A new energy automobile battery pack thermal runaway early warning method based on temperature gradient compensation

Abstract

The present application relates to the field of new energy automobile battery technology, and more particularly to a new energy automobile battery pack thermal runaway early warning method based on temperature gradient compensation, comprising the following steps: taking the temperature values obtained by all sampling points in the battery pack in the last inspection cycle as the reference, inhibiting the temperature values obtained in the current inspection cycle from environmental drift, and outputting the compensated temperature field after compensation; comparing the temperature gradient change rate with the preset adaptive threshold value, and outputting the suspected node sequence with thermal runaway suspicion; reversely tracking whether the suspected node will spread heat to the adjacent nodes in the next inspection cycle, and if the spread range exceeds the tracing threshold, issuing an unique thermal runaway early warning signal immediately. The present application, the node level degradation modeling and temperature compensation strategy constitute a personalized, evolution driven thermal state perception framework, which is the key point that is different from the traditional fixed threshold method or static temperature judgment method.

Description

Technical Field

[0001] This invention relates to the field of new energy vehicle battery technology, and in particular to a method for early warning of thermal runaway in new energy vehicle battery packs based on temperature gradient compensation. Background Technology

[0002] With the large-scale application of new energy vehicles, the safety of power battery systems has increasingly become a focus of industry attention. Thermal runaway, as one of the most dangerous failure modes in the operation of power batteries, is often caused by abnormal heating of local cells and may form an uncontrollable thermal diffusion chain in a short period of time. Therefore, establishing a high-precision, early-response thermal runaway early warning mechanism has become one of the core tasks of the power battery management system (BMS).

[0003] Existing technologies often employ fixed-value judgment methods based on cell temperature monitoring or multi-point temperature difference mutation detection methods. For example, a fixed temperature gradient threshold or upper limit for the temperature rise rate is set, and an alarm is triggered once the detected value exceeds the threshold. However, these methods have significant limitations in practical applications. During battery pack operation, factors such as ambient temperature fluctuations, thermal management system adjustments, or changes in vehicle operating conditions can cause global drift in all temperature sensors, easily masking local abnormal temperature rise signals or misinterpreting normal drift as anomalies. As the battery cell's lifespan progresses, its internal resistance gradually increases, and even under the same operating conditions, aged cells are more prone to generating additional heat. Existing methods do not differentiate between the aging states of different cells, leading to confusion between thermal anomalies and aging degradation signals, making it difficult to accurately determine the source of thermal runaway risk. Summary of the Invention

[0004] This invention provides a method for early warning of thermal runaway in new energy vehicle battery packs based on temperature gradient compensation. It can improve the accuracy of thermal runaway identification by using a robust, multi-dimensional dynamic judgment method based on temperature compensation, aging rewind, and diffusion tracing.

[0005] A method for early warning of thermal runaway in new energy vehicle battery packs based on temperature gradient compensation includes the following steps: S1. Gradient compensation step: Based on the temperature values ​​obtained by all sampling points in the battery pack in the previous inspection cycle, environmental drift suppression is performed on the temperature values ​​obtained in the current inspection cycle, and the compensated temperature field is output. S2. Gradient chain consistency verification step: Using the compensated temperature field as input, establish a gradient chain along the series and parallel conductive path of the battery pack, calculate the temperature gradient change rate between adjacent nodes in the gradient chain, compare the temperature gradient change rate with the preset adaptive threshold, and output the sequence of suspected nodes with thermal runaway suspicion. Before calculating the temperature gradient change rate, an aging offset rollback operation is introduced for each node in the gradient chain. This includes taking the moment when the temperature extreme value of the current node appears in the full life cycle historical data as the anchor point, extracting the internal resistance growth rate corresponding to the cumulative equivalent number of cycles from the anchor point, converting the internal resistance growth rate into Joule heat addition and superimposing it on the current compensation temperature of the current node to obtain the rollback temperature, and then recalculating the temperature gradient change rate between adjacent nodes with the rollback temperature. S3. Thermal runaway tracing step: Using the suspected node sequence as input, trace back whether the suspected node will spread heat to neighboring nodes in the next inspection cycle. If the spread exceeds the tracing threshold, a unique thermal runaway warning signal will be issued immediately.

[0006] Optionally, S1 includes calculating the set of temperature differences between all sampling points in the battery pack during the current inspection cycle and the previous inspection cycle, calculating the arithmetic mean of the set of temperature differences, and obtaining the average temperature drift value that characterizes the overall drift of the current ambient temperature.

[0007] Optionally, the average temperature drift value is subtracted from the original temperature value collected at each sampling point in the battery pack during the inspection cycle. The result is the compensated temperature value of the sampling point after environmental drift suppression. The compensated temperature values ​​of all sampling points together constitute the compensated temperature field.

[0008] Optionally, the construction of the gradient chain includes: determining the series and parallel conductive paths according to the electrical topology of the battery pack; mapping the temperature sampling points in the compensated temperature field to nodes on the series and parallel conductive paths; and connecting all nodes to form one or more continuous gradient chains according to the current flow sequence or potential relationship.

