A Fault Detection Method for Battery System

Abstract

The present invention discloses a fault detection method for a battery system, and relates to the field of batteries. The present invention introduces a current sensor to measure the balancing current between two adjacent battery groups, and based on the dynamic behavior characteristics that the time derivative of the balancing current changes continuously, uses each real-time time series data to obtain the adjacent battery group with faults and the corresponding fault type; for the adjacent battery group with faults, the upper and lower limits of the corresponding change rate threshold range are used to determine the corresponding fault degree of the battery group, that is, based on the dynamic behavior characteristics, the present invention uses the balancing current detected by a current detection device to obtain the adjacent battery group with faults and the corresponding fault type, and determines the corresponding fault degree of the battery group based on the upper and lower limits of the change rate threshold range, thereby effectively identifying the abnormal current fluctuation caused by the fault and improving the accuracy of fault detection.

Description

Technical Field

[0001] The present invention relates to the field of batteries, and in particular to a fault detection method for a battery system. Background Art

[0002] Under the dual pressures of energy crisis and environmental pollution, clean and economical energy solutions and efficient energy storage technologies have become the focus of research and development in today's society. Lithium-ion batteries have occupied an important position in the field of energy storage due to their advantages such as high energy density and long cycle life. However, lithium-ion batteries may encounter a variety of abnormal conditions during their use cycle, such as mechanical damage, internal short circuit failure, overcharge or over-discharge. These problems will not only cause the battery performance to gradually decline, but may also cause thermal runaway, and ultimately cause safety risks such as explosion or spontaneous combustion. Therefore, it is crucial to detect and identify these abnormal conditions in a timely manner at an early stage so that preventive measures can be taken to avoid potential safety accidents.

[0003] Abnormal battery conditions usually result from internal microstructural defects, changes in external environmental conditions, or accumulated damage from long-term use. The characteristic signals of these abnormal conditions, such as temperature changes, voltage fluctuations, or current anomalies, are relatively weak in the early stages and are easily masked by external noise. Traditional monitoring methods, such as thermal monitoring and voltage analysis, are not sensitive enough to capture these early abnormal signals, making it difficult to achieve effective fault warnings. Although there are some monitoring systems for battery safety on the market, they mainly focus on over-temperature protection and over-current protection. For more complex problems such as internal short-circuit faults, existing patented technologies are still relatively scarce. In particular, the technology of using the current balance characteristics between internal parallel battery packs to achieve real-time diagnosis is still in its infancy in terms of research and application, which provides a broad space for innovation.

[0004] In order to improve the safety and reliability of power battery packs, it is necessary to develop advanced monitoring technologies that can adapt to multi-level battery architectures, especially to accurately locate battery pack failures in large-scale parallel battery packs. Current monitoring technologies have obvious shortcomings in this regard, especially when dealing with complex large-scale battery systems. Therefore, it is necessary to explore new diagnostic methods based on fluctuations in the balancing current signal, because such fluctuations are often an important indicator of abnormalities inside the battery. Through further research and technological innovation, the shortcomings of existing technologies can be made up and the ability to detect abnormal battery conditions can be improved, thereby ensuring the safe operation of energy storage systems and promoting the application and development of clean energy. Summary of the invention

[0005] In order to accurately detect battery pack faults based on balancing current signal fluctuations, the present invention proposes a fault detection method for a battery system, which realizes fault detection based on a current detection device; the battery system includes N parallel battery packs, N is a positive integer greater than or equal to 2; the current detection device includes: N-1 current sensors, each current sensor measures the balancing current between two adjacent battery packs; the balancing current is the current difference flowing through the two parallel battery packs;

[0006] The fault detection method comprises:

[0007] By using the voltage, open circuit voltage and internal resistance of two standard battery packs in parallel and in normal working state, the time-varying characteristics of the balancing current are analyzed, and the dynamic behavior characteristics of the time derivative of the balancing current showing continuous changes are obtained;

[0008] The balancing current between two adjacent battery groups in the battery system is acquired in real time by a current detection device, and real-time time series data corresponding to the current sensor is obtained; the real-time time series data includes the balancing current corresponding to each time point and the time derivative, that is, the rate of change of the balancing current with time;

[0009] Based on the acquired dynamic behavior characteristics, the adjacent battery groups with faults and their corresponding fault types are obtained using each real-time time series data;

[0010] Initialize the change rate threshold range of each adjacent battery group, and update the change rate threshold range of the corresponding adjacent battery group in real time through real-time time series data. For adjacent battery groups with faults, use the upper and lower limits of the corresponding change rate threshold range to determine the fault degree corresponding to the battery group.

[0011] Furthermore, the fault types include: short circuit fault and extrusion fault.

[0012] Furthermore, based on the acquired dynamic behavior characteristics, the adjacent battery packs with faults and their corresponding fault types are acquired by using each real-time time series data, including:

[0013] Set time window , positive rate of change threshold θ up With negative rate of change threshold θ down ;

[0014] For each real-time time series data, sliding window processing is performed on it, and the time derivative exceeds the positive change rate threshold θ up The sudden rise event and its corresponding occurrence time t up , and the time derivative is below the negative rate of change threshold θ down The sudden drop event and its corresponding occurrence time t down, and analyze the continuous change trend of the time derivative. Based on the recorded data and the analysis results, it is determined whether the adjacent battery pack corresponding to the real-time time series data has a fault, and when a fault exists, its corresponding fault type.

[0015] Furthermore, the determination of whether an adjacent battery pack corresponding to the real-time time series data has a fault based on the recorded data and the analysis result, and the corresponding fault type when a fault exists, includes:

[0016] Determine at (any) time t up Preset time window after Is there a time derivative below the negative rate of change threshold θ? down If yes, it is determined that the adjacent battery pack corresponding to the real-time time series data has a fault, and the fault type is a short circuit fault.

