An immersed battery pack thermal safety monitoring system and method

Abstract

This invention discloses a thermal safety monitoring system and method for an immersion battery pack. The system includes: a battery pack housing containing multiple battery cells or modules filled with an insulating cooling immersion fluid; multiple pressure sensors located between the battery cells or between the battery cells and structural components; multiple voltage sensors and multiple temperature sensors; and a wireless battery management system including multiple wireless acquisition slave nodes and a wireless master node. Each slave node is connected to a corresponding sensor, and the master node receives signals and performs thermal safety monitoring via a wireless communication protocol. The method includes: establishing a rolling average and rolling standard deviation of a dynamic baseline based on pressure signals from the previous N cycles; calculating the deviation multiple of the real-time pressure signal and executing multi-level early warnings; and performing a fusion judgment based on pressure, temperature, and voltage signals. This invention achieves early warning of thermal runaway and improves the system reliability in immersion environments.

Description

Technical Field

[0001] This invention relates to the field of battery thermal safety management technology, and in particular to an immersion battery pack thermal safety monitoring system and method. Background Technology

[0002] Immersion liquid-cooled battery packs have become a cutting-edge direction in the field of battery thermal management due to their advantages of high efficiency, uniform temperature, and high energy density. In typical existing technical solutions, the cells are densely arranged and completely immersed in an insulating cooling immersion liquid (such as silicone oil, fluorinated liquid, etc.). The battery management system (BMS) mostly adopts a traditional wired architecture, connecting the voltage and temperature acquisition points of each cell through wiring harnesses.

[0003] However, existing technologies have the following drawbacks:

[0004] First, thermal runaway warnings are delayed. Battery thermal runaway is an inside-out process, with internal gas production and swelling preceding significant temperature rise. Current technologies rely solely on voltage and temperature monitoring, typically triggering alarms only in the later stages of thermal runaway. This leaves insufficient response time for safety measures, making early warning impossible.

[0005] Second, there are reliability issues with wired BMS in immersion environments. In a sealed environment filled with cooling immersion fluid, the numerous voltage and temperature acquisition harnesses require complex sealed connectors, which significantly increases the complexity of system design, manufacturing costs, and the risk of short circuits caused by leakage of cooling immersion fluid due to seal aging.

[0006] Third, there is a lack of in-situ sensing of the internal state of the battery cell. Voltage and temperature are external parameters and cannot directly reflect early signs of failure such as gas generation due to side reactions and increased internal pressure, resulting in insufficient accuracy and lead time for early warnings.

[0007] Therefore, how to achieve earlier and more sensitive thermal runaway warning in immersion battery packs, while ensuring the high reliability and easy deployment of the monitoring system in liquid environments, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0008] To address the issues of delayed thermal runaway early warning, poor reliability of wired BMS in immersion environments, and lack of in-situ sensing of the internal state of the battery cells in existing technologies, this invention proposes an immersion battery pack thermal safety monitoring system and method.

[0009] The specific technical solution is as follows:

[0010] A submersible battery pack thermal safety monitoring system includes:

[0011] A battery pack housing, which contains multiple battery cells or battery modules and is filled with an insulating cooling immersion liquid, wherein the battery cells or battery modules are at least partially immersed in the insulating cooling immersion liquid;

[0012] Multiple pressure sensors, each pressure sensor is positioned between battery cells or between a battery cell and a structural component;

[0013] Multiple voltage sensors and multiple temperature sensors are respectively installed on each cell or battery module;

[0014] Wireless battery management system, including:

[0015] Multiple wireless acquisition slave nodes, each of which corresponds to at least one cell in each battery module and is connected to the corresponding pressure sensor, voltage sensor and temperature sensor;

[0016] A wireless master node, connected to each of the aforementioned wireless acquisition slave nodes via a wireless communication protocol, is used to receive signals sent by each wireless acquisition slave node and perform thermal safety monitoring. By introducing pressure sensors and a wireless BMS architecture into the submersible battery pack, and using the inter-cell compression pressure as the core monitoring parameter, early signs of thermal runaway (gas production, bulging) can be directly detected, with a significantly earlier warning time than traditional temperature monitoring. At the same time, wireless communication eliminates the data harness inside the pack, avoiding the risk of sealing failure and short circuits caused by wired connections in the immersion environment, thus improving the reliability and safety of the system.