[0009] Optionally, the aging offset rollback operation includes performing the following operations sequentially for each node in the current gradient chain: Anchor point determination: retrieve the historical temperature data of the current node throughout its entire life cycle, identify its historical temperature extremes and the time when the extremes occurred, and use this time as the anchor point time for aging calculation; Accumulated aging amount acquisition: Calculate the cumulative equivalent number of cycles experienced by the current node from the anchor point time to the current time; based on the cumulative equivalent number of cycles, query the internal resistance growth rate corresponding to the current node through the pre-stored aging parameter mapping relationship; Temperature rewinding compensation: Based on the battery operating current and the internal resistance growth rate, the internal resistance growth is converted into a steady-state Joule heat addition caused by aging through a preset Joule heat calculation relationship; the Joule heat addition is superimposed on the current compensation temperature value of the node to obtain the rewinding temperature of the node.

[0010] Optionally, the rate of change of temperature gradient between adjacent nodes is calculated after the aging offset rollback operation. After the aging offset rollback operation is completed, the rollback temperature of each node is used to calculate the rate of change of temperature gradient between each pair of adjacent nodes in the gradient chain.

[0011] Optionally, each of the temperature gradient change rates is compared with an adaptive threshold obtained statistically from historical normal operating data; the corresponding nodes whose temperature gradient change rates exceed the adaptive threshold are marked as abnormal nodes and arranged according to their order in the gradient chain or the severity of the abnormality, and the output is the suspected node sequence.

[0012] Optionally, step S3 takes the suspected node sequence as input and sequentially selects each suspected node in the sequence as the current traceability source point; based on the physical arrangement of individual batteries or modules in the battery pack, it determines the node corresponding to at least one battery cell or module directly adjacent to the current traceability source point and defines it as the neighboring node of the current traceability source point.

[0013] Optionally, when the next inspection cycle arrives, the compensation temperature values ​​of the current traceable source point and all its neighboring nodes in the corresponding cycle are obtained; the compensation temperature difference between the current traceable source point and each neighboring node in the cycle is calculated respectively, and compared with the compensation temperature difference between the corresponding node pairs in the previous inspection cycle to obtain the change in temperature difference between each pair of nodes. The change is used to characterize the trend intensity of heat diffusion from the source point to the neighboring nodes.

[0014] Optionally, S3 further includes tracing judgment: if the change in temperature difference corresponding to at least one of the neighboring nodes exceeds a preset diffusion intensity threshold, it is determined that the current tracing source point has experienced heat diffusion; count the number of all neighboring nodes that have experienced heat diffusion, and if the number exceeds a preset diffusion range threshold, it is determined that the diffusion range exceeds the tracing threshold; when the tracing judgment condition is met, a unique, highest priority thermal runaway early warning signal is generated and issued.

[0015] The beneficial effects of this invention are: This invention constructs an average temperature drift model and performs uniform drift suppression processing on the temperature changes of all temperature sampling points in the current inspection cycle. This eliminates error interference caused by changes in ambient temperature, fluctuations in ventilation conditions, or the overall thermal inertia of the system, thereby obtaining a more realistic and stable compensated temperature field. On this basis, a gradient chain is further constructed along the electrical conductivity path, realizing the ability to identify local thermal anomalies without being affected by external disturbances, improving the sensitivity of temperature gradient abrupt change detection, and providing a data foundation for subsequent thermal runaway identification.

[0016] This invention proposes an aging offset rollback operation mechanism based on the temperature history of a node's lifecycle. By identifying extreme temperature anchor points, calculating the equivalent number of cycles, and mapping the internal resistance growth rate based on engineering calibration, the mechanism calculates the Joule heat gain and generates the rollback temperature. This mechanism is the first to quantitatively introduce the aging process of the battery cell into the temperature modeling process in thermal runaway early warning, accurately distinguishing between chronic temperature rises caused by aging and acute abrupt changes caused by anomalies, reducing false positives and false negatives. This node-level degradation modeling and temperature compensation strategy constitutes a personalized, evolution-driven thermal state perception framework, which is a key difference from traditional fixed threshold methods or static temperature judgment methods.

[0017] This invention establishes a set of physically neighboring nodes around a suspected node and dynamically compares the temperature difference evolution trend over two consecutive inspection cycles. It calculates the temperature difference change to form a thermal diffusion trend intensity index, and combines this with an adaptive diffusion intensity threshold and diffusion range threshold to complete a tracing process from anomaly identification to thermal diffusion tracking and then to quantitative judgment of risk spread. This overcomes the lack of thermal propagation behavior modeling in existing solutions, achieving a leap from point-like anomaly detection to area-like diffusion perception, ensuring that early warning signals are only triggered when a real thermal runaway risk occurs. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the early warning method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the gradient chain consistency verification steps in an embodiment of the present invention. Detailed Implementation

[0020] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. For some well-known technologies, those skilled in the art may also use other alternative methods to implement the invention. Moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0021] like Figures 1-2 As shown, a method for early warning of thermal runaway in a new energy vehicle battery pack based on temperature gradient compensation includes the following steps: S1. Gradient compensation step: Based on the temperature values ​​obtained by all sampling points in the battery pack in the previous inspection cycle, environmental drift suppression is performed on the temperature values ​​obtained in the current inspection cycle, and the compensated temperature field is output.