[0017] Furthermore, the method of determining whether an adjacent battery pack corresponding to the real-time time series data has a fault based on the recorded data and the analysis result, and the corresponding fault type when a fault exists, further includes:

[0018] Determine at (any) time t up or down Preset time window after Whether the time derivative within shows a trend of continuous slow change, if so, it is determined that the adjacent battery pack corresponding to the real-time time series data has a fault, and the fault type is a squeeze fault.

[0019] Furthermore, the judgment occurs at time t up or down Preset time window after Whether the time derivative within shows a trend of continuous slow change, specifically:

[0020] Calculate the preset time window standard deviation of the inner time derivative;

[0021] If the standard deviation is within the preset standard deviation threshold range, it means that the time derivative is within the preset time window. The internal trend is in a continuous and slow change.

[0022] Furthermore, the change rate threshold range of the corresponding adjacent battery group is updated in real time through the real-time time series data, specifically:

[0023] In real-time time series data, get the most recent n time windows The mean of the time derivative of With standard deviation , and update the change rate threshold range of the corresponding adjacent battery packs through the mean and standard deviation; the update formula is:

[0024] ; In the formula, , represents the control range factor, Indicates the updated change rate threshold range.

[0025] Furthermore, for the adjacent battery pack with a fault, the fault degree corresponding to the battery pack is determined by using the upper and lower limits of the corresponding change rate threshold range, specifically:

[0026] Based on the change rate threshold range [θ min ,θ max ] upper and lower limit settings:

[0027] Severe fault range, including:

[0028] Part above the upper limit: θ max ≤θ, and the part below the lower limit: θ≤θ min ;

[0029] Moderate failure range, including:

[0030] Part close to the upper limit: θ max -δ1≤θ<θ max , and the part close to the lower limit: θ min <θ≤θ min +δ1;

[0031] Minor fault ranges include:

[0032] θ min +δ1<θ≤θ min +δ2 and θ max -δ2≤θ<θ max -δ1; where δ1 and δ2 are preset small positive values, δ1<δ2, and θ is the time derivative;

[0033] According to the fault type of the adjacent battery pack, the fault degree is determined using the set fault interval. When the fault type is a squeeze fault, the determination conditions include:

[0034] When the time t up The corresponding time derivative is above the upper limit, and the time derivative is in the subsequent preset time window If the value is greater than the upper limit continuously, or occurs at time t down The corresponding time derivative is below the lower limit, and the time derivative is within the subsequent preset time window If the value is continuously less than the lower limit, the fault level of the battery pack is judged to be a severe fault;

[0035] When the time t up or down The corresponding time derivative is in the moderate fault interval, and the time derivative is in the subsequent preset time window If the battery pack is continuously in the moderate fault range, the fault level corresponding to the battery pack is determined to be moderate fault;

[0036] When the time t up or down The corresponding time derivative is in the mild fault interval, and the time derivative is in the subsequent preset time window If the battery pack is continuously in the mild fault range, the fault level of the battery pack is determined to be mild fault;

[0037] When the fault type is a short circuit fault, the determination conditions include:

[0038] When the time t up With t down When the corresponding time derivatives are all in the severe fault interval, it is determined that the fault level of the battery pack is severe fault;

[0039] When the time t up With t down If one of the corresponding time derivatives is in the severe fault interval and the other is in the moderate fault interval or the mild fault interval; or if both time derivatives are in the moderate fault interval, then it is determined that the fault level corresponding to the battery pack is a moderate fault;

[0040] When the time t up With t down If the corresponding time derivatives are both in the mild fault range, or one of them is in the mild fault range and the other is in the moderate fault range, then it is determined that the fault level corresponding to the battery pack is a mild fault.

[0041] Furthermore, among two adjacent battery packs, the positive terminal of one battery pack is connected to the positive terminal of the power supply after being wrapped around one side of the corresponding current sensor through a wire; the positive terminal of the other battery pack is simultaneously wrapped around the other side of the corresponding current sensor and connected to the positive terminal of the power supply; the negative terminals of the battery packs are all connected to the negative terminals of the power supply.

[0042] Furthermore, the current detection device also includes: a single chip microcomputer, which is communicatively connected with the current sensors and is used to collect the balancing current output by each current sensor.

[0043] Compared with the prior art, the present invention has at least the following beneficial effects:

[0044] (1) The present invention introduces N-1 current sensors to measure the balancing current between two adjacent battery groups, and uses the voltage, open circuit voltage and internal resistance of two standard battery groups in parallel and in normal working state to analyze the change characteristics of the balancing current over time, and obtains the dynamic behavior characteristics of the time derivative of the balancing current showing continuous change; based on the obtained dynamic behavior characteristics, the adjacent battery groups with faults and their corresponding fault types are obtained using each real-time time series data; for the adjacent battery groups with faults, the upper and lower limits of the corresponding change rate threshold range are used to determine the corresponding fault degree of the battery group, that is, based on the dynamic behavior characteristics, the present invention uses the balancing current detected by the current detection device to obtain the adjacent battery groups with faults and their corresponding fault types, and determines the corresponding fault degree of the battery group based on the upper and lower limits of the change rate threshold range, effectively identifying the abnormal current fluctuation caused by the fault, and improving the accuracy of fault detection;

[0045] (2) The present invention uses real-time time series data to update the change rate threshold range of the corresponding adjacent battery groups in real time. For the adjacent battery group with faults, the upper and lower limits of the corresponding change rate threshold range are used to determine the fault degree of the battery group. That is, the present invention enhances the adaptability to different working conditions and aging states by dynamically adjusting the change rate threshold range, ensuring high sensitivity and reliability in complex and changeable working environments.

