An electrochemical energy storage liquid hybrid submerged guard distributed control system

Abstract

The present application relates to the technical field of distribution control, and particularly relates to a kind of electrochemical energy storage's liquid integration immersed protection distribution control system, including physical partition module, composite sensing module, feature extraction and authentication module, hierarchical response and flow resistance reconstruction module and failure protection module, through acoustic-optic dual-mode composite sensing collection immersion liquid acoustics and optical parameters, four-layer progressive authentication verification extracts early abnormal characteristics of thermal runaway, combined with flow-opening-flow resistance quantitative relationship dynamic adjustment electromagnetic proportional valve opening, reconstructs flow distribution to realize targeted cooling or thermal spread block, while setting failure protection module to realize global bubble exhaust and standby monitoring.The present application determines the reasonable value of threshold, reference value and coefficient through multiple tests and completes calculation, can early thermal runaway warning node, and reduces false alarm rate, gives consideration to the safety protection and continuous stable operation of energy storage system, and all parameters and structures are adapted to industrial production landing demand.

Description

Technical Field

[0001] This invention relates to the field of distributed control technology, specifically to a liquid-integrated immersion protection distributed control system for electrochemical energy storage. Background Technology

[0002] Electrochemical energy storage, as a crucial component of new power systems, relies heavily on lithium-ion batteries due to their high energy density and superior charge-discharge efficiency. Liquid cooling technology, particularly immersion liquid cooling, has become a core thermal management solution for large-capacity energy storage battery packs due to its high heat exchange efficiency and uniform cell heat dissipation. Immersion liquid cooling completely submerges the battery cluster in an insulating cooling medium, utilizing the medium's convective heat transfer for heat transfer. Its core technical challenge lies in early warning and precise protection against thermal runaway. Battery thermal runaway begins with the thermal decomposition of the SEI film. This stage only generates micron-sized microbubbles and localized weak thermal convection, without significant temperature rise or voltage fluctuations. Existing monitoring and protection technologies for immersion liquid cooling systems are mostly designed for the significant characteristics of the later stages of thermal runaway, making early identification and targeted intervention difficult. This has become a key issue restricting the safe operation of energy storage systems.

[0003] The existing technology is Chinese invention patent publication number CN119627291A, entitled "A Modular Immersion Liquid-Cooled Energy Storage Battery Thermal Management System with an Inverse Opal Structure Surface." This solution improves heat exchange area and efficiency through a modular immersion chamber design combined with an inverse opal structure heat dissipation surface. It monitors the thermal state based on parameters such as coolant inlet and outlet temperatures and individual battery cell voltages, and achieves full-area liquid cooling control by adjusting the total flow rate of the cooling pump. The core drawback of this solution is: Firstly, thermal runaway monitoring relies solely on conventional electrophysical parameters such as temperature and voltage, which cannot capture early warning signs such as microbubbles and thermal convection during the thermal decomposition stage of the SEI film. It can only provide early warning after the cell shows a significant temperature rise, resulting in a serious lag in the warning and missing the best opportunity to deal with thermal runaway. Secondly, the coolant flows in a series in a co-current manner in each modular immersion chamber, without an independent flow regulation and flow resistance reconstruction mechanism. It is impossible to achieve targeted flow distribution for local abnormal areas, the cooling efficiency is limited by the overall flow field, and it cannot effectively prevent the high-temperature fluid and thermal runaway products in abnormal areas from spreading to healthy areas, which can easily lead to a chain of thermal spread. Third, there is no dedicated signal authentication and multimodal verification mechanism. The collected monitoring signals are easily affected by non-thermal runaway factors such as cooling pump speed fluctuations, pipeline bubbles, and mechanical vibrations, which can easily lead to false judgments and false alarms, affecting the accuracy and reliability of the system protection. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a liquid-integrated immersion protection distribution control system for electrochemical energy storage.

[0005] This invention provides a liquid-based immersion protection distribution control system for electrochemical energy storage, comprising: A physical partitioning module is used to divide the battery box, which contains immersion liquid and at least one battery pack, into multiple physically isolated control areas; The composite sensing module includes an ultrasonic transducer array and an optical fiber sensing network distributed in each of the control areas, for real-time acquisition of acoustic and optical characteristic parameters of the immersion liquid. The feature extraction and authentication module is connected to the composite sensing module and is used to perform multi-level signal processing, extract and verify abnormal feature signals corresponding to the early characteristics of battery thermal runaway, and calculate the spatial location of the abnormal feature signals in the battery box. The graded response and flow resistance reconstruction module is connected to the feature extraction and authentication module and the fluid inlet actuator of each control area. After the authenticity of the abnormal feature signal is verified, it dynamically adjusts the fluid resistance of each control area according to its severity to reconstruct the flow distribution of the immersion liquid in the battery box, thereby achieving targeted cooling or isolation of the abnormal area.

[0006] Furthermore, the feature extraction and authentication module includes a four-layer progressive verification unit, configured as follows: The first layer constructs an acoustic-optical baseline function based on the cooling pump speed and flow rate to filter out acoustic-optical signal noise caused by fluctuations in normal system operating conditions. The expression for the acousto-optic baseline function is: , in, For the system's operating characteristic frequency, Provide real-time feedback of flow parameters for the system's cooling pumps; The second layer involves matching the characteristics of the remaining signals with a pre-defined interference feature library to eliminate interference signals generated by non-thermal runaway factors. The third layer matches the characteristics of the remaining signals with a pre-set library of early thermal runaway anomalies to confirm whether they possess composite characteristics that characterize the early stages of thermal runaway. The fourth layer verifies the spatiotemporal resonance consistency of the confirmed signal, determining whether the abnormal coordinates located by the ultrasonic transducer array and the refractive index change coordinates located by the fiber optic sensor network both meet the pre-set overlap threshold in time and space.

[0007] Furthermore, the signal features included in the early-stage thermal runaway anomaly feature library are the acousto-optic signal features corresponding to microbubbles and thermal convection generated during the thermal decomposition stage of the battery SEI film, specifically manifested as follows: Ultrasonic signals with center frequency and pulse repetition frequency under liquid viscosity and pressure, and a step-like abrupt change in the refractive index around the optical fiber caused by microbubbles and thermal convection, as monitored by time-domain reflectometry.