[0022] S1 aims to address the issue of suppressing environmental temperature interference during the early warning process of thermal runaway in new energy vehicle battery packs. In the operational monitoring of new energy vehicle battery packs, temperature is a core indicator for assessing the risk of thermal runaway. However, the external ambient temperature of the battery pack may drift over time, for example, due to changes in season, weather, or vehicle driving conditions, causing all sensor-measured temperature values ​​to collectively rise or fall. Directly using this raw temperature data for analysis may misjudge local thermal anomalies as normal drift or mask true signs of thermal runaway. Therefore, it is essential to first compensate for the influence of the global ambient temperature to ensure the accuracy of subsequent temperature gradient analysis.

[0023] Therefore, a temperature drift suppression mechanism based on the temperature difference of all sampling points is designed, which is carried out in three steps: First, the temperature changes of all temperature sampling points within the battery pack are recorded over two adjacent inspection cycles. Specifically, for each sampling point, the temperature difference between the current inspection cycle and the previous cycle is calculated. This set of differences comprehensively reflects the temperature change trend of the entire battery pack during that time period, including the overall impact of external temperature changes.

[0024] After obtaining the temperature differences at all sampling points, the method further calculates the arithmetic mean of these differences to obtain a scalar value called the average temperature drift value, which reflects the overall trend of the ambient temperature of the entire battery pack during this inspection cycle. The core idea is that if the temperature at all points is rising or falling to similar degrees, then this change is likely due to external environmental factors rather than localized abnormal heating.

[0025] After identifying the average drift, it is subtracted from the raw temperature value currently collected at each sampling point. That is, if the temperature at a sampling point rises by 1.5°C, but 1.2°C of that is due to ambient temperature rise, then only 0.3°C should be considered evidence of a potential local heat source at that point. After this compensation, the temperature value at each sampling point more accurately reflects its local state, removing noise caused by environmental factors. All compensated temperature values ​​are reorganized into a spatial distribution field, called the compensated temperature field. This temperature field preserves the true thermal anomaly distribution inside the battery while eliminating uniform external temperature disturbances.

[0026] This compensation mechanism avoids the misleading effect of environmental changes on temperature gradient analysis, and strengthens the specific signals generated by local temperature abrupt changes, such as internal short circuits or local heat accumulation, thus laying a data foundation for the next step of gradient chain construction and consistency verification.

[0027] The specific calculations are as follows: S11, Calculate the temperature difference set: Assume that the battery pack is equipped with... The temperature sampling point is recorded as the nth temperature sampling point. The sampling temperature at each sampling point during the current inspection cycle is: The temperature in the previous inspection cycle was Then the set of temperature difference values ​​is: ;in, Indicates the first The raw temperature values ​​collected at each sampling point during the current inspection cycle. Indicates the first The temperature value of each sampling point in the previous inspection cycle. Indicates the first Temperature difference between sampling points.

[0028] S12, Calculate the average temperature drift value: For the above... The arithmetic mean of the temperature differences is used to obtain the average temperature drift value, which represents the overall temperature drift of the current environment. ; This represents the average temperature difference across all sampling points, and the average temperature drift.

[0029] S13, perform drift compensation on the current temperature value of each sampling point to obtain the compensated temperature value: ; Indicates the first Compensated temperature values ​​of each sampling point after drift suppression; all Together they constitute the compensation temperature field .

[0030] S2. Gradient chain consistency verification step: Using the compensated temperature field as input, establish a gradient chain along the series and parallel conductive path of the battery pack, calculate the temperature gradient change rate between adjacent nodes in the gradient chain, compare the temperature gradient change rate with the preset adaptive threshold, and output the sequence of suspected nodes with thermal runaway suspicion. Before calculating the rate of change of temperature gradient, an aging offset rollback operation is introduced for each node in the gradient chain. This includes taking the moment when the temperature extreme value of the current node appears in the historical data of the entire life cycle as the anchor point, extracting the internal resistance growth rate corresponding to the cumulative equivalent number of cycles from the anchor point, converting the internal resistance growth rate into Joule heat addition and superimposing it on the current compensation temperature of the current node to obtain the rollback temperature, and then recalculating the rate of change of temperature gradient between adjacent nodes with the rollback temperature.

[0031] S21, Gradient Chain Construction: Based on the electrical topology of the battery pack, determine its series and parallel conductive paths. This includes compensating for the temperature field. Each temperature sampling point in the data is mapped to a node on the conductive path according to the direction of current flow or the order of potential levels. And form one or more continuous gradient chains. .