[0046] (3) For adjacent battery packs with faults, the present invention further uses the upper and lower limits of the corresponding change rate threshold range to determine the severity of the fault (mild, moderate or severe), which provides clear guidance for subsequent maintenance and repair, allowing maintenance personnel to formulate reasonable response strategies based on the specific circumstances of the fault, thereby reducing unnecessary downtime and costs;

[0047] (4) The fault detection method of the present invention is applicable to complex battery systems comprising multiple parallel battery packs, especially large-scale parallel battery packs. It can effectively monitor the status of the battery packs and quickly locate the faulty battery packs to ensure the safe operation of the entire battery system. This is particularly important for application scenarios such as large-scale energy storage systems and electric vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a flow chart of a fault detection method for a battery system in an embodiment of the present invention;

[0049] Figure 2 A structural diagram of a battery system and a current detection device in an embodiment of the present invention;

[0050] Figure 3 Another structural diagram of a battery system and a current detection device in an embodiment of the present invention;

[0051] Figure 4 A schematic diagram of a balancing current and voltage curve of a parallel battery pack in a 1C charging state according to an embodiment of the present invention;

[0052] Figure 5 A schematic diagram of a curve of balancing current of battery packs with different capacities under 7mm compression in an embodiment of the present invention;

[0053] Figure 6 Schematic diagram of curves showing different battery characteristics of battery packs with different capacities under stress operation in an embodiment of the present invention;

[0054] Figure 7 Schematic diagram of the curve of the corresponding balancing current when the battery packs with different capacities are externally connected to a short-circuit resistor of 1000Ω in the embodiment of the present invention.

[0055] In the figure:

[0056] 1. Battery pack; 2. Current sensor; 3. Microcontroller; 4. Power supply. DETAILED DESCRIPTION

[0057] The following are specific embodiments of the present invention and the accompanying drawings to further describe the technical solution of the present invention, but the present invention is not limited to these embodiments.

[0058] In order to effectively identify abnormal current fluctuations caused by faults and improve the accuracy of fault detection, such as Figure 1 As shown, an embodiment of the present invention proposes a fault detection method for a battery system, which implements fault detection based on a current detection device; the battery system includes N parallel battery packs 1, N is a positive integer greater than or equal to 2; the current detection device includes: N-1 current sensors 2, each current sensor 2 measures the balancing current between two adjacent battery packs 1; the balancing current is the current difference flowing through the two parallel battery packs 1; specifically, in the battery system, the branch currents of any two symmetrical battery packs (ideally, the two battery packs are completely consistent) are equal in magnitude and consistent in direction, and are in a balanced state. However, when there is a difference in internal resistance, the balanced state is broken, and the difference between the two branch currents is called the balancing current.

[0059] It should be noted that in order to accurately measure the current, this embodiment chooses to use a leakage current sensor based on the Hall effect, which works on the principle of magnetic modulation and magnetic balance. By adopting a high magnetic permeability iron core and achieving alternating saturation under the excitation of an alternating modulated magnetic field, the sensor can detect the weak magnetic flux signal generated by the external current and measure a trace amount of DC signal accordingly. In a specific application, the wire is wound around both sides of the sensor in the form of a spiral coil. When a current passes through the wire, the spiral coil formed will generate a magnetic field. The clever design is that this arrangement allows the magnetic fields on both sides to cancel each other out, leaving only the differential signal, that is, the value of the balanced current. This method not only improves the measurement accuracy, but also effectively reduces the influence of external electromagnetic interference, ensuring accurate measurement of the balanced current. This feature is crucial for monitoring small current changes in battery systems, especially providing reliable data support for fault diagnosis and preventive maintenance.

[0060] Among two adjacent battery groups 1, the positive terminal of one battery group 1 is connected to the positive terminal of the power source 4 after being wrapped around one side of the corresponding current sensor 2 through a wire; the positive terminal of the other battery group 1 is connected to the positive terminal of the power source 4 after being wrapped around the other side of the corresponding current sensor 2; the negative terminals of the battery groups 1 are all connected to the negative terminals of the power source 4.

[0061] The current detection device further includes: a single chip computer 3 , which is in communication connection with the current sensor 2 and is used for collecting the balancing current output by each current sensor 2 .

[0062] Specifically, this embodiment provides a battery system comprising two battery packs 1, and the connection relationship between the battery system and the current detection device is as follows: Figure 2 As shown:

[0063] Figure 2 In the figure, a battery pack 1 includes N battery cells; the positive terminal of one battery pack 1 is wound with one side of the corresponding current sensor 2 and then connected to the positive terminal of the power source 4, and the positive terminal of another battery pack 1 is wound with the other side of the current sensor 2 and then connected to the positive terminal of the power source 4; the negative terminals of the battery packs 1 are all connected to the negative terminal of the power source 4; the single-chip computer 3 collects the balancing current through the output port (RS485 interface) of the current sensor 2.

[0064] This embodiment also provides a battery system comprising three battery packs, and the connection relationship between the battery system and the current detection device is as follows: Figure 3 As shown:

[0065] Among the three battery groups connected in parallel, the first and second battery groups 1 are a first pair of adjacent battery groups, and the second and third battery groups 1 are a second pair of adjacent battery groups;

[0066] For the convenience of explanation, the current sensors 2 are numbered, and the current sensor 2 corresponding to the first pair of adjacent battery groups is set as current sensor 2 No. 1, and the current sensor 2 corresponding to the second pair of adjacent battery groups is set as current sensor 2 No. 2;

[0067] Figure 3 In the first pair of adjacent battery groups, the positive terminal of the first battery group 1 is wound around one side of the No. 1 current sensor 2 and then connected to the positive terminal of the power source 4, and the positive terminal of the other battery group 1 is wound around one side of the No. 2 current sensor 2 and the other side of the No. 1 current sensor 2 and then connected to the positive terminal of the power source 4; in the second pair of adjacent battery groups, the positive terminal of the third battery group 1 is wound around the other side of the No. 2 current sensor 2 and then connected to the positive terminal of the power source 4; the negative terminal of each battery group 1 is connected to the negative terminal of the power source 4; the single-chip computer 3 collects the corresponding balancing current through the output port (RS485 interface) of each current sensor 2.