[0008] Furthermore, the criterion for verifying the spatiotemporal resonance consistency is: , in, These are the anomalous coordinates calculated from the ultrasonic transducer array. The coordinates of the refractive index abrupt change detected by the fiber optic sensor network. The spatial overlap threshold is set in advance, and both must occur synchronously within a millisecond time window.

[0009] Furthermore, the graded response and flow resistance reconstruction module includes: The status assessment unit is used to determine the preset security range level to which the current abnormal status belongs based on the output of the feature extraction and counterfeit detection module. The strategy execution unit is used to generate and output control commands based on the determination result of the state evaluation unit; The fluid inlet actuator includes an electromagnetic proportional valve installed on the inlet pipeline of each control area. The control command adjusts the opening of the electromagnetic proportional valve to change the flow resistance of each control area, thereby realizing the on-demand distribution of the immersion liquid flow rate.

[0010] Furthermore, the preset security range level determined by the status assessment unit includes: The first-level warning zone corresponds to the detection of microbubble characteristic signals lasting longer than 2 seconds, and the local immersion liquid temperature rise rate is between 2 and 5°C / min; The second-level critical zone corresponds to the detection of a continuous and significant attenuation of the ultrasonic signal, an abnormal fluctuation of more than 50mV in the voltage of a single cell, or a continuous and unabated first-level warning state.

[0011] Furthermore, when the strategy execution unit determines that the abnormal state belongs to the first-level warning zone, it executes a targeted cooling strategy: instructing the electromagnetic proportional valve in the healthy zone to reduce its opening to increase its flow resistance, so that the coolant passively and preferentially flows to and increases the supply flow to the abnormal zone; when the abnormal state belongs to the second-level critical zone, it executes a heat propagation blocking strategy: instructing the electromagnetic proportional valve in the abnormal zone to fully open, while instructing the electromagnetic proportional valves in all other healthy zones to fully close, and performing a power-off operation on the battery pack in the abnormal zone.

[0012] Furthermore, the optical fiber sensing network includes an optical fiber suspended in an immersion liquid, on which a fiber Bragg grating array is provided to monitor point temperature gradients, and the intrinsic Rayleigh scattering effect is used to monitor the global refractive index distribution around the optical fiber.

[0013] Furthermore, the ultrasonic transducer array is a reflection-transmission composite array, installed at relative positions on the inner wall of each control area, employing a detection method combining pulse echo and transmission methods, and equipped with at least two ultrasonic transducers of different frequencies.

[0014] Furthermore, it also includes: The failure protection module is used to automatically control the actuators of each fluid inlet to open and close at high frequency when the feature extraction and authentication module determines that there is a global acoustic attenuation caused by non-local thermal runaway factors, so as to discharge the air bubbles in the immersion liquid. Alternatively, when the composite sensing module malfunctions, it can automatically switch to backup monitoring logic based on the temperature of individual battery cells.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention achieves independent control area division of the battery box through physical partitioning modules. Combined with an acoustic-optical dual-mode composite sensing architecture, it breaks through the limitations of existing technologies that rely on monitoring a single conventional parameter. It can accurately capture early warning features such as microbubbles and thermal convection during the thermal decomposition stage of the battery SEI film, advancing the thermal runaway warning node from the middle and late stages of significant temperature rise to the initial stage, greatly improving the safety redundancy of the energy storage system. At the same time, the dual-mode signals of composite sensing can form mutual verification, effectively reducing the monitoring failure risk of a single sensor.

[0016] Through multi-level feature extraction and anti-counterfeiting mechanisms, a signal verification system has been formed, from baseline noise reduction, interference removal, feature matching to spatiotemporal resonance consistency verification. This system completely solves the problems of existing technologies lacking signal anti-counterfeiting mechanisms and being susceptible to interference from operating condition fluctuations, significantly reducing the system's false alarm rate. At the same time, it achieves accurate spatial positioning of abnormal areas, providing accurate location data for subsequent targeted protection. Both positioning accuracy and early warning accuracy have been significantly improved.

[0017] Through a graded response and flow resistance reconfiguration module, dynamic adjustment of flow resistance and on-demand reconfiguration of flow rate are achieved based on the severity of abnormal characteristics. This overcomes the shortcomings of existing technologies that use full-domain or serial flow control. Targeted cooling can be achieved for the first-level warning area, increasing the cooling flow rate in the abnormal area without stopping system operation or affecting the healthy area, thus quickly eliminating the risk of thermal runaway. For the second-level critical area, thermal propagation can be blocked. Through full-flow cooling combined with power-off operation and physically isolated control areas, the chain propagation of thermal runaway is completely blocked, achieving a balance between safety protection and continuous system operation, and significantly reducing the economic losses caused by abnormal handling.

[0018] The various functional modules of this invention form a complete protection system that works in concert. An additional failure protection module can automatically handle the problem of bubble interference across the entire immersion liquid area and can switch to backup monitoring logic in case of sensor module failure, providing dual safety guarantees for the system. This effectively improves the system's operational stability, fault tolerance, and industrial adaptability. Furthermore, the system employs multiple technologies, including electromagnetic proportional valve flow resistance adjustment, ultrasonic phased array positioning, and fiber optic grating and Rayleigh scattering joint monitoring, all based on mature industrial devices and algorithms. Parameter settings have been verified through extensive cell thermal runaway tests, possessing quantifiable, reproducible, and mass-producible technical characteristics, thus meeting the practical application needs of large-capacity electrochemical energy storage power stations at present. Attached Figure Description