[0032] A battery pack typically consists of multiple cells arranged in a series-parallel configuration to meet the dual voltage and capacity requirements of new energy vehicles. A battery pack may consist of 100 cells, with 10 cells connected in parallel to form a battery pack, and 10 battery packs connected in series to form a complete high-voltage platform. A series path refers to the sequential flow of current through each individual cell during operation, resulting in a sequential voltage superposition. A parallel path refers to multiple cells sharing a current path, resulting in the same voltage but superimposed capacity. The overall system's conductive path is a composite structure of these two connection methods. Knowing this topology, the current flow sequence throughout the battery pack can be determined, and the potential hierarchy between each cell or module can be calculated, such as the voltage distribution path from the high-voltage end to the low-voltage end. This conductive path is not a spatial route of physical location but rather an electrical logical conduction channel; it serves as a reference framework for constructing the temperature gradient chain.

[0033] The sampling points are mapped to nodes on the conductive path: Battery packs typically have multiple temperature sampling points. These sensors may be installed at critical locations such as the cell surface, between cells, and at the connection points of battery modules to monitor temperature changes. Each sampling point can be considered a temperature observation unit, and the value it collects represents the thermal performance of the battery material at that location under current operating conditions. In this invention, we abstract these sampling points as nodes, that is, "electrically critical points with temperature observation capabilities" along the current conductive path. Each node has both a physical spatial location and an electrical sequence position in the topology. For example: Node V1: This may be the positive electrode temperature of the first cell on the high-voltage side. Node V5: This may be the connection temperature point of a parallel branch in the middle section; Node V10: This may be a temperature point near the busbar at the low-pressure end.

[0034] By mapping and pairing all sampling points with electrical conduction paths, they can be numbered and placed into the electrical path chain to form a logically continuous sequence of gradient nodes.

[0035] As current flows along the electrical path, each node (i.e., a temperature sampling point) can be considered as a thermal resistance point or a heat transfer point on that path. If the node values ​​in the compensated temperature field are arranged in the order of conductivity, a temperature gradient chain along the current flow direction can be formed. Assuming the battery pack electrical topology contains 10 series-connected cells with 10 temperature sampling points, corresponding to nodes V1 to V10, then the gradient chain can be represented as: The arrows here do not indicate the direction of data transmission, but rather the electrical connection sequence of the nodes and the direction of current flow. In complex structures, there may be multiple parallel branches or modular arrangements, forming multiple gradient chains. For example, each module may form a local gradient chain, which is then aggregated to form a global gradient chain across modules, or parallel modules may each form an independent gradient chain, which is then evaluated in parallel. Ultimately, each gradient chain reflects the changing trend of local and global temperature distribution under the current conductive path, providing a basic data channel for subsequent judgment of whether there are any abnormal abrupt changes in the local area.

[0036] Constructing a gradient chain based on the conductive path is one of the key points of this invention. Unlike traditional methods of dividing temperature analysis regions according to spatial location or lateral areas, this invention emphasizes a heat transfer logic structure formed by the current flow path. This is because thermal runaway is essentially caused by abnormalities within the battery cell or at connection points, leading to the localized generation of large amounts of Joule heat. The generation of Joule heat is closely related to the current path; therefore, to determine whether a thermal anomaly is abnormal, it is more beneficial to prioritize tracing temperature gradient changes along the current flow direction for high-precision early warning and location.

[0037] S22, Aging Offset Rewind Operation: In thermal runaway early warning of new energy vehicle battery packs, relying solely on the current temperature or temperature gradient is often insufficient to accurately determine potential risks. This is because battery cells undergo varying degrees of aging as their lifespan progresses, such as increased internal resistance and decreased thermal capacity. These aging differences lead to inconsistent temperature changes in different cells even under identical external conditions. Without differentiation, the system may misjudge a slow temperature rise caused by aging as a sign of thermal runaway, or mask true latent anomalies within normal fluctuations. Therefore, this invention proposes an aging offset rewind operation based on traditional gradient analysis. Its goal is to proactively reconstruct the thermal offset caused by aging within the battery cell before judging a temperature anomaly, thereby extracting a more accurate thermal runaway signal.

[0038] Each temperature sampling point (node) undergoes several high-temperature operations during the battery's lifespan. By accessing its historical temperature records, the extreme temperature values ​​experienced by that node are identified, and the time of these values ​​is used as the aging anchor point. This anchor point represents the earliest or most significant aging inflection point for the battery cell. From this moment on, its material properties (especially electrochemical internal resistance) may begin to deviate from the factory standard. Starting from this anchor point, we can track the entire aging evolution process. From the anchor point time, the equivalent number of cycles experienced by that node is counted, i.e., the standardized count of charge-discharge cycles per unit time. Combined with a preset internal resistance growth model, the current internal resistance growth rate is mapped. This growth rate quantitatively describes the degree of degradation of the battery cell and serves as a bridge connecting electrical aging and thermal behavior. Increased internal resistance leads to more Joule heat generated per unit current. This step calculates the additional heat generated due to aging, i.e., the extra heat energy, based on the current operating current and the increase in internal resistance. This heat is then converted into a temperature rise value, i.e., the thermal offset. Finally, this offset is superimposed on the current compensation temperature value to obtain the rewinding temperature. This rewinding temperature is not an actual measured value, but an equivalent temperature value that takes into account the thermal offset of aging, which more accurately reflects the thermal performance of the cell under physically equivalent conditions.