[0017] Furthermore, the pressure sensor is a thin-film pressure sensor, which is clamped or attached between the large surfaces of adjacent battery cells. The thin-film pressure sensor has a very small thickness, so it does not affect the tight arrangement and heat dissipation of the battery cells; specifically, it can be arranged in the middle area between the large surfaces of adjacent battery cells, which can sensitively capture the changes in compressive pressure caused by internal gas generation or expansion of the battery cells, thereby improving the sensitivity and accuracy of signal acquisition.

[0018] Furthermore, the wireless acquisition slave node is configured to operate in an environment immersed in an insulating cooling immersion fluid, and its housing material is chemically compatible with the insulating cooling immersion fluid.

[0019] A method for monitoring the thermal safety of an immersion battery pack includes the following steps:

[0020] Dynamic baseline establishment steps: During the normal cyclic operation of the battery pack, for each monitoring point, the rolling average and rolling standard deviation of the pressure signal are calculated using historical pressure signal data from the N complete charge-discharge cycles prior to the current moment as a sliding window; where N is a preset positive integer, N≥2;

[0021] Real-time monitoring and deviation calculation steps: acquire the real-time pressure signal within the current cycle, and calculate the deviation multiple of the real-time pressure signal relative to the rolling average value;

[0022] Multi-level early warning steps: Based on the preset threshold range into which the deviation multiple falls, trigger the corresponding level of early warning signal;

[0023] Fusion Judgment Steps: Upon triggering an early warning, the system comprehensively considers the abnormal states of at least two of the current pressure, temperature, and voltage signals, performs weighted or logical operations, and outputs the final risk level. This allows for adaptation to battery aging and changes in operating conditions, avoiding false alarms or missed alarms caused by fixed thresholds. By using pressure signal deviation multiple-level early warning and combining multi-physical quantity fusion judgment, the system can improve the lead time and accuracy of thermal runaway early warnings, providing more time for proactive safety intervention.

[0024] Furthermore, the rolling average μx(k) and rolling standard deviation δX(k):

[0025] ;

[0026] ;

[0027] Where X(i) is the effective mean or characteristic value of the pressure signal in the i-th cycle, k is the current cycle number, and N is the set baseline learning window size.

[0028] Furthermore, in the multi-level early warning steps, the preset threshold range includes:

[0029] When the deviation multiple is greater than or equal to 1 and less than 2, a level one warning is triggered;

[0030] When the deviation multiple is greater than or equal to 2 and less than 3, a level 2 warning is triggered;

[0031] When the deviation multiple is greater than or equal to 3, a level 3 warning is triggered.

[0032] Furthermore, the multi-level early warning process also includes an independent early warning condition: when the pressure signal decreases by more than a preset sudden drop threshold within a unit of time, a level three early warning is directly triggered. The sudden drop threshold corresponds to the pressure drop caused by the abnormal opening of the battery cell safety valve.

[0033] Furthermore, in the fusion judgment step, the risk index K is calculated using the comprehensive index method:

[0034] ;

[0035] Where δVpt and δVp are the current rate of pressure change and the maximum rate of pressure change corresponding to the current SOC of the system settings, respectively, and k1 is the pressure change factor; δVtt and δVt are the current rate of temperature change and the maximum rate of temperature change corresponding to the current SOC of the system settings, respectively, and k2 is the temperature change factor; δVvt and δVv are the current rate of voltage change and the maximum rate of voltage change corresponding to the current SOC of the system settings, respectively, and k3 is the voltage change factor;

[0036] Based on the different ranges in which the K value falls, risks are classified as low-level, medium-level, and high-level.