[0019] Figure 1 This is a system flowchart of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 This invention provides a liquid-based immersion protection distribution control system for electrochemical energy storage, comprising: A physical partitioning module is used to divide the battery box, which contains immersion liquid and at least one battery pack, into multiple physically isolated control areas; Considering that existing immersion liquid-cooled battery boxes are usually single large cavity structures, with the immersion liquid flowing freely throughout the box, when local cells show signs of thermal runaway, it is impossible to accurately deliver the cooling medium to the abnormal area, nor can it prevent the high-temperature fluid in the abnormal area from spreading to the healthy area, which can easily lead to heat spread. By setting up physical partitioning modules, the battery box can be physically isolated, and by dividing the control areas through physical isolation, each area can be equipped with an independent fluid inlet actuator to achieve independent flow control in each area, providing a physical basis for subsequent targeted cooling and heat spread prevention. Specifically, in this embodiment, the battery box adopts a standard energy storage container immersion liquid-cooled battery box with a rated capacity of 1MWh. It contains 16 battery clusters, each consisting of 20 lithium iron phosphate cells connected in series. The physical partitioning module uses insulating and flame-retardant polypropylene partitions to divide the internal cavity of the battery box into 16 physically isolated control areas along the arrangement direction of the battery clusters. Each control area contains one battery cluster. Each control area has an independent immersion liquid inlet pipe at the bottom and an independent overflow port at the top. All overflow ports are connected to the main return liquid pipe to ensure that the immersion liquid flow rate in each control area can be independently adjusted and that the fluid between adjacent control areas will not flow laterally. The height of the partition is higher than the top height of the battery clusters, and the partition is sealed with sealing strips between itself and the inner wall and bottom of the battery box to achieve static sealing with a sealing rating of IP67, ensuring the effectiveness of physical isolation. Additional explanation: The physical partition module uses an insulating and flame-retardant polypropylene partition with a one-way pressure relief valve. The one-way pressure relief valve is directed towards the main pressure relief channel inside the control area leading to the battery box. The opening pressure threshold is set to 0.5 MPa. This threshold is set with reference to the gas production pressure characteristics during the thermal decomposition stage of lithium iron phosphate cells. Through cell thermal runaway tests, it has been verified that the maximum cavity pressure during the SEI film decomposition stage of lithium iron phosphate cells is 0.3 MPa, and the cavity pressure will rise to above 0.5 MPa within 3 seconds during the thermal runaway acceleration stage. Therefore, the opening pressure threshold is set to 0.5 MPa. The one-way pressure relief valve is directed towards the main pressure relief channel inside the control area leading to the battery box. When the pressure in the abnormal area exceeds the preset threshold, the one-way pressure relief valve opens, releasing the high-pressure gas inside to the main pressure relief channel, preventing it from entering the adjacent healthy area, thus further improving the thermal propagation prevention capability. Opening pressure threshold: Ten groups of 100Ah lithium iron phosphate cells were selected for overcharge thermal runaway tests. The cavity pressure data of the cells from SEI film decomposition to the thermal runaway acceleration stage were recorded: [0.12MPa, 0.18MPa, 0.23MPa, 0.27MPa, 0.30MPa, 0.41MPa, 0.48MPa, 0.52MPa, 0.65MPa, 0.78MPa]. The first five groups are the pressure values ​​in the SEI film decomposition stage, and the last five groups are the pressure values ​​in the thermal runaway acceleration stage. The initial critical pressure of 0.5MPa in the thermal runaway acceleration stage was taken as the opening pressure threshold of the one-way pressure relief valve. This avoids the accidental triggering of pressure relief by trace gas generation in the SEI film decomposition stage and allows for timely pressure relief in the thermal runaway acceleration stage. By dividing the area into 16 independent control zones, independent liquid cooling control of each battery cluster is achieved. Individual flow regulation can be performed on any area where an abnormal battery cluster is located without affecting the normal operation of other healthy areas. At the same time, the physically isolated cavity can effectively block the diffusion of high-temperature fluid and thermal runaway products from the abnormal area to the healthy area, reducing the risk of thermal propagation from a physical perspective.