[0039] Specifically, S22 includes the gradient chain Each node in Perform the following operations in sequence: S221, Anchor Point Determination: Anchor point determination is the starting point of the entire aging offset rollback operation. Its purpose is to find the moment that best reflects the onset characteristics of degradation from the historical operating data of each temperature sampling node, for use in the subsequent calculation of cumulative aging. The essence of this operation is to provide an individualized aging reference point for each node, ensuring that the aging assessment has a physical basis, rather than mechanically starting from the initial time or a random moment. The battery aging process is driven by multiple factors, such as current, temperature, and cycle depth. Among them, high-temperature operation is a key factor that accelerates aging. During the operation of the battery throughout its entire life cycle, at certain moments, due to high load, ambient temperature rise, or failure of the thermal management system, the local cell temperature may reach a peak in a short period of time. These extreme moments are usually accompanied by stronger electrochemical reactions and material degradation, and therefore have the following characteristics: high temperature extremes are often the starting point for significant changes in cell state; compared with invisible electrochemical losses, temperature extremes are easier to be recorded by sensors for a long time and have high reusability; this moment, as an anchor point, facilitates subsequent cumulative modeling based on time span or cycle number. Therefore, this invention designs the moment when the historical temperature extreme value occurs as the aging anchor point moment for each node, that is, it is considered that from this moment on, the cell has entered a significant aging stage.

[0040] The specific solution is as follows: retrieve node Full lifecycle historical temperature data sequence Determine its historical temperature extremes: ; This represents the historical temperature extreme value of a node. The extreme temperature value, i.e., the maximum temperature point, is found among all recorded data for this node. If the sampling frequency is high, multiple approximate extreme values ​​may appear. The moment the first peak value is reached can be selected as the earliest observable evidence of aging initiation. To consider robustness, a moving average of the temperature series can be performed and the maximum value selected to avoid mistakenly selecting short-term noise peaks. The time point at which this extreme value occurs is recorded as the anchor point time. ; This indicates the time anchor point where historical temperature extremes occurred.

[0041] Traditional battery aging models typically use a uniform start time or number of charge cycles as a starting point, ignoring the differences in operating conditions between different cells. This invention, however, establishes anchor points at the node level, ensuring that the aging rollback operation of each cell is based on its own key historical nodes. By establishing these anchor points, subsequent analysis can focus on the accumulated charge-discharge behavior from the anchor point to the present, i.e., the equivalent number of cycles and the aging rate during that period, and can also calculate thermal offset, etc.

[0042] S222, Accumulated Aging Amount Acquisition: In S221, an aging anchor point has been determined for each node. However, the anchor point itself is only a starting point in time. It cannot directly tell us the current aging level of the node, whether it is mildly or severely aged compared to other nodes, or how much this aging will affect the current heat dissipation behavior. The core task of S222 is to transform the time span into a quantifiable, calculable, and thermally analyzeable aging index, that is, how many effective aging processes the node has undergone from the anchor point to the present. The reason why this invention does not use simple running time but uses equivalent cycle count is because running time cannot truly reflect aging. Battery aging is not necessarily older the longer the time, but is highly dependent on whether charging and discharging occur, the depth of charging and discharging, the current magnitude, and the temperature level. If only the time elapsed from the anchor point to the present is used, a problem will arise: long-term parking with almost no operation and short-term high-frequency high-power charging and discharging have little difference in time, but the degree of aging is completely different. Equivalent cycle count is a standardized indicator that converts various irregular charge and discharge behaviors into the equivalent number of complete standard charge and discharge cycles, which is closer to the real aging mechanism than time.

[0043] Therefore, S222 consists of two steps: S2221: Calculate the time of node self-anchor point Up to the current moment The equivalent number of loops experienced The calculation is as follows: a retrieves the continuous charge / discharge current of the node (cell or module) from the anchor point to the current time from the battery management system (BMS). ,Voltage SOC (State of Charge) or charge change data.

[0044] b. Calculate the depth of discharge (DoD) for each time segment: Divide the entire time axis into several time segments, and calculate the percentage of the cell's rated capacity corresponding to the change in charge (charging or discharging) within each segment, i.e.: ;in, Indicates the first Changes in the amount of electricity in the segment, This indicates the rated capacity of the cell at that node. Indicates the first The depth of discharge in the segment.

[0045] c squares the discharge depth of all the small segments sequentially and sums them up to obtain the equivalent number of complete cycles: This formula is a simplified model of the classic Rainflow-counting method in thin-cell battery life assessment, reflecting the cumulative damage effects of partial deep cycling. For example: One 100% depth discharge = one complete cycle; Four 50% discharges = one complete cycle; 25 discharges at 20% ≈ 1 complete cycle.

[0046] at this time, This is the cumulative equivalent number of cycles from the anchor point to the present, used to characterize the total degradation intensity of the node from the aging inflection point.

