery cell obtained during the current detection period;

[0147] Determine whether the short-circuit resistance of a single cell exceeds the range of the adaptive alarm threshold in the current detection. If so, it is determined that the single cell has a fault.

[0148] In this embodiment, for the lithium battery system to be evaluated during operation, when the < of the i-th single cell, an alarm for abnormal decrease in short-circuit internal resistance is issued:

[0149] (1) represents the fluctuation degree of the short-circuit resistance deviation. When it exceeds the limit in the j-th detection period, an alarm is issued at the end of the period.

[0150] (2) The detection period is t, but the start interval between two adjacent detection periods can be less than t, that is, the sliding window method is used for detection.

[0151] S5. According to the circuit parameters of the equivalent circuit model and the short-circuit resistance data set, calculate the average heat generation power of each battery cell, obtain the maximum heat dissipation power of the battery cell, and predict the time when the heat dissipation capacity limit of each cell in the system is reached;

[0152] Step S5 includes the steps of:

[0153] S51. For each detection period, calculate the average heat generation power of each single cell according to the circuit parameters of the equivalent circuit model and the short-circuit resistance data set;

[0154] The average heat generation power is specifically calculated as:

[0155] ;

[0156] Among them, and represent the q - order equivalent circuit parameters of battery cell i, is the rated current of the battery, is the equivalent average short - circuit current of cell i at time t in the current detection cycle, and are the sampled voltages of battery cell i at the initial time and time t in the current detection cycle respectively;

[0157] S52. Calculate the maximum heat dissipation power of each battery cell according to the total refrigeration power of the cooling system for the battery;

[0158] The calculation of the maximum heat dissipation power is specifically as follows:

[0159] ;

[0160] Among them, is the total refrigeration power of the cooling system for the battery, and m is the number of battery cells of the battery;

[0161] S53. Combine the average heat generation power calculated in several previous detection cycles to construct an average heat generation power data set for each battery cell;

[0162] S54. Perform curve fitting on the average heat generation power data set, and according to the fitting result, judge the number of detection cycles required to reach the maximum heat dissipation power, and predict the time to reach the heat dissipation capacity limit of the system for each battery cell.

[0163] In this embodiment, calculate the heat generation of each cell, evaluate whether the heat generation limit caused by the increase in the short - circuit resistance of the cell exceeds the thermal management ability of the battery system; and according to the estimated short - circuit resistance values obtained in several detection cycles, predict the time to reach the heat dissipation capacity limit of the system for each battery cell, and give a long - term short - circuit fault time prediction and alarm.

[0164] Specifically, according to the obtained resistance R0, R1 parameters (first - order equivalent circuit), or R0, R1, R2 parameters (second - order equivalent circuit), calculate the average heat generation power of each cell within the detection cycle .

[0165] Taking the second - order equivalent circuit as an example,

[0166] ;

[0167] Among them, are the second - order equivalent circuit parameters of cell i, is the battery rated current.

[0168] Calculate the maximum heat dissipation power of each single cell according to the capacity of the battery thermal management system . Assuming that the heat dissipation power of each single cell is equal, then:

[0169] ;

[0170] is the total refrigeration power of the cooling system for the battery.

[0171] Calculate for n consecutive detection cycles , and obtain the average heat generation power evaluation data set of single cell i .

[0172] ;

[0173] For The data sequence formed is fitted, and reference can be made to Figure 3 , and obtain the number of detection cycles x required to reach ; when x*y < z, trigger the long-cycle short-circuit fault time prediction alarm. Where y is the starting point interval time between two adjacent detection cycles, and z is the early warning time limit.

[0174] Effect verification:

[0175] For the calculation in the method Carry out simulation verification:

[0176] Connect a parallel resistor at the output position in the single cell battery model to simulate the occurrence of a short circuit. Set the resistor to 10 ohms and the battery capacity to 280 Ah. Simulate the difference between the estimated_SOC and the ampere-hour integrated SOC. During the detection cycle with a total duration of 200000 s (t = 200000 s), is 0.0525.

[0177] 0.0525 * 280 / 200000 * 3600 = 0.265 A;

[0178] ;

[0179] The error is 22.6%.

[0180] Adjust the short-circuit resistance to 5 ohms, 0.103 * 280 / 200000 * 3600 = 0.5192 A;

[0181] 6.26 ohms, the error is 25%.

[0182] Adjust the short - circuit resistance to 20 ohms. 0.0295 * 280 / 200000 * 3600 = 0.149 A;

[0183] 21.82 ohms, with an error of 9%.