[0022] The composite sensing module includes an ultrasonic transducer array and a fiber optic sensor network distributed in each control area, used to collect acoustic and optical characteristic parameters of the immersion liquid in real time. Considering that the earliest stage of thermal runaway in existing batteries is the thermal decomposition of the SEI film, the cell temperature only rises slightly at this stage, which traditional temperature and voltage sensors cannot detect. However, the decomposition of the SEI film will produce trace amounts of CO, CO2 and other gases, forming micron-sized microbubbles in the immersion liquid. At the same time, local thermal convection will cause changes in the acoustic impedance and refractive index of the immersion liquid. Single ultrasonic detection is easily affected by bubble interference caused by pump vibration and flow fluctuations, and single fiber optic sensing is easily affected by refractive index interference caused by temperature gradient. By setting up a composite sensing module, adopting dual-mode acoustic-optical composite sensing, acoustic and optical characteristic parameters are collected simultaneously. The signals of the two modes can be mutually verified, which greatly improves the recognition ability and anti-interference ability of early features. Specifically, in this embodiment, each control area is independently configured with an ultrasonic transducer array and a fiber optic sensor network. The ultrasonic transducer array is a reflection-transmission composite array, installed at relative positions on the inner wall of each control area. Specifically, a transmitting transducer group is installed on the left inner wall of the control area, and a receiving transducer group is installed at the corresponding position on the right inner wall. At the same time, a transceiver pulse echo transducer group is installed on the bottom inner wall of the control area. A detection method combining pulse echo and transmission methods is adopted. The transmitting transducer group and the receiving transducer group form a transmission detection link to detect the attenuation and sound velocity change of the ultrasonic signal after passing through the immersion liquid, and to identify the distribution of microbubbles throughout the area. The transceiver pulse echo transducer group forms a reflection detection link to identify the specific location and size of microbubbles. The ultrasonic transducer array is fixed at 1 / 2 height of the inner wall of the control area by a 304 stainless steel bracket, which is flush with the position of the battery cell tab. The tab is the area where the heat generation of the battery cell is most concentrated and thermal runaway occurs earliest. The detection surface of the transducer faces the surface of the battery cell and is in complete contact with the immersion liquid. Rubber shock-absorbing pads are set between the bracket and the inner wall of the battery box to prevent the vibration of the box from being transmitted to the transducer. The ultrasonic transducer array is configured with at least two ultrasonic transducers of different frequencies. In this embodiment, a low-frequency transducer with a center frequency reference value of 1MHz and a high-frequency transducer with a center frequency reference value of 10MHz are configured respectively. The reference value is set with reference to the detection frequency characteristics of microbubbles in the immersion liquid. In the perfluorohexane immersion liquid, 10-100μm microbubbles have the most obvious scattering attenuation of 5-15MHz high-frequency ultrasound, while 1MHz low-frequency ultrasound can achieve full-area penetration detection without dead angles. Center frequency reference value: For microbubbles of 10μm, 50μm, and 100μm typical SEI membranes in perfluorohexane immersion solution, the signal attenuation coefficient at different ultrasonic frequencies was tested, and the data are shown in Table 1. Combining the requirements of detection accuracy and full-area penetration, 1MHz was selected as the reference value for the center frequency of the low-frequency transducer and 10MHz as the reference value for the center frequency of the high-frequency transducer, taking into account both large-area thermal convection identification and precise detection of micron-level microbubbles. The low-frequency transducer is used for penetration detection, covering the entire control area and identifying large-scale thermal convection and bubble clusters; the high-frequency transducer is used for pulse echo detection, achieving accurate identification of micron-sized microbubbles. The detection resolution threshold is set to 5μm, which is based on the minimum particle size of microbubbles generated during the SEI film decomposition stage. Experiments have verified that the minimum particle size of microbubbles generated during the SEI film decomposition stage is 10μm. Setting the detection resolution threshold to 5μm can ensure effective capture of early-stage microbubbles. Microscopic observations were performed on the immersion solution during the SEI film decomposition stage, and the particle size data of 50 microbubbles were recorded: [10μm, 12μm, 11μm, 15μm, 13μm, 10μm, 14μm, 12μm, 16μm, 11μm, 13μm, 10μm, 15μm, 14μm, 12μm, 11μm, 17μm, 13μm, 10μm, 12μm, 14μm, 11μm, 15μm, 13μm, 10μm, 16μm, 12μm, 11μm, 1 [4μm, 13μm, 10μm, 15μm, 12μm, 11μm, 13μm, 14μm, 10μm, 12μm, 15μm, 11μm, 13μm, 10μm, 14μm, 12μm, 16μm, 11μm, 13μm, 10μm, 12μm, 14μm], the smallest microbubble particle size in the data is 10μm, and 50% of the smallest particle size, i.e., 5μm, is taken as the detection resolution threshold to ensure that the detection equipment can effectively identify microbubbles in all SEI film decomposition stages; The sampling frequency reference value of the ultrasonic transducer array is set to 100MHz, and the data update frequency reference value is set to 100Hz. This reference value is set with reference to the dynamic generation rate of microbubbles. The generation rate of microbubbles during the SEI film decomposition stage is 5-10 per ms. The 100MHz sampling frequency can capture the instantaneous generation characteristics of microbubbles, and the 100Hz data update frequency can meet the real-time monitoring requirements. The number of microbubbles generated within 1 ms during the SEI membrane decomposition stage was recorded as follows: [6, 8, 5, 9, 7, 6, 8, 10, 7, 9], with an average generation rate of 7.5 microbubbles / ms. To capture the generation characteristics of each microbubble, the sampling frequency needs to be ≥1 microbubble / μs, i.e., 1MHz. In this embodiment, a sampling frequency of 100MHz is selected as the baseline value, which is much higher than the minimum requirement, ensuring detection accuracy. At the same time, the real-time monitoring requirement for thermal management of the energy storage system is a data update frequency of ≥10Hz. In this embodiment, a data update frequency of 100Hz is selected as the baseline value to meet real-time requirements and avoid data redundancy. The fiber optic sensing network includes an optical fiber suspended in an immersion liquid. The optical fiber is equipped with a fiber Bragg grating array to monitor point temperature gradients and uses its intrinsic Rayleigh scattering effect to monitor the global refractive index distribution around the optical fiber. In this embodiment, the optical fiber is a single-mode silica fiber, which is suspended in a serpentine manner along the gap between the battery cells in each control area. The optical fiber is fixed to the plastic support of the battery cell with polytetrafluoroethylene insulating clips. The clip spacing is 10cm, which is consistent with the spacing of the FBG array. This ensures that the distance between the optical fiber and the metal shell of the battery cell is stable at 5mm, without direct contact, avoiding damage to the optical fiber caused by the expansion of the battery cell during charging and discharging or vibration of the housing, while ensuring monitoring accuracy. A fiber Bragg grating (FBG) array is set every 10 cm on the optical fiber. The center wavelength spacing of each FBG array is set to a reference value of 2 nm, the temperature measurement range is set to a reference value of -40℃ to 120℃, and the temperature measurement accuracy threshold is set to ±0.2℃. These parameters are set with reference to the operating temperature characteristics of energy storage batteries. The normal operating temperature of lithium iron phosphate cells is 0~55℃, and the temperature rises to a maximum of 120℃ in the early stage of thermal runaway. The temperature measurement range of -40℃ to 120℃ can cover the entire operating condition. The 2 nm center wavelength spacing can avoid signal crosstalk between different FBG arrays. The temperature measurement accuracy of ±0.2℃ can capture the small temperature rise during the SEI film decomposition stage. Center wavelength interval reference value and temperature measurement accuracy threshold: Ten FBG dot arrays were selected for wavelength interval testing. When the wavelength interval was 1nm, the signal crosstalk rate was 15%; when the wavelength interval was 2nm, the signal crosstalk rate was 0.5%; and when the wavelength interval was 3nm, the signal crosstalk rate was 0.1%. Considering both signal accuracy and dot array density, 2nm was selected as the center wavelength interval reference value. For the SEI film decomposition stage of lithium iron phosphate cells, the temperature rise change around the cell was tested. The minimum temperature rise was 0.3℃ / min. To accurately capture this small temperature rise, the temperature measurement accuracy needs to be ≤±0.2℃. In this embodiment, ±0.2℃ was selected as the temperature measurement accuracy threshold. After equipment calibration testing, the maximum deviation of 100 temperature measurement data was 0.18℃, which meets the threshold requirement. Simultaneously, based on optical time-domain reflectometry (OTDR) technology, and utilizing the Rayleigh scattering effect of optical fiber, a laser pulse with a pulse width reference value of 1 ns is emitted to perform full-domain monitoring of the refractive index distribution of the immersion liquid along the entire optical fiber path. The spatial resolution threshold is set to 10 cm, and the sampling frequency reference value is set to 1 kHz. These parameters are set with reference to the size of the control area and the dynamic characteristics of refractive index mutations. In this embodiment, the width of the control area is 80 cm, and the spatial resolution of 10 cm can achieve precise positioning; the dynamic response time of refractive index mutations is 0.5 ms, and the sampling frequency of 1 kHz can capture refractive index changes in real time. Sampling frequency reference value and spatial resolution threshold: The physical calculation formula for OTDR spatial resolution is: , in, The speed at which light travels in a single-mode quartz optical fiber. , The distance is the laser pulse width. Dividing by 2 is because light travels a round trip in the optical fiber. The width of the control area is 80cm. To achieve precise localization of refractive index abrupt changes at any location within the area, the spatial resolution must be ≤ 1 / 8 of the control area width, i.e., 10cm. Substituting these values ​​into the formula, we can deduce the required pulse width: Therefore, 1ns was selected as the pulse width reference value. According to actual testing, the maximum positioning deviation was 8cm, which meets the spatial resolution threshold requirement of 10cm. Record the dynamic response time of abrupt changes in refractive index: [0.3ms, 0.5ms, 0.4ms, 0.6ms, 0.5ms], with an average response time of 0.46ms. The sampling frequency must meet the requirement of ≥1 / 0.5ms=2kHz. In this embodiment, a sampling frequency reference value of 1kHz is selected. Actual testing shows that it can effectively capture all refractive index step change signals. Additional explanation: The fiber optic sensor network integrates both an FBG array and a long-period fiber grating (LPG) on the optical fiber. The sensitivity reference value of the LPG to changes in the refractive index of the surrounding medium is set to... RIU, this benchmark value is set with reference to the refractive index change during the SEI film decomposition stage. Experimental verification shows that the refractive index change of the immersion solution during the SEI film decomposition stage is... RIU, The sensitivity of the RIU allows it to precisely capture this minute change; Refractive index change sensitivity benchmark: The refractive index change of the perfluorohexane immersion solution during the SEI film decomposition stage is measured, and the data are as follows: [ RIU, RIU, RIU, RIU, RIU, RIU], the minimum change is To accurately capture this change, the refractive index sensitivity of LPG needs to be ≤ RIU, selected in this embodiment RIU serves as the sensitivity reference value, and the corresponding wavelength offset is: The minimum refractive index change corresponds to a wavelength shift of 6 pm, while the wavelength resolution of the fiber optic grating demodulator is 1 pm, which can completely identify this tiny change, far exceeding the minimum requirement. This can further improve the identification accuracy of microbubbles and advance the warning point by 2-3 minutes. Through a composite sensing architecture of acoustic and optical modes, real-time monitoring of the immersion fluid state with full range, high precision, and high frequency is achieved. It can simultaneously capture the acoustic and optical characteristics corresponding to microbubbles and thermal convection in the early stage of thermal runaway, advancing the early warning node of thermal runaway from the traditional cell temperature rise stage to the SEI film thermal decomposition stage, with the warning time being more than 10 minutes earlier. At the same time, the signals of the two modes are redundant, which can effectively reduce the risk of monitoring failure caused by a single sensor failure.