[0047] S2222: Obtain Then, combined with the preset internal resistance growth mapping function Query the corresponding internal resistance growth rate Used for subsequent Joule thermal rewind compensation: The internal resistance growth mapping function uses a piecewise linear function: ; in, This represents the growth slope at different stages, that is, how many milliohms the internal resistance increases with each additional equivalent cycle. In the initial stage, growth is slow, and a small value is usually set. Assuming a medium-term acceleration of the recession, For the later, rapid degradation stage, a large value is set. Specifically, an experimental calibration method can be used. Several samples of the same type of new battery cells are taken and subjected to life cycle tests at constant temperature and rate. The curves of their internal resistance changing with the number of cycles are recorded. A piecewise linear fitting method is used to determine the growth slope at different stages. The threshold segmentation point for the equivalent number of iterations. This represents the initial equivalent number of cycles required for aging to enter a significant growth phase. The initial equivalent number of cycles required to enter the rapid degradation zone during aging can be approximated as a percentage of the design life: If the cell's design life is 1000 complete cycles: take Cth1=300, indicating that mid-term degradation begins at approximately 30% of the lifespan; take Cth2=700, indicating that late-term degradation begins at approximately 70% of the lifespan. This indicates the increase relative to the factory condition. The constant is the intra-segment smoothing connection constant. Since it is a piecewise linear model, the function form of each segment is a linear line. If no processing is performed, numerical jumps or discontinuities may occur between different segments. Therefore, the smoothing connection constant is introduced. This ensures that each curve segment connects continuously at the segmentation point: Make the end of the first paragraph equal to the beginning of the second paragraph; : Make the end of the second paragraph equal to the beginning of the third paragraph.

[0048] S223, Temperature Rewind Compensation: Based on the current operating current. The relationship is calculated using the Joule heat gain: ;in, This is the inspection cycle time. Convert this additional heat load into a temperature rise compensation value (heat capacity is constant). : ; This indicates the additional Joule heat caused by aging. Add it to the current compensated temperature value to obtain the node. Rewinding temperature: .

[0049] Compared to traditional methods, this invention introduces the aging time dimension. Traditional temperature gradient analysis only makes judgments based on the temperature difference between two time points, without considering the characteristics of battery cell evolution over time. In contrast, this invention introduces historical temperature data throughout the entire battery lifecycle before gradient analysis, using aging anchor points as a starting point to model and compensate for the degradation behavior of nodes, making the judgment more temporally continuous and physically causal. It establishes a physical link between electrochemical aging and thermal anomalies, rather than simply performing a moving average or filtering of temperature. Instead, it starts from the physical source of aging (internal resistance) and combines it with current operating conditions such as current to quantitatively calculate the additional heat, demonstrating a deep-level cross-domain coupling modeling capability. By rewinding the temperature, it can reconstruct the temperature distribution that the battery should exhibit under ideal healthy conditions, and then compare it with the actual temperature difference. This helps to eliminate slow heating caused by aging, highlight sudden heating caused by anomalies, improve the accuracy of early warning, and avoid misjudgments or omissions.

[0050] S23, Gradient Change Rate Calculation and Comparison: After obtaining the rewind temperature of all nodes in the entire gradient chain... Then, calculate the rate of change of the temperature gradient between each pair of adjacent nodes: ;in, Represents a node and Geometric spacing along the actual deployment path.

[0051] Rate of change of each gradient Adaptive threshold obtained from statistics based on normal operating conditions Compare: This is an abnormal node.

[0052] S24, Suspect Sequence Generation: Generate all sequences that satisfy... nodes These nodes are marked as anomalous, forming a set of anomalous nodes. Based on their physical order in the gradient chain or the magnitude of the anomaly, the output is a sequence of suspect nodes: .

[0053] In the temperature gradient consistency analysis of this invention, the rate of temperature change between adjacent nodes in each gradient chain Each node in the gradient chain needs to be compared with a corresponding threshold to determine whether there is abnormal heat conduction at that point. During actual vehicle operation, the rate of change of temperature gradient under non-abnormal conditions is continuously recorded. Establish its historical distribution model and set the adaptive threshold as follows: ;in, For the first time under normal historical conditions The average temperature gradient of the nodes, For the corresponding standard deviation, The constant is used to adjust the sensitivity, and is taken between 1 and 2.

[0054] S3. Thermal runaway tracing step: Using the suspected node sequence as input, trace back whether the suspected node will spread heat to neighboring nodes in the next inspection cycle. If the spread exceeds the tracing threshold, a unique thermal runaway warning signal will be issued immediately.

[0055] S31, Node Traversal and Proximity Delineation: The core objective is to determine whether these suspected nodes have diffused heat to their surrounding structures. This involves constructing a heat diffusion discrimination range for this judgment process, which means identifying the monitoring points that are physically adjacent to each suspected node and establishing clear diffusion targets for subsequent temperature difference evolution analysis.

[0056] Specifically, based on the sequence of suspected nodes As input, select each suspect node in the sequence sequentially. As the current source of traceability.