[0068] For a battery system including more battery packs, the connection with the current sensor in the current detection device is the same as Figure 3 Connection method shown.

[0069] The fault detection method comprises:

[0070] By using the voltage, open circuit voltage and internal resistance of two standard battery packs (i.e., undamaged and fault-free battery packs) in parallel and in normal working condition, the time-varying characteristics of the balancing current are analyzed, and the dynamic behavior characteristics of the continuous change of the time derivative of the balancing current are obtained; specifically:

[0071] When the battery pack adopts a series-first-then-parallel structure, the current difference flowing through the two parallel battery packs is defined as the balancing current. Figure 2 The figure shows the current flow of two parallel battery packs. In this structure, battery packs Pack1 and Pack2 are in an equal position. Let the voltages of battery packs Pack1 and Pack2 be V1 and V2 respectively. According to the parallel circuit characteristics, the terminal voltage of each battery pack is determined by its open circuit voltage and internal resistance. Let the open circuit voltage of the battery pack be OCV and the internal resistance be , then the current relationship and terminal voltage satisfy the following formula:

[0072] ;

[0073] In the formula, OCV1 and OCV2 represent the open circuit voltages of battery packs Pack1 and Pack2 respectively. and They represent the internal resistance of battery packs Pack1 and Pack2 respectively, and I1 and I2 represent the operating current of battery packs Pack1 and Pack2 respectively;

[0074] The balancing current is defined as , then the calculation formula for the balancing current under charging conditions is:

[0075] ;

[0076] In the formula, Indicates the total current;

[0077] The internal resistance changes dynamically with the state of the battery (such as state of charge SOC, temperature, etc.), while the change in open circuit voltage is a smooth function of SOC (approximately a linear or exponential function). Therefore, based on the calculation formula of the balancing current, it can be seen that the change in internal resistance will cause dynamic adjustment of the balancing current. The change in balancing current mainly depends on OCV and internal resistance. When two series battery packs are connected in parallel with the same terminal voltage, the state of charge (SOC) change trend of the battery will remain consistent when they are charged and discharged at the same time. Assuming that the internal resistance changes smoothly with time t, the rate of change of the balancing current with time can be calculated. Specifically, the time derivative of the balancing current is affected not only by the rate of change of OCV (because OCV is a smooth function of SOC and its change is continuous), but also by the rate of change of internal resistance (internal resistance changes slowly with SOC and temperature).

[0078] Assuming that the internal resistance changes smoothly with time t, the formula for calculating the rate of change of the balanced current with time is:

[0079] ;

[0080] In the formula, , Respectively represent the sum and difference of the internal resistance of two battery packs (battery packs Pack1 and Pack2); Represents the difference in open circuit voltage between two battery packs; It represents the rate of change of the equilibrium current with time, that is, the time derivative.

[0081] From the above analysis, it can be seen that the rate of change of the balancing current is mainly determined by two parts: first, the rate of change of the open circuit voltage (OCV). Since OCV is a smooth function of the state of charge (SOC), its change is continuous and gradual, which means that as the SOC changes, the OCV will change in a smooth and predictable way; second, the rate of change of the internal resistance. Under normal circumstances, the internal resistance will change slowly and continuously with the change of SOC and temperature, that is, this change is also continuous, although the speed is slow, but it still affects the balancing current. Therefore, the time derivative of the balancing current is the result of the combined effect of these two, which manifests as a continuous change process. This characteristic enables the effective detection of abnormal conditions in the battery system by monitoring the balancing current and its time derivative, and timely measures to ensure the safety and reliability of the system.

[0082] Below through Figures 4 to 7 The curve diagram shown in the figure (in the figure, average current represents the balancing current, voltage represents the voltage, and Time represents the time) explains in detail why abnormal conditions in the battery system (such as shelving, squeezing, and internal short circuit) can be effectively detected by monitoring the balancing current:

[0083] 1. Figure 4 It consists of two stages:

[0084] 1) Charging stage:

[0085] During the charging stage (0-4000 seconds), the value of the balance current shows a smooth change. As can be seen from the figure, the current gradually decreases from the initial value and maintains a relatively stable change trend during the charging process. The voltage gradually increases from a lower value until it reaches a stable value close to 4.0V.

[0086] 2) Shelving stage:

[0087] When entering the rest stage (after 4000 seconds), the balancing current shows a significant mutation due to the sudden disconnection of the charging voltage. This mutation is a transient response caused by the immediate removal of the charging power supply, which causes the original charging current to disappear quickly, thereby causing a short current fluctuation. Subsequently, due to the inherent inconsistency between the two batteries, the battery with higher voltage begins to supplement the battery with lower voltage. This process causes the balancing current to gradually decrease and eventually approach zero. Specifically, as the rest time increases, the voltage difference between the two battery groups gradually decreases until a new equilibrium state is reached, at which time the balancing current is almost zero, indicating that the state of charge between the two battery groups is basically synchronized.

[0088] 2. Figure 5The figure shows the changes in the balancing current of battery packs with different states of charge (SOC, 20%, 40%, 60% and 80% respectively) when they are placed in a standby state and subjected to two 3.5mm squeezes. As can be seen from the figure, all SOC curves show obvious fluctuations in three key stages: the first squeeze occurs at about 750 seconds, and all curves show significant current transients at this time, indicating that the squeeze has an immediate impact on the internal structure of the battery; the second squeeze occurs at about 1500 seconds, which triggers similar current fluctuations again, reflecting the response of the battery under the second squeeze; finally, after about 2100 seconds, it enters the lifting stage (referring to the process in which the mechanical pressure applied to the battery is released after the squeeze test is completed), at which time all curves gradually recover and approach zero, indicating that as the squeeze ends, the internal state of the battery gradually returns to a stable state, and the balancing current returns to normal. By monitoring these current changes, we can gain a deep understanding of the response characteristics of the battery to mechanical stress under different SOCs, providing an important basis for fault diagnosis and safety assessment.