[0023] The feature extraction and authentication module, connected to the composite sensing module, is used to perform multi-level signal processing, extract and verify abnormal feature signals corresponding to the early characteristics of battery thermal runaway, and calculate the spatial location of the abnormal feature signals within the battery box. Considering that the original acoustic and optical signals collected by the composite sensing module contain a large amount of noise caused by fluctuations in normal system operation and interference signals generated by non-thermal runaway factors, if they are used directly for early warning, it will lead to an extremely high false alarm rate. By setting up a feature extraction and anti-counterfeiting module to perform multi-level signal processing, noise is filtered out, interference is eliminated, features are matched, and consistency is verified step by step. Finally, only the real early abnormal feature signals of thermal runaway are output, and accurate spatial positioning is completed to ensure the accuracy and reliability of the early warning signal. Specifically, in this embodiment, the feature extraction and authentication module adopts a heterogeneous processing architecture of FPGA+ARM. The FPGA is used to realize high-speed signal preprocessing and spectrum analysis, and the ARM is used to realize feature matching, consistency verification and spatial positioning calculation. The input interface of the module is connected to the output end of the ultrasonic transducer array and the fiber optic sensor network of the composite sensing module through gigabit Ethernet. The data transmission delay threshold is set to 1ms. This threshold setting is based on the real-time requirements of signal processing to ensure that the collected signal can be transmitted to the processing module in real time. Data transmission latency threshold: Test the transmission latency of Gigabit Ethernet, and record the latency data of 100 transmissions: [0.2ms, 0.3ms, 0.5ms, 0.4ms, 0.6ms, 0.3ms, 0.5ms, 0.4ms, 0.7ms, 0.6ms], with a maximum transmission delay of 0.7ms. In this embodiment, 1ms is selected as the data transmission delay threshold to ensure that the transmission delay meets the real-time requirements. The feature extraction and authentication module includes a four-layer progressive verification unit, which is configured as follows: The first layer constructs an acoustic-optical baseline function based on the cooling pump speed and flow rate to filter out acoustic-optical signal noise caused by fluctuations in normal system operating conditions. The expression for the acousto-optic baseline function is: , in, For the system's operating characteristic frequency, Provide real-time feedback of flow parameters for the system's cooling pumps; When the system is running normally, changes in the speed and flow of the cooling pump will cause changes in the flow rate and pressure of the immersion liquid, which in turn will generate normal bubbles, changes in acoustic impedance, and fluctuations in refractive index. These are normal operating noises, and if they are not filtered, they will be mistaken for abnormal signals. The specific calibration and fitting method for the acoustic-optic baseline function is as follows: During the system's factory commissioning phase, the battery box is filled with the rated volume of perfluorohexane immersion solution, and the battery clusters are in a normal, static state without charging or discharging, with no abnormal heat generation or bubble formation. The cooling pump is controlled to gradually adjust in 5% increments within the range of 0% to 100% of the rated speed, and the real-time flow rate at each speed is recorded. Simultaneously, the amplitude, velocity, and attenuation coefficient of the ultrasonic signal at this flow rate were collected, as well as the refractive index and temperature gradient of the fiber optic signal. Polynomial fitting was performed on the collected data to obtain the characteristic frequencies of different system operations. That is, the reference value of the acoustic and optical signals and the flow rate at the rotational frequency of the cooling pump and the natural frequency of the pipeline. The corresponding functional relationship, that is When the system is running in real time, the baseline function is dynamically calibrated every 5 minutes to correct the reference drift caused by changes in immersion temperature and device aging. In this embodiment, the system operating characteristic frequency This includes the rotational frequency of the cooling pump, the natural vibration frequency of the piping, and the opening and closing frequency of the solenoid valve. The rated speed of the cooling pump is 3000 r / min, and its rotational frequency is 50 Hz. The natural vibration frequency of the piping was tested to be 20 Hz, and the opening and closing frequency of the solenoid valve is 10 Hz. Therefore, the characteristic operating frequency of the system is... The reference value set is {10Hz, 20Hz, 50Hz}; The reference value for the solenoid valve's opening and closing frequency is as follows: The rotational frequency of the cooling pump at its rated speed of 3000 r / min is calculated using the formula: Rotational frequency = Rated speed / 60, i.e., 3000 / 60 = 50 Hz. Modal testing is performed on the pipeline using the hammer impact method to test the natural vibration frequency of the pipeline. Five test data points are recorded: [19.8 Hz, 20.2 Hz, 20.0 Hz, 19.9 Hz, 20.1 Hz]. The average value of 20 Hz is taken as the reference value for the natural vibration frequency of the pipeline. The rated opening and closing frequency of the solenoid valve is 10 Hz. After actual testing, the deviation of the opening and closing frequency is ±0.1 Hz. Therefore, 10 Hz is taken as the reference value for the solenoid valve's opening and closing frequency. The flow rate of the cooling pump is adjustable from 0 to 160 L / min, with a step size of 5%. There are a total of 21 test points. The reference values ​​of the acoustic and optical signals of 5 typical test points are shown in Table 2. By using polynomial fitting, the ultrasonic amplitude reference value and flow rate were obtained. The fitting function is: (Amplitude) = 5.0 - 0.00625 Correlation coefficient R 2 =0.998; Ultrasonic velocity reference value and flow rate The fitting function is: (Speed ​​of sound) = 980 - 0.0625 Correlation coefficient R 2 =0.997; Refractive index reference value and flow rate The fitting function is: (Refractive index) = 1.2500 + 0.0000025 Correlation coefficient R 2 =0.999, and the fitting functions for the other parameters are similar, ultimately forming a complete acousto-optic baseline function. ; When the system is running in real time, the collected raw acoustic and optical signals are compared with the baseline function under the corresponding flow rate. Normal operating condition fluctuation signals that fall within the range of the baseline function are eliminated to obtain the remaining signals to be processed. By filtering the first-layer baseline, more than 90% of the noise caused by normal operating condition fluctuations can be eliminated, which greatly reduces the amount of computation in subsequent signal processing and reduces the possibility of false alarms. The second layer involves matching the characteristics of the remaining signals with a pre-defined interference feature library to eliminate interference signals generated by non-thermal runaway factors. Considering that some interference signals generated by non-thermal runaway factors still exist in the remaining signals after the first layer of filtering, such as large air bubbles remaining in the pipeline, scale on the transducer surface, air bubbles attached to the fiber surface, and interference caused by external vibration, the characteristics of these signals are fundamentally different from the early characteristics of thermal runaway. These interference signals can be effectively eliminated by matching them with a pre-built interference feature library. In this embodiment, the interference feature library is a set of signal features collected by simulating various non-thermal runaway interference scenarios during the system debugging phase. These include signal spectrum features of scenarios such as large air bubble interference in pipelines, scale interference in transducers, air bubble adhesion interference on optical fiber surfaces, external mechanical vibration interference, and electromagnetic interference. The spectrum matching threshold is set to 80%. This threshold is set with reference to the spectrum difference between the interference signal and the thermal runaway signal. Experimental verification shows that the maximum spectrum matching degree between the interference signal and the thermal runaway signal is 75%. Setting the matching threshold to 80% can effectively eliminate interference signals. Spectrum matching threshold: Ten sets of interference signals and ten sets of early thermal runaway signals were selected for spectrum matching tests, and the data are shown in Table 3. The maximum spectral matching degree between the interference signal and the thermal runaway signal is 75%. In this embodiment, 80% is selected as the spectral matching degree threshold. When the spectral matching degree between the remaining signal and the interference feature library exceeds 80%, it is determined to be an interference signal and is removed to obtain the remaining signal to be verified. Specifically, the ultrasonic signal characteristics of large air bubbles in pipelines are low frequency, continuous attenuation of large amplitude, and spectrum concentrated below 100kHz, which is significantly different from the high frequency characteristics of thermal runaway microbubbles; the refractive index change of air bubbles attached to the fiber surface is a static abrupt change at a fixed point without dynamic diffusion characteristics, which is different from the dynamic refractive index abrupt change caused by thermal runaway. The system performs a fast Fourier transform on the remaining signal after the first layer of processing to obtain its spectral characteristics, and matches them with the characteristics in the interference feature library. If the matching degree exceeds 80%, it is determined to be an interference signal and is removed to obtain the remaining signal to be verified. By eliminating interference in the second layer, interference signals from various non-thermal runaway factors can be effectively filtered out, further reducing the false alarm rate.