[0057] A battery pack consists of multiple cells or modules arranged according to a specific physical structure, such as a two-dimensional grid, side-by-side array, or honeycomb structure. Each structural unit may have one or more temperature sensors. After the source point is determined, based on the cell or module's position in the physical structure, other cells that are directly adjacent to it in space are located, and the temperature sampling points corresponding to these adjacent cells are designated as their neighboring nodes. For example: If the cells are arranged in a row and column structure, then the left, right, top, and bottom cells are the neighboring nodes; If the module structure is a three-dimensional block shape, it may include six adjacent units in six directions of a hexahedron; If only some measuring points are set up, then only the nodes that actually have temperature sampling devices should be selected.

[0058] Ultimately, each traceable source will be bound to a set of neighboring nodes, forming a set of heat diffusion determination areas.

[0059] Based on the physical layout of the battery pack, identify the traceability point. The temperature sampling nodes corresponding to at least one directly adjacent cell or module constitute a neighboring node set: .

[0060] S32, Diffusion Trend Calculation: In S31, a neighboring observation area was established for each suspected thermal runaway node, which means knowing which adjacent units or modules may be affected by heat diffusion. The core of this step is not to look at the temperature difference at a single time point, but to analyze the evolution trend of the temperature difference.

[0061] The temperature data is continuously recorded for each inspection cycle. In the next inspection cycle, it is recorded as the number of cycles. When the cycle arrives, acquire the compensated temperature values ​​for two cycles of the trace source point and all its neighboring nodes, including: 1. Source point temperature: ; 2. Temperature of neighboring nodes: ,in .

[0062] Calculate the temperature difference between the trace source and each neighboring node: Current cycle temperature difference: ; The temperature difference of the previous cycle is denoted as the number of cycles. cycle: ; Temperature difference change: The change in temperature difference Used to characterize the heat traced from its source. to neighboring nodes The strength of the transmission trend.

[0063] like This indicates that the temperature of neighboring nodes is converging towards the source point, which means that the source point may be transferring heat outward. like This indicates that heat diffusion is not significant, which may be due to transient heat fluctuations or the heat not dissipating outwards.

[0064] The above comparison focuses on the amount of temperature difference change, rather than directly observing the temperature difference. Directly comparing temperature differences only tells us which temperature is higher, but it cannot determine whether this difference is evolving. Heat diffusion is essentially a dynamic process, and we cannot judge whether it is spreading based on a single moment. Comparing the amount of temperature difference change between two periods can reveal the direction and intensity of heat diffusion.

[0065] S33, Tracing Judgment: Set the preset diffusion intensity threshold as follows: If there exists a neighboring node that satisfies: If the neighboring node detects effective heat diffusion from the trace source, it is considered that the neighboring node has detected effective heat diffusion from the trace source. Essentially, this means that if the temperature difference between the source point and its neighboring nodes decreases by more than this value between two inspection cycles, heat is considered to be spreading. This threshold should not be too high or too low, otherwise it will lead to missed or false alarms. Based on the distribution of temperature difference changes for each pair of nodes under no-thermal-runaway conditions in a large amount of historical normal operation data, the upper quantile (95%) is selected as the threshold. ; Count the number of all neighboring nodes that meet the above conditions. and diffusion range threshold Compare the changes. If only a single neighboring node shows a change, it indicates a small diffusion range, possibly a localized thermal disturbance. If multiple nodes respond simultaneously, it indicates that the thermal diffusion has formed a spreading trend, with a higher risk level. Then the source point is determined. The resulting thermal diffusion exceeded the traceability threshold.

[0066] Based on the total number of neighboring nodes, if there are 6 neighboring nodes around a certain traceability point, the threshold is set to half or more of them, which can allow 1 to 2 nodes to be thermally coupled normally, but if the diffusion exceeds half, it constitutes a risk.

[0067] S34, Warning Trigger: Once any traceable source meets the diffusion range determination condition of S33, a unique and highest priority thermal runaway warning signal is immediately generated and issued, indicating that the current battery pack may be undergoing an uncontrollable heat diffusion process. Emergency measures should be taken immediately, including cutting off high voltage, starting active cooling, and triggering physical isolation.

[0068] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0069] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for early warning of thermal runaway in a new energy vehicle battery pack based on temperature gradient compensation, characterized in that, Includes the following steps: S1. Based on the temperature values ​​obtained from all sampling points in the battery pack in the previous inspection cycle, environmental drift suppression is applied to the temperature values ​​obtained in the current inspection cycle, and the compensated temperature field is output after compensation. S2. Using the compensated temperature field as input, establish a gradient chain along the series-parallel conductive path of the battery pack, calculate the rate of change of temperature gradient between adjacent nodes in the gradient chain, compare the rate of change of temperature gradient with the preset adaptive threshold, and output the sequence of suspected nodes with thermal runaway. Before calculating the temperature gradient change rate, an aging offset rollback operation is introduced for each node in the gradient chain. This includes taking the moment when the temperature extreme value of the current node appears in the full life cycle historical data as the anchor point, extracting the internal resistance growth rate corresponding to the cumulative equivalent number of cycles from the anchor point, converting the internal resistance growth rate into Joule heat addition and superimposing it on the current compensation temperature of the current node to obtain the rollback temperature, and then recalculating the temperature gradient change rate between adjacent nodes with the rollback temperature. The construction of the gradient chain includes: determining the series and parallel conductive paths according to the electrical topology of the battery pack; mapping the temperature sampling points in the compensated temperature field to nodes on the series and parallel conductive paths; and connecting all nodes to form one or more continuous gradient chains according to the current flow sequence or potential relationship. S3. Using the suspected node sequence as input, reverse track whether the suspected node will spread heat to neighboring nodes in the next inspection cycle. If the spread exceeds the traceability threshold, a unique thermal runaway warning signal will be issued immediately.