[0037] Furthermore, in the fusion judgment step, a signal threshold method is adopted: if only one signal exceeds its corresponding normal range, it is judged as low risk; if two signals exceed their corresponding normal range, it is judged as medium risk; if three or more signals exceed their corresponding normal range, it is judged as high risk.

[0038] An immersion battery pack includes a thermal safety monitoring system for the immersion battery pack.

[0039] The above technical solution has the following advantages or technical effects:

[0040] 1. By introducing inter-cell pressure sensors and a wireless BMS architecture, this invention enables in-situ and direct sensing of early signs of thermal runaway in immersion battery packs. The warning time is significantly earlier than that of traditional temperature monitoring, providing more time for proactive safety measures.

[0041] 2. This invention adopts a pure wireless communication scheme of wireless acquisition slave nodes and wireless master control nodes, which eliminates the data harness inside the battery pack, solves the problems of complex sealing, high leakage risk and poor reliability of wired BMS in immersion environment, and simplifies the structure and manufacturing process.

[0042] 3. The dynamic baseline multi-level early warning algorithm based on a rolling window proposed in this invention can adapt to battery aging and changes in operating conditions. It provides early warning by pressure deviation multiple, and combines independent criteria for sudden pressure drop and multi-signal fusion judgment to improve the lead time, accuracy and anti-interference ability of thermal runaway early warning. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of the system structure of the present invention;

[0044] Figure 2 This is an exploded view of the pressure sensor arrangement of the present invention;

[0045] Figure 3 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0046] To make the technical solution of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0047] Example 1

[0048] An immersion battery pack thermal safety monitoring system includes: a battery pack housing, multiple battery cells or battery modules, an insulating cooling immersion liquid, multiple pressure sensors, multiple voltage sensors, multiple temperature sensors, and a wireless battery management system.

[0049] The battery pack casing contains multiple battery cells or battery modules. For example... Figure 1 As shown, the battery cells or modules can be arranged in a grid pattern, a cross pattern, or other compact configurations. The battery pack casing is filled with an insulating cooling immersion fluid, and the battery cells or modules are at least partially immersed in this fluid. The insulating cooling immersion fluid can be silicone oil, mineral oil, fluorinated liquid, or other liquids with insulating and good thermal conductivity properties. To ensure long-term chemical stability, all components in contact with the cooling immersion fluid (including the cell casing, housing, sensor encapsulation, adhesives, etc.) are made of materials compatible with the cooling immersion fluid to prevent corrosion or precipitation.

[0050] like Figure 2 As shown, multiple pressure sensors are arranged between adjacent battery cells or between a battery cell and a module structural component. Each pressure sensor is used to sense changes in compressive pressure caused by the expansion of the battery cell volume or internal gas generation. In this embodiment, the pressure sensors are preferably thin-film pressure sensors, which have a very small thickness and will not significantly affect the tight arrangement and heat dissipation of the battery cells. Each thin-film pressure sensor is clamped or attached between the large surfaces of adjacent battery cells and is located in the middle region of the battery cell height direction. The middle region is the most sensitive location for battery cell expansion and deformation, and can more sensitively capture pressure changes. The number and specific arrangement of pressure sensors can be optimized according to the size of the battery module and the characteristics of the battery cells. Generally, a uniform arrangement is adopted to ensure that each battery module has at least one corresponding pressure monitoring point. The number and arrangement of pressure sensors are not limited to between battery cells; they can also be placed between the module and the housing, or between the battery cell and the end plate, as long as they can sense the pressure changes caused by the expansion of the battery cells.

[0051] In addition, a voltage sensor and a temperature sensor are respectively installed on each cell or battery module. The voltage sensor is used to collect the terminal voltage of the cell in real time, and the temperature sensor is used to collect the temperature of the cell surface or the tab in real time. The specific types of voltage and temperature sensors can be conventional voltage detection circuits and negative temperature coefficient thermistors or thermocouples.