[0089] 3. Figure 6 middle:

[0090] Figure (a): shows the balance current change curves of battery packs with different SOC (20%, 40%, 60% and 80%) when they are squeezed twice with 3.5mm in a standby state. As can be seen from the figure, at the three key time points of about 750 seconds, 1500 seconds and 2100 seconds, all curves have significant fluctuations, indicating that squeezing has an immediate impact on the internal structure of the battery.

[0091] Figure (b): shows the temperature change of the electrode of the battery pack with different SOC (20%, 40%, 60% and 80%) under the same test conditions. Unlike the balance current curve, the electrode temperature does not fluctuate significantly at these three key time points, but remains relatively stable.

[0092] Figure (c): shows the overall temperature change of the battery packs with different SOC (20%, 40%, 60% and 80%) under the same test conditions. Similarly, the overall battery temperature did not change significantly at these three key time points, maintaining a relatively stable trend.

[0093] Figure (d) shows the voltage changes of battery packs with different SOC (20%, 40%, 60% and 80%) under the same test conditions. The voltage curve did not show obvious fluctuations at these three key time points and remained relatively stable.

[0094] In summary, Figure 6Figures a, b, c, and d in the figure show the changes in the balance current, electrode temperature, overall battery temperature, and voltage of the battery pack at different SOCs. The balance current curve fluctuates greatly under the three stress changes, while the electrode temperature, overall battery temperature, and voltage do not change significantly at these three locations.

[0095] 4. In the abnormal state of the battery, internal short circuit failure is a common phenomenon. Since the short circuit resistance is large in the early stage of internal short circuit, a short circuit resistance of 1000Ω is selected to simulate the situation of early internal short circuit failure. Figure 7 The figure shows the changes in the balancing current of lithium battery packs with different SOCs (20%, 40%, 60% and 80%) when an external 1000Ω short-circuit resistor is connected. As can be seen from the figure, when the short circuit occurs (about 600 seconds), the balancing current fluctuates greatly, and the effect is more obvious. Specifically, the balancing current of all battery packs with different SOCs rises rapidly at the beginning of the short circuit, and then drops rapidly after the short circuit ends (about 700 seconds) and gradually returns to a stable state. This shows that the internal short circuit has an immediate impact on the internal structure of the battery, and this impact is consistent at different SOCs.

[0096] In this embodiment, the internal short circuit fault refers to the abnormal conduction phenomenon between the internal electrodes of the battery (i.e., the positive and negative electrodes of the battery cell), which causes the current to flow directly inside the battery instead of through the normal charging and discharging path. Specifically, the internal short circuit fault can be caused by the following situations:

[0097] 1. Mechanical damage:

[0098] Due to mechanical stress such as external extrusion and collision, the internal structure of the battery is deformed or damaged, causing the positive and negative electrode materials to be in direct contact or connected through tiny conductive paths.

[0099] 2. Diaphragm damage:

[0100] The separator inside the battery (used to separate the positive and negative electrodes and prevent short circuits) is damaged due to aging, thermal runaway or other reasons, resulting in direct or near contact between the positive and negative electrodes.

[0101] 3. Metal impurities:

[0102] Metal impurities or foreign matter mixed in during the battery manufacturing process may form tiny conductive paths inside the battery, causing internal short circuits.

[0103] The balancing current between two adjacent battery groups in the battery system is acquired in real time by a current detection device, and real-time time series data corresponding to the current sensor is obtained; the real-time time series data includes the balancing current corresponding to each time point and the time derivative, that is, the rate of change of the balancing current with time;

[0104] Based on the acquired dynamic behavior characteristics, each real-time time series data is used to obtain adjacent battery packs with faults and their corresponding fault types; the fault types include: short circuit faults (specifically internal short circuit faults) and extrusion faults.

[0105] The method of obtaining adjacent battery packs with faults and their corresponding fault types based on the acquired dynamic behavior characteristics using each real-time time series data includes:

[0106] Set time window , positive rate of change threshold θ up With negative rate of change threshold θ down ;

[0107] For each real-time time series data, sliding window processing is performed on it, and the time derivative exceeds the positive change rate threshold θ up The sudden rise event and its corresponding occurrence time t up , and the time derivative is below the negative rate of change threshold θ down The sudden drop event and its corresponding occurrence time t down , and analyze the continuous change trend of the time derivative. Based on the recorded data and the analysis results, it is determined whether the adjacent battery pack corresponding to the real-time time series data has a fault, and when a fault exists, its corresponding fault type.

[0108] Specifically:

[0109] Characteristics and judgment of internal short circuit fault:

[0110] In the case of an internal short circuit fault, the balancing current usually increases rapidly and approaches the peak value, and then the current drops rapidly due to changes in the internal resistance of the battery or the action of the protection mechanism. This phenomenon is manifested as a significant local peak (positive value) on the current curve, marking a sudden increase in current; followed by a steep drop inflection point (negative rate of change), which is the result of a rapid decrease in current due to an increase in internal resistance or the activation of a protection mechanism. The amplitude and duration of the extreme point can reflect the severity of the short circuit: if the short-circuit resistance is small, the slope of the balancing current curve will change more steeply. The specific judgment method is to calculate the local rate of change of the balancing current through a sliding window. When a combination of a sudden increase and a sudden drop in the local rate of change is detected, it can be judged that an internal short circuit fault has occurred.