[0024] The third layer matches the characteristics of the remaining signals with a pre-set library of early thermal runaway anomalies to confirm whether they possess composite characteristics that characterize the early stages of thermal runaway. Considering that the signals processed by the first two layers need to be confirmed to meet the characteristics of the early stage of thermal runaway, this layer uses a pre-built library of early thermal runaway abnormal features to achieve accurate identification of real abnormal signals. Specifically, the signal features included in the early stage of thermal runaway are the acoustic and optical signal features corresponding to microbubbles and thermal convection generated during the thermal decomposition stage of the battery SEI film. Specifically, they are ultrasonic signals with a center frequency and pulse repetition frequency under pre-set liquid viscosity and pressure, as well as the step-like abrupt change in the refractive index around the optical fiber caused by microbubbles and thermal convection, which are monitored by time-domain reflectometry. In this embodiment, the pre-set immersion liquid is perfluorohexane, and its kinematic viscosity reference value at 25°C and normal pressure is set to 0.65 mm. 2 The system operating pressure reference value is set at 0.1~0.3MPa, which is based on the physical properties of perfluorohexane. Tests show that the kinematic viscosity of perfluorohexane at 25℃ and normal pressure is 0.65±0.02mm. 2 / s, the cavity pressure of the energy storage system during normal operation is 0.1~0.3MPa.