[0052] The wireless battery management system comprises multiple wireless acquisition slave nodes and one wireless master node. Each wireless acquisition slave node corresponds to at least one cell in each battery module and is electrically connected to the pressure sensor, voltage sensor, and temperature sensor corresponding to that cell or module. The wireless acquisition slave node integrates an analog-to-digital converter (ADC) to convert the analog signals output from the pressure, voltage, and temperature sensors into digital signals. Each slave node also includes a wireless communication unit and a power management unit. The power management unit can draw power from within the battery pack (e.g., from the corresponding cell) or be self-powered (e.g., using a coin cell). The slave node's housing material is configured to be chemically compatible with the insulating cooling immersion fluid and has excellent sealing properties, ensuring stable operation of the slave node in environments where it is immersed in the insulating cooling immersion fluid for extended periods.

[0053] The wireless master node connects to each wireless acquisition slave node via a wireless communication protocol. This protocol can be Bluetooth Mesh, Zigbee, or a dedicated ISM band protocol. The wireless master node receives pressure, voltage, and temperature signals from all slave nodes and performs status assessments and early warnings based on a pre-defined thermal safety monitoring strategy. The wireless master node also connects to the vehicle controller or energy storage system controller via a wired communication interface (such as a CAN interface) to upload early warning information and battery status data to the upper-level system.

[0054] There are no wired data harnesses inside the battery pack housing for transmitting pressure, voltage, and temperature signals. All signals are transmitted wirelessly from the acquisition slave node to the master node, thus avoiding the leakage risks and reliability issues associated with using numerous wiring harnesses and sealed connectors in immersion environments.

[0055] Example 2

[0056] like Figure 3 As shown, a method for monitoring the thermal safety of an immersion battery pack based on the above system is presented. This method mainly includes four core steps: dynamic baseline establishment, real-time monitoring and deviation calculation, multi-level early warning, and fusion judgment. These steps are described in detail below.

[0057] Dynamic baseline establishment

[0058] During normal battery pack operation, the system maintains a dynamic baseline for each monitoring point (e.g., the pressure monitoring point between each cell). This baseline is updated based on a sliding historical data window. The window contains historical pressure signal data from N complete charge-discharge cycles prior to the current moment. N is a preset positive integer, N≥2, and its value can be adjusted according to the battery aging rate. In this embodiment, N is between 3 and 5, preferably N=3.

[0059] Specifically, for any monitoring parameter X (in this embodiment, X represents the pressure signal P), its rolling average μx(k) and rolling standard deviation δX(k) at the end of the kth cycle are:

[0060] ;

[0061] ;

[0062] Where X(i) is the effective mean or characteristic value of the pressure signal in the i-th cycle, k is the current cycle number, and N is the set baseline learning window size. Through this method, the system can track the fluctuation range of the pressure signal in real time during the normal aging process, avoiding the maladaptive effects caused by using a fixed threshold.

[0063] In this embodiment, the dynamic baseline can also establish its own rolling average and rolling standard deviation for the temperature signal and voltage signal respectively, for subsequent fusion judgment.

[0064] Real-time monitoring and deviation calculation

[0065] During actual operation of the battery pack, the wireless acquisition slave node collects the real-time pressure signal Xreal(m) (where m is the current cycle number) within the current cycle and transmits it wirelessly to the master control node. The master control node calculates the deviation factor Z of this real-time pressure signal relative to the dynamic baseline:

[0066] ;

[0067] in, Based on the number The rolling average and standard deviation are calculated from historical data up to a certain period. The deviation multiple reflects the degree to which the current pressure value deviates from the normal fluctuation range, and corresponds to the alarm triggering conditions, such as the range setting of Zx(m), which can be defined according to the product requirements.