[0111] Characteristics and judgment of extrusion failure:

[0112] For extrusion failure, the balancing current gradually decreases during extrusion and recovers to a limited extent after release. This is because extrusion causes damage to the active material, resulting in a continuous increase in internal resistance. At the beginning of extrusion, the balancing current curve may show a sudden rise or fall inflection point, which reflects the sudden change in battery internal resistance or the local electrode short circuit or rapid change in contact state caused by extrusion. During the extrusion maintenance stage, the curve enters a slow change stage, and multiple slow-changing inflection points may appear, indicating that the internal resistance gradually increases, for example, due to the shedding of active materials or changes in electrolyte distribution. At the moment of extrusion release, a significant extreme inflection point (positive or negative) may appear, and then the curve gradually returns to a stable state, reflecting the partial recovery of the battery internal resistance after the mechanical stress is released, and the local short circuit that may disappear. In order to judge the extrusion fault, the method of calculating the local change rate of the balancing current using a sliding window is also used. When the local change rate shows a trend of continuous slow change after a single sudden rise or drop, it can be judged as an extrusion fault.

[0113] The method of determining whether an adjacent battery pack corresponding to the real-time time series data has a fault based on the recorded data and the analysis result, and the corresponding fault type when a fault exists, includes:

[0114] Determine at (any) time t up Preset time window after Is there a time derivative below the negative rate of change threshold θ? down If yes, it is determined that the adjacent battery pack corresponding to the real-time time series data has a fault, and the fault type is a short circuit fault.

[0115] The step of determining whether an adjacent battery pack corresponding to the real-time time series data has a fault based on the recorded data and the analysis result, and the corresponding fault type when a fault exists, further includes:

[0116] Determine at (any) time t up or down Preset time window after Whether the time derivative within shows a trend of continuous slow change, if so, it is determined that the adjacent battery pack corresponding to the real-time time series data has a fault, and the fault type is a squeeze fault.

[0117] The judgment occurs at the time t up or down Preset time window after Whether the time derivative within shows a trend of continuous slow change, specifically:

[0118] Calculate the preset time window standard deviation of the inner time derivative;

[0119] If the standard deviation is within the preset standard deviation threshold range, it means that the time derivative is within the preset time window. The internal trend is in a continuous and slow change.

[0120] Initialize the change rate threshold range of each adjacent battery group, and update the change rate threshold range of the corresponding adjacent battery group in real time through real-time time series data. For adjacent battery groups with faults, use the upper and lower limits of the corresponding change rate threshold range to determine the fault degree corresponding to the battery group.

[0121] The present invention uses real-time time series data to update the change rate threshold range of corresponding adjacent battery groups in real time. For adjacent battery groups with faults, the upper and lower limits of the corresponding change rate threshold range are used to determine the fault degree corresponding to the battery group. That is, the present invention enhances the adaptability to different working conditions and aging states by dynamically adjusting the change rate threshold range, ensuring high sensitivity and reliability in complex and changeable working environments; at the same time, this provides clear guidance for subsequent maintenance and repair, allowing maintenance personnel to formulate reasonable response strategies based on the specific circumstances of the fault, thereby reducing unnecessary downtime and costs.

[0122] The real-time time series data is used to update the change rate threshold range of the corresponding adjacent battery group in real time, specifically:

[0123] In real-time time series data, get the most recent n time windows The mean of the time derivative of With standard deviation , and update the change rate threshold range of the corresponding adjacent battery packs through the mean and standard deviation; the update formula is:

[0124] ; In the formula, , represents the control range factor, Indicates the updated change rate threshold range.

[0125] For the adjacent battery pack with fault, the fault degree corresponding to the battery pack is determined by using the upper and lower limits of the corresponding change rate threshold range, specifically:

[0126] Based on the change rate threshold range [θ min ,θ max ] upper and lower limit settings:

[0127] Severe fault range, including:

[0128] Part above the upper limit: θ max ≤θ, and the part below the lower limit: θ≤θ min ;

[0129] Moderate failure range, including:

[0130] Part close to the upper limit: θ max -δ1≤θ<θ max , and the part close to the lower limit: θ min <θ≤θ min +δ1;

[0131] Minor fault ranges include:

[0132] θ min +δ1<θ≤θ min +δ2 and θ max -δ2≤θ<θ max -δ1; where δ1 and δ2 are preset small positive values, δ1<δ2, θ is the time derivative; the middle part is θ min +δ2<θ<θ max -δ2 is the normal interval, where θ min +δ2 is the negative rate of change threshold θ down ,θ max -δ2 is the positive rate of change threshold θ up ;

[0133] According to the fault type of the adjacent battery pack, the fault degree is determined using the set fault interval. When the fault type is a squeeze fault, the determination conditions include:

[0134] When the time t up The corresponding time derivative is above the upper limit, and the time derivative is in the subsequent preset time window If the value is greater than the upper limit continuously, or occurs at time t down The corresponding time derivative is below the lower limit, and the time derivative is within the subsequent preset time window If the fault level of the battery pack is continuously less than the lower limit within t up The corresponding time derivative is above the upper limit, but in the subsequent preset time windows The value is not greater than the upper limit continuously within t; or the time when the down The corresponding time derivative is below the lower limit, but in the subsequent preset time windows If the value is not continuously less than the lower limit, it is a moderate fault;

[0135] When the time t up or down The corresponding time derivative is in the moderate fault interval, and the time derivative is in the subsequent preset time window If the battery pack is continuously in the moderate fault range within t, the fault level corresponding to the battery pack is determined to be moderate fault; when the time t up or downThe corresponding time derivative is in the moderate fault interval, but the time derivative is in the subsequent preset time window If the battery pack gradually recovers to the mild fault range or normal level within a certain period of time, the fault level of the battery pack is determined to be mild fault;

[0136] When the time t up or down The corresponding time derivative is in the mild fault interval, and the time derivative is in the subsequent preset time window If the battery pack remains in the mild fault range or returns to normal level, the fault level of the battery pack is determined to be mild fault.

[0137] When the fault type is a short circuit fault, the determination conditions include:

[0138] When the time t up With t down When the corresponding time derivatives are all in the severe fault interval, it is determined that the fault level of the battery pack is severe fault;

[0139] When the time t up With t down If one of the corresponding time derivatives is in the severe fault interval and the other is in the moderate fault interval or the mild fault interval; or if both time derivatives are in the moderate fault interval, then it is determined that the fault level corresponding to the battery pack is a moderate fault;

[0140] When the time t up With t down If the corresponding time derivatives are both in the mild fault range, or one of them is in the mild fault range and the other is in the moderate fault range, then it is determined that the fault level corresponding to the battery pack is a mild fault.