2. The method for early warning of thermal runaway of a new energy vehicle battery pack based on temperature gradient compensation according to claim 1, characterized in that, S1 includes calculating the set of temperature differences between all sampling points in the battery pack in the current inspection cycle and the previous inspection cycle, calculating the arithmetic mean of the set of temperature differences, and obtaining the average temperature drift value that represents the overall drift of the current ambient temperature.

3. The method for early warning of thermal runaway of a new energy vehicle battery pack based on temperature gradient compensation according to claim 2, characterized in that, The average temperature drift value is subtracted from the original temperature value collected at each sampling point in the battery pack during the inspection cycle. The result is the compensated temperature value of the sampling point after environmental drift suppression. The compensated temperature values ​​of all sampling points together constitute the compensated temperature field.

4. The method for early warning of thermal runaway of a new energy vehicle battery pack based on temperature gradient compensation according to claim 1, characterized in that, The aging offset rollback operation includes performing the following operations sequentially on each node in the current gradient chain: Anchor point determination: retrieve the historical temperature data of the current node throughout its entire life cycle, identify its historical temperature extremes and the time when the extremes occurred, and use this time as the anchor point time for aging calculation; Accumulated aging amount acquisition: Calculate the cumulative equivalent number of cycles experienced by the current node from the anchor point time to the current time; based on the cumulative equivalent number of cycles, query the internal resistance growth rate corresponding to the current node through the pre-stored aging parameter mapping relationship; Temperature rewinding compensation: Based on the battery operating current and the internal resistance growth rate, the internal resistance growth is converted into a steady-state Joule heat addition caused by aging through a preset Joule heat calculation relationship; the Joule heat addition is superimposed on the current compensation temperature value of the node to obtain the rewinding temperature of the node.

5. The method for early warning of thermal runaway of a new energy vehicle battery pack based on temperature gradient compensation according to claim 4, characterized in that, The rate of change of temperature gradient between adjacent nodes is calculated after the aging offset and rollback operation. After the aging offset and rollback operation is completed, the rollback temperature of each node is used to calculate the rate of change of temperature gradient between each pair of adjacent nodes in the gradient chain.

6. The method for early warning of thermal runaway of a new energy vehicle battery pack based on temperature gradient compensation according to claim 5, characterized in that, Each of the temperature gradient change rates is compared with an adaptive threshold obtained based on historical normal operating condition data. The nodes whose temperature gradient change rate exceeds the adaptive threshold are marked as anomalous nodes and arranged according to their order in the gradient chain or the severity of the anomalousness, and the output is the suspected node sequence.

7. The method for early warning of thermal runaway of a new energy vehicle battery pack based on temperature gradient compensation according to claim 1, characterized in that, S3 takes the suspected node sequence as input, and sequentially selects each suspected node in the sequence as the current traceability source point; based on the physical arrangement of individual batteries or modules in the battery pack, it determines the node corresponding to at least one battery cell or module directly adjacent to the current traceability source point, and defines it as the neighboring node of the current traceability source point.

8. The method for early warning of thermal runaway of a new energy vehicle battery pack based on temperature gradient compensation according to claim 7, characterized in that, When the next inspection cycle arrives, obtain the compensation temperature values ​​of the current traceability source point and all its neighboring nodes in the corresponding cycle; The compensation temperature difference between the current traceable source point and each neighboring node in the current cycle is calculated and compared with the compensation temperature difference between the corresponding node pairs in the previous inspection cycle to obtain the change in temperature difference between each pair of nodes. The change is used to characterize the trend intensity of heat diffusion from the source point to the neighboring nodes.

9. A method for early warning of thermal runaway in a new energy vehicle battery pack based on temperature gradient compensation according to claim 8, characterized in that, The S3 further includes a traceability judgment: if the change in temperature difference corresponding to at least one of the neighboring nodes exceeds a preset diffusion intensity threshold, it is determined that the current traceability source point has experienced heat diffusion; the number of all neighboring nodes that have experienced heat diffusion is counted, and if the number exceeds a preset diffusion range threshold, it is determined that the diffusion range exceeds the traceability threshold. When the conditions for retrospective judgment are met, a unique, highest-priority thermal runaway early warning signal is generated and issued.

Images