[0068] Multi-level early warning

[0069] The system triggers a warning signal of the corresponding level based on the preset threshold range into which the deviation multiple Zx(m) falls. This embodiment sets up three levels of warnings:

[0070] When the deviation multiple is greater than or equal to 1 and less than 2, a Level 1 warning (alert) is triggered. At this time, the system issues an alert signal, indicating that the cell status may have a slight deviation, and maintenance personnel are advised to pay attention.

[0071] When the deviation multiple is greater than or equal to 2 and less than 3, a level 2 warning is triggered. The system issues a warning signal and may optionally activate local enhanced cooling or limit the charging and discharging power of the cell.

[0072] When the deviation multiple is greater than or equal to 3, a Level 3 warning (critical alarm) is triggered. The system determines that the risk of thermal runaway is extremely high, immediately issues the highest level alarm, and executes preset safety strategies, such as system power failure and activation of the entire area fire suppression system.

[0073] The above threshold ranges (1σ, 2σ, 3σ) can be adjusted appropriately according to the specific characteristics of the battery cell and product requirements.

[0074] Furthermore, this embodiment also sets an independent early warning condition: when the pressure signal decreases by more than a preset sudden drop threshold within a unit time, a level three early warning is directly triggered. This sudden drop threshold corresponds to the pressure drop characteristic caused by the abnormal opening of the cell safety valve. For example, when the pressure sensor detects that the pressure drops by more than a certain set value (e.g., 10 kPa) within a very short time (e.g., 0.5 seconds), the system directly determines that the risk of thermal runaway is extremely high, without waiting for the deviation multiple calculation.

[0075] In addition to the multi-level early warning based on pressure signals mentioned above, this system also retains independent monitoring and early warning mechanisms for temperature and voltage signals as a supplement and verification of pressure early warning.

[0076] Temperature warning

[0077] The system monitors the temperature signal of each cell or module in real time. Temperature warning mainly includes two aspects: absolute temperature threshold and temperature rise rate.

[0078] For absolute temperature thresholds, the system presets high-temperature alarm thresholds (e.g., 60℃ for Level 1 warning, 70℃ for Level 2 warning, and 80℃ for Level 3 warning) and low-temperature alarm thresholds (e.g., 0℃). When the real-time temperature exceeds the corresponding threshold, the corresponding level of temperature warning is triggered.

[0079] For the temperature rise rate, the system calculates the rate of temperature change over time in real time. During the normal operating period, the formula for calculating the temperature rise rate δT is:

[0080] (T is temperature, t is time);

[0081] Where T1 and T2 are the temperature values ​​at times t1 and t2, respectively. A temperature warning is triggered when the rate of temperature rise exceeds a preset threshold (e.g., 1℃ / s).

[0082] During the period when the cooling unit is on, the rate of temperature rise may differ from normal operating conditions. For example, when the battery temperature reaches a specific value (such as 20°C or 25°C), the liquid cooling unit starts cooling, at which point the temperature may drop rapidly or the rate of temperature rise may slow down. In low-temperature environments, the cooling unit may start heating, and the rate of temperature rise may increase abnormally. Therefore, when calculating the rate of temperature rise and determining whether to trigger an alarm, this system simultaneously acquires the operating status of the cooling unit. If the cooling unit is operating, the system will make a judgment based on a preset compensation coefficient or an adjusted threshold to avoid false alarms caused by normal operation of the cooling unit. Specifically, the system can pre-calibrate the normal temperature rise rate range when the cooling unit is on, and only trigger an alarm when the actual temperature rise rate exceeds this range.

[0083] Voltage warning

[0084] The system monitors the voltage signal of each battery cell in real time. Voltage warnings are based on the battery's own charging and discharging characteristics.

[0085] During the battery's plateau period (typically between 20% and 90% SOC), voltage changes are relatively gradual. Within this range, if an abnormal voltage jump occurs (e.g., a drop exceeding a set value in a short period) or the voltage change rate exceeds a preset threshold, a voltage warning is triggered.