[0141] It should be noted that this embodiment also provides another implementation method of fault detection: based on the acquired dynamic behavior characteristics, the adjacent battery packs with faults and their corresponding fault types are acquired using each real-time time series data, including:

[0142] 1. Data acquisition and preprocessing

[0143] Get the dataset:

[0144] Data collected during the experiment: including balancing current data, temperature data, voltage data and time series data (including the balancing current and time derivatives corresponding to each time point). These data are derived from the operation records of the actual battery system, especially the measurements under different working conditions (such as charging, storage, short circuit simulation, etc.) and fault conditions (such as extrusion fault, internal short circuit fault).

[0145] Data preprocessing:

[0146] Noise removal: Use filters (such as low-pass filters or wavelet transforms) to clean up the noise in the data to improve the accuracy of subsequent analysis. Especially for balanced current data, it is critical to reduce the impact of high-frequency noise on fault detection.

[0147] Data normalization: Normalize data of different dimensions (such as equilibrium current, temperature, voltage) to a uniform range (such as [0,1] or [-1,1]) so that the model can better handle multi-dimensional inputs.

[0148] Window division: Sliding window technology is used for time series data to generate fixed-length segments as training samples. Each window contains all sensor data within a period of time to ensure that sufficient time-dependent features are captured.

[0149] Data labeling: Add a label to each training sample (including fixed-length time series data and its corresponding balanced current data, temperature data, and voltage data) to identify its corresponding fault type and fault severity.

[0150] 2. Training set and test set division

[0151] Divide the dataset:

[0152] Divide the data set according to a certain ratio (e.g. 70% for training set and 30% for test set) to ensure that the data is evenly distributed and the number of various fault samples in the training set and test set is as consistent as possible. This helps to evaluate the generalization ability and reliability of the model.

[0153] 3. Model selection and training

[0154] Select Model:

[0155] Long Short-Term Memory (LSTM): Since LSTM is good at capturing long-term dependencies in time series data, it is very suitable for processing the dynamic behavior characteristics of battery systems, such as the change of balancing current over time.

[0156] Hybrid model (such as LSTM+CNN): Combining the advantages of LSTM and convolutional neural network (CNN), it can not only capture the long-term dependencies of time series, but also extract local features. CNN can effectively extract important local patterns from the data within the time window, further improving the performance of the model.

[0157] Training objectives:

[0158] The selected model is trained and tested using the training set and the test set. During the training process, the cross entropy loss function or other loss functions suitable for the classification task are used to optimize the model parameters to ensure that the model can accurately identify different types of faults.

[0159] 4. Input the real-time time series data to be predicted into the trained model to predict the fault type and fault severity.

[0160] Through the above steps, the present invention not only realizes the effective identification of internal faults of the battery system (such as internal short circuit and extrusion faults), but also improves the accuracy and efficiency of fault diagnosis by using deep learning technology. This method can not only be applied to fault simulation experiments in laboratory environments, but can also be extended to practical applications, providing important guarantees for the safe operation of large-scale parallel battery systems.

[0161] The present invention introduces N-1 current sensors to measure the balancing current between two adjacent battery groups, and uses the voltage, open circuit voltage and internal resistance of two parallel standard battery groups in normal working state to analyze the change characteristics of the balancing current over time, and obtains the dynamic behavior characteristics of the time derivative of the balancing current changing continuously; based on the acquired dynamic behavior characteristics, the adjacent battery groups with faults and their corresponding fault types are obtained by using each real-time time series data; for the adjacent battery groups with faults, the upper and lower limits of the corresponding change rate threshold range are used to determine the corresponding fault degree of the battery group, that is, based on the dynamic behavior characteristics, the present invention uses the balancing current detected by the current detection device to obtain the adjacent battery groups with faults and their corresponding fault types, and determines the corresponding fault degree of the battery group based on the upper and lower limits of the change rate threshold range, effectively identifying the abnormal current fluctuation caused by the fault, and improving the accuracy of fault detection. At the same time, the fault detection method of the present invention is suitable for complex battery systems including multiple parallel battery groups, especially large-scale parallel battery groups. It can effectively monitor the state of the battery group and realize the rapid positioning of the faulty battery group to ensure the safe operation of the entire battery system, which is particularly important for application scenarios such as large energy storage systems and electric vehicles.

[0162] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0163] In addition, in the present invention, descriptions such as "first", "second", "one", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0164] In the present invention, unless otherwise clearly specified and limited, the terms "connection", "fixation", etc. should be understood in a broad sense. For example, "fixation" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0165] In addition, the technical solutions between the various embodiments of the present invention can be combined with each other, but it must be based on the fact that ordinary technicians in the field can implement it. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

Claims

1. A fault detection method for a battery system, characterized in that: Fault detection is implemented based on a current detection device; the battery system includes N parallel-connected battery packs, where N is a positive integer greater than or equal to 2; The current detection device comprises: N-1 current sensors, each current sensor measures the balancing current between two adjacent battery packs; the balancing current is the current difference flowing through the two parallel battery packs; The fault detection method comprises: By using the voltage, open circuit voltage and internal resistance of two standard battery packs in parallel and in normal working state, the time-varying characteristics of the balancing current are analyzed, and the dynamic behavior characteristics of the time derivative of the balancing current showing continuous changes are obtained; The balancing current between two adjacent battery groups in the battery system is acquired in real time by a current detection device, and real-time time series data corresponding to the current sensor is obtained; the real-time time series data includes the balancing current corresponding to each time point and the time derivative, that is, the rate of change of the balancing current with time; Based on the acquired dynamic behavior characteristics, the adjacent battery groups with faults and their corresponding fault types are obtained using each real-time time series data; Initialize the change rate threshold range of each adjacent battery group, and update the change rate threshold range of the corresponding adjacent battery group in real time through real-time time series data. For adjacent battery groups with faults, use the upper and lower limits of the corresponding change rate threshold range to determine the fault degree corresponding to the battery group.