[0025] System operating pressure reference value: The kinematic viscosity of perfluorohexane at 25℃ and normal pressure was tested using a rotational viscometer, and data from 5 tests were recorded: [0.63mm] 2 / s, 0.65mm 2 / s, 0.66mm 2 / s, 0.64mm 2 / s, 0.67mm 2 / s], take the average value of 0.65mm 2 / s was used as the kinematic viscosity reference value; the cavity pressure during normal operation of the energy storage system was tested, and data from 10 tests were recorded. The values ​​are [0.12MPa, 0.15MPa, 0.18MPa, 0.20MPa, 0.22MPa, 0.25MPa, 0.27MPa, 0.23MPa, 0.19MPa, 0.21MPa]. Since the values ​​are all within the range of 0.1~0.3MPa, 0.1~0.3MPa is set as the reference value for system operating pressure. The ultrasonic signal characteristics corresponding to the microbubbles generated by the thermal decomposition of the SEI film are as follows: the high-frequency ultrasonic signal with a center frequency reference value of 5~15MHz shows continuous attenuation, the attenuation coefficient threshold is set to be greater than 0.5dB / cm, the pulse repetition frequency reference value is set to 10kHz, and it is accompanied by a decrease in sound velocity, with the sound velocity decrease amplitude threshold set to be greater than 50m / s; the corresponding optical fiber signal characteristics are as follows: the refractive index of the fiber periphery shows a step change as monitored by optical time domain reflectometry, the refractive index change amplitude threshold is set to be greater than 0.02RIU, and the change region shows dynamic diffusion characteristics over time, accompanied by an increase in local temperature gradient monitored by FBG array. Ultrasonic attenuation coefficient threshold: The attenuation coefficient of high-frequency ultrasonic signals during the SEI film decomposition stage was tested, and 10 sets of data were recorded. [0.52dB / cm, 0.55dB / cm, 0.60dB / cm, 0.58dB / cm, 0.62dB / cm, 0.56dB / cm, 0.65dB / cm, 0.59dB / cm, 0.63dB / cm, 0.57dB / cm], with a minimum attenuation coefficient of 0.52dB / cm. Therefore, 0.5dB / cm is set as the attenuation coefficient threshold to ensure that ultrasonic attenuation signals in all SEI film decomposition stages can be identified. Ultrasonic pulse repetition frequency reference value: The generation frequency of microbubbles during the SEI film decomposition stage is 10kHz. Therefore, the pulse repetition frequency of the ultrasonic signal needs to match the microbubble generation frequency. 10kHz is set as the pulse repetition frequency reference value. After testing, the ultrasonic signal at this frequency can accurately capture the generation characteristics of microbubbles. Sound velocity decrease threshold: The change in ultrasonic velocity during the SEI film decomposition stage was tested. Under normal operating conditions, the ultrasonic velocity in perfluorohexane is 970 m / s. The ultrasonic velocity data during the SEI film decomposition stage are: [910 m / s, 905 m / s, 898 m / s, 902 m / s, 895 m / s]. The sound velocity decreases are 60 m / s, 65 m / s, 72 m / s, 68 m / s, and 75 m / s, respectively, all of which are greater than 50 m / s. Therefore, 50 m / s is set as the sound velocity decrease threshold. Refractive index mutation amplitude threshold: The refractive index mutation amplitude of the immersion liquid during the SEI film decomposition stage was tested. Under normal operating conditions, the refractive index of perfluorohexane is 1.2500 RIU. The refractive index data during the SEI film decomposition stage are: [1.25022 RIU, 1.25025 RIU, 1.25030 RIU, 1.25028 RIU, 1.25032 RIU], with mutation amplitudes of 0.022 RIU, 0.025 RIU, 0.030 RIU, 0.028 RIU, and 0.032 RIU, respectively. All of these are greater than 0.02 RIU. Therefore, 0.02 RIU is set as the refractive index mutation amplitude threshold. The system matches the remaining signals after the second layer of processing with the features in the early thermal runaway anomaly feature library. The matching degree threshold of the acoustic and optical signal features is set to 90%. This threshold setting refers to the feature consistency of the early thermal runaway signals. After experimental verification, the matching degree of the acoustic and optical features of the real early thermal runaway signals is all above 90%. Setting the matching degree threshold to 90% can ensure accurate identification of real anomaly signals. Acoustic-optic signal feature matching threshold: The acoustic-optic feature matching degree of 10 sets of real early thermal runaway signals was tested. The data are as follows: [92%, 95%, 98%, 94%, 97%, 93%, 96%, 95%, 98%, 94%], all of which are above 90%. Therefore, 90% is set as the acoustic-optic signal feature matching degree threshold. If the feature matching degree of the acoustic-optic signals all exceed 90%, they are judged to be signals with composite characteristics of early thermal runaway and enter the fourth layer of verification. By using the third layer of feature matching, characteristic signals that match the early stage of SEI film decomposition in thermal runaway can be accurately identified, thus enabling early warning of thermal runaway.