[0086] In the final stages of battery charging and discharging (SOC below 20% or above 90%), voltage changes are rapid, which is normal. Therefore, the system dynamically adjusts the voltage warning threshold based on the current SOC. Specifically, the system pre-stores the normal voltage change rate ranges corresponding to different SOC intervals, and triggers a voltage warning only when the actual voltage change rate exceeds this range.

[0087] Voltage warnings can be issued independently or integrated with pressure and temperature warnings to improve the overall accuracy and reliability of the warnings.

[0088] Fusion judgment

[0089] When an early warning is triggered, the system considers the abnormal states of at least two of the current pressure, temperature, and voltage signals, performs weighted or logical operations, and outputs the final risk level. This embodiment provides two fusion judgment methods.

[0090] Method 1: Signal Threshold Method

[0091] The system sets independent normal ranges for pressure, temperature, and voltage. When the real-time value of a signal exceeds its corresponding normal range, the signal is considered abnormal. The risk level is then determined based on the number of abnormal signals.

[0092] If only one signal exceeds the normal range, it is judged as a low-risk level.

[0093] If two signals are outside the normal range, the risk level is determined to be medium.

[0094] If three or more signals are outside the normal range, it is considered a high-risk situation.

[0095] The signal threshold method is simple to implement and requires little computation, making it suitable for embedded systems that are sensitive to cost or computing power.

[0096] Method 2: Comprehensive Index Method

[0097] The system calculates the comprehensive risk index K by weighting and fusing the rates of change of pressure, temperature, and voltage signals. The formula for calculating the K value is:

[0098] ;

[0099] Where δVpt and δVp are the current rate of pressure change and the maximum rate of pressure change corresponding to the current SOC of the system settings, respectively, and k1 is the pressure change factor; δVtt and δVt are the current rate of temperature change and the maximum rate of temperature change corresponding to the current SOC of the system settings, respectively, and k2 is the temperature change factor; δVvt and δVv are the current rate of voltage change and the maximum rate of voltage change corresponding to the current SOC of the system settings, respectively, and k3 is the voltage change factor;

[0100] Risk levels are categorized into low, medium, and high risks based on the different ranges in which the K value falls. Specifically, three thresholds A1, A2, and A3 (A1 < A2 < A3) can be set: low risk occurs when A1 ≤ K < A2, medium risk when A2 ≤ K < A3, and high risk when K ≥ A3. The specific values ​​of A1, A2, and A3 can be determined based on product definition and experimental data.

[0101] The comprehensive index method can more comprehensively assess the risk of battery thermal runaway, and the weighting factors can be flexibly adjusted, improving the intelligence and accuracy of early warning.

[0102] In practical applications, one of the two methods mentioned above can be selected as needed, or both methods can be used simultaneously and the results compared. Furthermore, other signals, such as smoke concentration and gas composition, can be added to the fusion judgment to further improve the reliability of the early warning.

[0103] Example 3

[0104] An immersion battery pack includes the immersion battery pack thermal safety monitoring system described in Example 1. Specifically, the battery pack includes a housing, multiple battery cells or modules, an insulating cooling immersion fluid, a pressure sensor network, voltage and temperature sensors, wireless acquisition slave nodes, and a wireless master node. The connection relationships and functions of all components are the same as in Example 1. This battery pack can be applied to electric vehicles or energy storage systems. Due to the integration of the aforementioned thermal safety monitoring system, the battery pack can provide early warning of thermal runaway and possesses high reliability and maintenance-free characteristics.