2. A fault detection method for a battery system according to claim 1, characterized in that: The fault types include: short circuit fault and extrusion fault.

3. A fault detection method for a battery system according to claim 2, characterized in that: The method of obtaining adjacent battery packs with faults and their corresponding fault types based on the acquired dynamic behavior characteristics using each real-time time series data includes: Set time window , positive rate of change threshold θ up With negative rate of change threshold θ down ; For each real-time time series data, sliding window processing is performed on it, and the time derivative exceeds the positive change rate threshold θ up The sudden rise event and its corresponding occurrence time t up , and the time derivative is below the negative rate of change threshold θ down The sudden drop event and its corresponding occurrence time t down , and analyze the continuous change trend of the time derivative. Based on the recorded data and the analysis results, it is determined whether the adjacent battery pack corresponding to the real-time time series data has a fault, and when a fault exists, its corresponding fault type.

4. A fault detection method for a battery system according to claim 3, characterized in that: The method of determining whether an adjacent battery pack corresponding to the real-time time series data has a fault based on the recorded data and the analysis result, and the corresponding fault type when a fault exists, includes: Determine the occurrence time t up Preset time window after Is there a time derivative below the negative rate of change threshold θ? down If yes, it is determined that the adjacent battery pack corresponding to the real-time time series data has a fault, and the fault type is a short circuit fault.

5. A fault detection method for a battery system according to claim 4, characterized in that: The step of determining whether an adjacent battery pack corresponding to the real-time time series data has a fault based on the recorded data and the analysis result, and the corresponding fault type when a fault exists, further includes: Determine the occurrence time t up or down Preset time window after Whether the time derivative within shows a trend of continuous slow change, if so, it is determined that the adjacent battery pack corresponding to the real-time time series data has a fault, and the fault type is a squeeze fault.

6. A fault detection method for a battery system according to claim 5, characterized in that: The judgment occurs at the time t up or down Preset time window after Whether the time derivative within shows a trend of continuous slow change, specifically: Calculate the preset time window standard deviation of the inner time derivative; If the standard deviation is within the preset standard deviation threshold range, it means that the time derivative is within the preset time window. The internal trend is in a continuous and slow change.

7. A fault detection method for a battery system according to claim 6, characterized in that: The real-time time series data is used to update the change rate threshold range of the corresponding adjacent battery group in real time, specifically: In real-time time series data, get the most recent n time windows The mean of the time derivative of With standard deviation , and update the change rate threshold range of the corresponding adjacent battery packs through the mean and standard deviation; the update formula is: ; In the formula, k represents the control range factor, Indicates the updated change rate threshold range.

8. A fault detection method for a battery system according to claim 7, characterized in that: For the adjacent battery pack with fault, the fault degree corresponding to the battery pack is determined by using the upper and lower limits of the corresponding change rate threshold range, specifically: Based on the change rate threshold range [θ min ,θ max ] upper and lower limit settings: Severe fault range, including: Part above the upper limit: θ max ≤θ, and the part below the lower limit: θ≤θ min ; Moderate failure range, including: Part close to the upper limit: θ max -δ1≤θ<θ max , and the part close to the lower limit: θ min <θ≤θ min +δ1; Minor fault ranges include: θ min +δ1<θ≤θ min +δ2 and θ max -δ2≤θ<θ max -δ1; where δ1 and δ2 are preset small positive values, δ1<δ2, and θ is the time derivative; According to the fault type of the adjacent battery pack, the fault degree is determined using the set fault interval. When the fault type is a squeeze fault, the determination conditions include: When the time t up The corresponding time derivative is above the upper limit, and the time derivative is in the subsequent preset time window If the value is greater than the upper limit continuously, or occurs at time t down The corresponding time derivative is below the lower limit, and the time derivative is within the subsequent preset time window If the value is continuously less than the lower limit, the fault level of the battery pack is judged to be a severe fault; When the time t up or down The corresponding time derivative is in the moderate fault interval, and the time derivative is in the subsequent preset time window If the battery pack is continuously in the moderate fault range, the fault level corresponding to the battery pack is determined to be moderate fault; When the time t up or down The corresponding time derivative is in the mild fault interval, and the time derivative is in the subsequent preset time window If the battery pack is continuously in the mild fault range, the fault level of the battery pack is determined to be mild fault; When the fault type is a short circuit fault, the determination conditions include: When the time t up With t down When the corresponding time derivatives are all in the severe fault interval, it is determined that the fault level of the battery pack is severe fault; When the time t up With t down If one of the corresponding time derivatives is in the severe fault interval and the other is in the moderate fault interval or the mild fault interval; or if both time derivatives are in the moderate fault interval, then it is determined that the fault level corresponding to the battery pack is a moderate fault; When the time t up With t down If the corresponding time derivatives are both in the mild fault range, or one of them is in the mild fault range and the other is in the moderate fault range, then it is determined that the fault level corresponding to the battery pack is a mild fault.

9. A fault detection method for a battery system according to claim 1, characterized in that: Among two adjacent battery packs, the positive terminal of one battery pack is connected to the positive terminal of the power supply after being wrapped with one side of the corresponding current sensor through a wire; the positive terminal of the other battery pack is connected to the positive terminal of the power supply after being wrapped with the other side of the corresponding current sensor; the negative terminals of the battery packs are all connected to the negative terminals of the power supply.

10. A fault detection method for a battery system according to claim 9, characterized in that: The current detection device further includes: a single chip microcomputer, which is in communication connection with the current sensors and is used for collecting the balancing current output by each current sensor.

Images