[0184] The actual additional heat generation of each cell under the condition of three resistors is 1.05 W, 2.11 W, and 0.52 W respectively, all of which are relatively small and cannot be quickly detected by temperature detection. The method plays a role in early detection.

[0185] The error generated comes from the acquisition accuracy, the estimation errors of the Kalman filter algorithm and the ampere - hour integration algorithm, which are in the order of several ohms to dozens of ohms. The method can effectively detect the value of the short - circuit resistance.

[0186] Please refer to Figure 4 , the second embodiment of the present invention is:

[0187] A terminal 1 for short - circuit identification and prediction of a battery circuit, including a processor 2, a memory 3, and a computer program stored in the memory 3 and operable on the processor 2. When the processor 2 executes the computer program, it implements the steps in the method for short - circuit identification and prediction of a battery circuit described in the above - mentioned first embodiment.

[0188] To sum up, the method and terminal for short - circuit identification and prediction of a battery circuit provided by the present invention use circuit - level modeling to simplify the calculation complexity, estimate the SOC difference by combining the Kalman filter and the ampere - hour integration method to indirectly calculate the short - circuit resistance, without relying on complex electrochemical models or historical fault data; directly reflect the capacity loss of the short - circuit current through the SOC difference to provide a quantitative basis for short - circuit detection; at the same time, combine heat generation prediction and heat dissipation capacity analysis to early - warning the risk of thermal runaway and solve the detection delay problem caused by insufficient data or complex models in traditional methods.

[0189] 1. In the present invention, the Kalman filter method for dealing with nonlinear problems is adopted to identify the equivalent circuit of the battery cell containing the short - circuit resistance as the battery capacity loss, and the SOC estimation accuracy is not affected by the change of the short - circuit resistance.

[0190] 2. The SOC estimation method of ampere - hour integration cannot identify the current consumed in the internal loop of the battery cell. When there is a short - circuit, the SOC value is higher than the identification result of the UKF method; under the condition of initial state alignment in each evaluation period, the evolution of the deviation can reflect the capacity consumption of the short - circuit current, thus reflecting the magnitude of the short - circuit current.

[0191] 3. The present invention gives the calculation result of the short - circuit resistance, which is convenient to accumulate the data of the short - circuit resistance during the system operation and lays a foundation for the data - based evaluation method.

[0192] 4. The adaptive short-circuit resistance determination threshold given by the present invention in combination with the gradual change of the system's own operating state has an adaptive filtering effect and can filter out the errors introduced by individual systems.

[0193] 5. According to the thermal runaway prevention mechanism of the present invention, the internal short circuit of the battery is a gradual and irreversible process. When the heat generation of a single cell is greater than the heat dissipation, the temperature of the battery cell will continue to rise until thermal runaway occurs. This method does not rely on temperature acquisition and analysis, but calculates the heat generation to achieve the prediction function, providing a new evaluation basis and direction for the judgment of the early state of the internal short circuit.

[0194] The above are only the embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in the relevant technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A method for identifying and predicting short circuits in a battery circuit, characterized in that: Includes steps: S1. For each battery cell in the series-connected lithium battery system, an equivalent circuit model including a short-circuit equivalent resistance is established; S2. According to the circuit parameters of the equivalent circuit model, in each detection cycle, for each battery cell, SOC estimation is performed by using a Kalman family algorithm to obtain a first SOC data set, and SOC estimation is performed by using an ampere-hour integration method to obtain a second SOC data set; S3, calculating a short-circuit resistance data set according to the first SOC data set and the second SOC data set; Step S3 is specifically as follows: According to the first SOC data set and the second SOC data set, the short-circuit resistance is calculated: ； ； The short-circuit resistance of the i-th battery cell in the j-th detection cycle is recorded as , construct a short-circuit resistance dataset; in, It represents the equivalent short-circuit resistance of battery cell i from the initial moment to moment t in a certain detection cycle. It represents the equivalent average short-circuit current of battery cell i at time t in the current detection cycle, and They represent the sampled voltages of battery cell i at the initial moment and at moment t in the current detection cycle, respectively. represents the rated capacity of battery cell i, It represents the SOC value of battery cell i at time t estimated by the ampere-hour integration method in the current detection cycle. It represents the SOC value of battery cell i at time t estimated by the Kalman family algorithm in the current detection cycle; S4. Calculate the deviation between the short-circuit resistance of each battery cell and the short-circuit resistance mean value according to the short-circuit resistance data set, and determine whether there is a fault in the short-circuit resistance of the cell; S5. Calculate the average heat generation power of each battery cell according to the circuit parameters of the equivalent circuit model and the short-circuit resistance data set, obtain the maximum heat dissipation power of the battery cell, and predict the time to reach the heat dissipation capacity limit of the system for each battery cell.