[0105] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. An immersion battery pack thermal safety monitoring system, characterized by, include: A battery pack housing, which contains multiple battery cells or battery modules and is filled with an insulating cooling immersion liquid, wherein the battery cells or battery modules are at least partially immersed in the insulating cooling immersion liquid; Multiple pressure sensors, each pressure sensor is positioned between battery cells or between a battery cell and a structural component; Multiple voltage sensors and multiple temperature sensors are respectively installed on each cell or battery module; Wireless battery management system, including: Multiple wireless acquisition slave nodes, each of which corresponds to at least one cell in each battery module and is connected to the corresponding pressure sensor, voltage sensor and temperature sensor; A wireless master node is connected to each of the wireless acquisition slave nodes via a wireless communication protocol, and is used to receive signals sent by each wireless acquisition slave node and perform thermal safety monitoring.

2. The immersion battery pack thermal safety monitoring system of claim 1, wherein, The pressure sensor is a thin-film pressure sensor, and it is clamped or attached between the large surfaces of adjacent battery cells.

3. The immersion battery pack thermal safety monitoring system of claim 1, wherein, The wireless acquisition slave node is configured to operate in an environment immersed in an insulating cooling immersion fluid, and its housing material is chemically compatible with the insulating cooling immersion fluid.

4. A method for monitoring the thermal safety of an immersion battery pack, characterized in that, Includes the following steps: Dynamic baseline establishment steps: During the normal cyclic operation of the battery pack, for each monitoring point, the rolling average and rolling standard deviation of the pressure signal are calculated using historical pressure signal data from the N complete charge-discharge cycles prior to the current moment as a sliding window; where N is a preset positive integer, N≥2; Real-time monitoring and deviation calculation steps: acquire the real-time pressure signal within the current cycle, and calculate the deviation multiple of the real-time pressure signal relative to the rolling average value; Multi-level early warning steps: Based on the preset threshold range into which the deviation multiple falls, trigger the corresponding level of early warning signal; Fusion judgment steps: When an early warning is triggered, the abnormal states of at least two of the current pressure, temperature and voltage signals are considered, and a weighted or logical operation is performed to output the final risk level.

5. The method of claim 4, wherein the method further comprises: The rolling average μx(k) and rolling standard deviation δX(k): ； ； Where X(i) is the effective mean or characteristic value of the pressure signal in the i-th cycle, k is the current cycle number, and N is the set baseline learning window size.

6. The method of claim 4, wherein the method further comprises: In the multi-level early warning steps, the preset threshold range includes: When the deviation multiple is greater than or equal to 1 and less than 2, a level one warning is triggered; When the deviation multiple is greater than or equal to 2 and less than 3, a level 2 warning is triggered; When the deviation multiple is greater than or equal to 3, a level 3 warning is triggered.

7. The method of claim 4, wherein the method further comprises: The multi-level early warning process also includes an independent early warning condition: when the pressure signal drops by more than a preset sudden drop threshold per unit time, a level three early warning is directly triggered. The sudden drop threshold corresponds to the pressure drop caused by the abnormal opening of the battery cell safety valve.

8. The method for monitoring the thermal safety of an immersion battery pack according to claim 4, characterized in that, In the fusion judgment step, the risk index K is calculated using the comprehensive index method: ； Where δVpt and δVp are the current rate of pressure change and the maximum rate of pressure change corresponding to the current SOC of the system settings, respectively, and k1 is the pressure change factor; δVtt and δVt are the current rate of temperature change and the maximum rate of temperature change corresponding to the current SOC of the system settings, respectively, and k2 is the temperature change factor; δVvt and δVv are the current rate of voltage change and the maximum rate of voltage change corresponding to the current SOC of the system settings, respectively, and k3 is the voltage change factor; Based on the different ranges in which the K value falls, risks are classified as low-level, medium-level, and high-level.

9. The method for monitoring the thermal safety of an immersion battery pack according to claim 4, characterized in that, In the fusion judgment step, the signal threshold method is adopted: if only one signal exceeds its corresponding normal range, it is judged as low risk; if two signals exceed their corresponding normal range, it is judged as medium risk; if three or more signals exceed their corresponding normal range, it is judged as high risk.

10. An immersion battery pack, characterized by, The immersion battery pack thermal safety monitoring system includes any one of claims 1 to 3.

Images