2. The method for short circuit identification and prediction of a battery circuit according to claim 1, characterized in that: Step S4 comprises the steps of: According to the short-circuit resistance data set, the short-circuit resistance of each battery cell obtained in the current detection cycle is obtained, and variance and mean calculation is performed; Calculate the adaptive alarm threshold according to the calculated variance and mean; It is determined whether the short-circuit resistance of a single cell exceeds the adaptive alarm threshold range in the current detection, and if so, it is determined that the single cell is faulty.

3. The method for short circuit identification and prediction of a battery circuit according to claim 2, characterized in that: The adaptive alarm threshold includes a lower threshold; The lower threshold The calculation is as follows: ； in, and They respectively represent the mean and variance of the short-circuit resistance of each battery cell obtained in the current detection cycle.

4. The method for short circuit identification and prediction of a battery circuit according to claim 1, characterized in that: Step S5 comprises the steps of: S51, for each detection cycle, calculating the average heat generation power of each monomer according to the circuit parameters of the equivalent circuit model and the short-circuit resistance data set; S52, calculating the maximum heat dissipation power of each battery cell according to the total cooling power of the cooling system for the battery; S53, constructing an average heat generation power data set for each battery cell based on the average heat generation power calculated in several previous detection cycles; S54, performing curve fitting on the average heat generation power data set, determining the number of detection cycles required to reach the maximum heat dissipation power according to the fitting result, and predicting the time to reach the heat dissipation capacity limit of the system for each battery cell.

5. The method for short circuit identification and prediction of a battery circuit according to claim 4, characterized in that: The average heat generation power The calculation is as follows: ； in, and represents the q-order equivalent circuit parameters of battery cell i, is the battery rated current, is the equivalent average short-circuit current of monomer i at time t in the current detection cycle, and are the sampled voltages of battery cell i at the initial moment and at moment t in the current detection cycle, It represents the equivalent short-circuit resistance of battery cell i from the initial moment to moment t in the current detection cycle.

6. A method for short circuit identification and prediction of a battery circuit according to claim 4, characterized in that: The maximum heat dissipation power The calculation is as follows: ； in, is the total cooling power of the cooling system for the battery, and m is the number of battery cells.

7. A method for short circuit identification and prediction of a battery circuit according to claim 1, characterized in that: The estimation of the first SOC data set comprises the steps of: According to the circuit parameters of the q-order equivalent circuit of the battery cell, the discretized state equation is constructed: ； in, Q c Represents the monomer capacity, , , as well as Represents the equivalent circuit resistance and capacitance at temperature T and state of charge SOC, SOC t 、U1 t , Uq t and SOC t+1 、U1 t+1 , Uq t+1 Respectively represent the SOC state of the battery cell at time t and time t+1 and the voltage state of the equivalent capacitor C1 and Cq, I t is the monomer current at time t, T s is the calculation step size, W is the process noise; Construct the measurement equation for the monomer: ； in, e t+1 It indicates the voltage measured at the output end of the monomer at time t+1. U2 t+1 represents the voltage state of the equivalent capacitor C2 at time t+1, U OCV is the obtained single cell open circuit voltage, U OCV_t+1 represents the single cell open circuit voltage at time t+1, I t+1 is the monomer current at time t+1, R 0_t+1 is the DC internal resistance of the monomer at time t+1, and V is the measurement noise; Use the UKF algorithm or the EKF algorithm to calculate the i-th of all m monomers being evaluated at time t SOC i_t The estimated value of , and form a state of charge set to obtain the first data set : 。 8. The method for short circuit identification and prediction of a battery circuit according to claim 1, characterized in that: The estimation of the second SOC data set comprises the steps of: The state of charge of all battery cells is estimated using the ampere-hour integration method: ； in, and are the state of charge of battery cell i at time t-1 in the detection cycle and the state of charge at the initial moment of the detection cycle, is the charging or discharging efficiency of battery cell i at time t, is the average charge or discharge current of battery cell i from time t-1 to t, is the duration from t-1 to time t, is the rated capacity of battery cell i; According to the state of charge of m battery cells, the second SOC data set is obtained : ； Among them, at the initial moment of the current detection cycle, the SOC calculated by the ampere-hour integration method and the SOC calculated by the Kalman family algorithm have the same initial value.

9. A terminal for identifying and predicting short circuits in a battery circuit, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps in the method for short circuit identification and prediction of a battery circuit as described in any one of claims 1 to 8 are implemented.

Images