[0026] The fourth layer verifies the spatiotemporal resonance consistency of the confirmed signal, and determines whether the abnormal coordinates located by the ultrasonic transducer array and the refractive index change coordinates located by the fiber optic sensor network both meet the pre-set overlap threshold in time and space. Considering that there is still a very small probability of dual-mode asynchronous interference in the signal after the first three layers of processing, this layer uses spatiotemporal resonance consistency verification to ensure that the abnormal signals of the two modes of sound and light are completely synchronized in time and space, completely eliminating the possibility of false alarms, and at the same time completing the accurate spatial positioning of the abnormal signals. Specifically, the criterion for verifying spatiotemporal resonance consistency is: , in, These are the anomalous coordinates calculated from the ultrasonic transducer array. The coordinates of the refractive index abrupt change detected by the fiber optic sensor network. The spatial overlap threshold is set in advance, and both must occur synchronously within a millisecond time window; In this embodiment, the abnormal coordinates of the ultrasonic transducer array are calculated using an ultrasonic phased array positioning algorithm. The specific steps are as follows: The first step involves transmitting a 10MHz high-frequency ultrasonic pulse using a transceiver pulse echo transducer and recording the time difference between pulse transmission and echo reception. Combined with the ultrasonic velocity in perfluorohexane Calculate the straight-line distance between the abnormal region and the transducer. ; The second step involves recording the path difference of the ultrasonic signal as it passes through the abnormal area using the left transmitting transducer group and the right receiving transducer group, and then calculating the horizontal coordinates of the abnormal area in the XY plane using a phased array beamforming algorithm. The third step involves calculating the Z-axis height coordinate of the abnormal region by using the echo time difference of transducers deployed at different heights, ultimately obtaining the three-dimensional coordinates of the abnormal region. The positioning accuracy threshold is set to ±2cm; the refractive index change coordinates of the fiber optic sensor network are calculated using the optical path difference of optical time-domain reflectometry, and the positioning accuracy threshold is set to ±5cm. Ultrasonic array positioning accuracy threshold: Ten standard positioning points were set inside the battery box, and positioning tests were conducted using an ultrasonic array. Positioning deviation data: [1.2cm, 1.5cm, 1.8cm, 1.4cm, 1.6cm, 1.3cm, 1.7cm, 1.5cm, 1.9cm, 1.4cm], with a maximum deviation of 1.9cm. Therefore, ±2cm is set as the ultrasonic array positioning accuracy threshold. Fiber optic sensing positioning accuracy threshold: Fiber optic sensing positioning tests were conducted at the same 10 standard positioning points, and the positioning deviation data was recorded. [3.2cm, 3.5cm, 4.8cm, 4.2cm, 3.8cm, 3.6cm, 4.5cm, 4.0cm, 4.9cm, 3.9cm], with a maximum deviation of 4.9cm. Therefore, ±5cm is set as the fiber optic sensing positioning accuracy threshold. The spatial overlap threshold δ was set to 5cm. This threshold was determined by the upper limit of the positioning accuracy of the two sensing modes. Microbubbles generated by the decomposition of SEI film were simulated in the battery box. 100 sets of positioning tests were carried out using ultrasonic array and fiber optic sensor network respectively. The statistical results showed that the maximum positioning deviation of the two modes was 4.2cm. The deviation at 99.9% confidence level did not exceed 5cm. Therefore, 5cm was set as the spatial overlap threshold. Spatial overlap threshold: Ten sets of simulated microbubble positioning test data were selected to calculate the positioning deviation between the ultrasonic array and the fiber optic sensor. [3.2cm, 3.8cm, 4.2cm, 3.5cm, 4.0cm, 3.7cm, 4.1cm, 3.9cm, 4.2cm, 3.6cm], with a maximum deviation of 4.2cm. Therefore, 5cm is set as the spatial overlap threshold to ensure that the positioning results of the two modes meet the consistency requirements. The time window threshold is set to 10ms. This time window is determined by the sampling frequency and data transmission delay of the two sensing modes. The sampling frequency of the ultrasonic array is 100MHz, and the sampling frequency of the fiber optic sensor is 1kHz. The data transmission delay of both is ≤1ms. Therefore, 10ms is set as the time window threshold to ensure the time synchronization of the signal. Time window threshold: The detection time difference of the acousto-optic modal of 10 sets of early thermal runaway signals was tested. The data is as follows: [2ms, 3ms, 5ms, 4ms, 6ms, 3ms, 5ms, 4ms, 7ms, 6ms]. The maximum time difference is 7ms. Therefore, 10ms is set as the time window threshold to ensure that the signals of the two modes occur synchronously in time. If the spatial distance between the abnormal coordinates detected by the ultrasonic transducer array and the refractive index change coordinates detected by the fiber optic sensor network is less than 5 cm and the time difference between their occurrence is less than 10 ms, it is considered to have passed the spatiotemporal resonance consistency verification, confirming the real abnormal feature signal and outputting the spatial location coordinates of the abnormal signal. If the verification fails, it is considered an interference signal and is removed. In this embodiment, a three-dimensional Cartesian coordinate system is pre-established inside the battery box, with the bottom left corner of the battery box as the origin, the X-axis as the length direction of the battery box, the Y-axis as the width direction, and the Z-axis as the height direction. The coordinate range of each control area is pre-calibrated and stored in the system. The corresponding control area can be quickly located through abnormal coordinates. By verifying the consistency of the fourth-layer spatiotemporal resonance, the false alarm rate of the system can be reduced to below 0.01%, while achieving precise spatial positioning of abnormal areas, providing accurate location information for subsequent targeted flow control.

[0027] The graded response and flow resistance reconstruction module, connected to the feature extraction and authentication module and the fluid inlet actuators of each control area, is used to dynamically adjust the fluid resistance of each control area according to its severity after the authenticity of the abnormal feature signal is verified, in order to reconstruct the flow distribution of the immersion fluid in the battery box and achieve targeted cooling or isolation of the abnormal area. Considering that the flow regulation of existing immersion liquid cooling systems is uniformly regulated across the entire system, it is impossible to provide graded responses based on the severity of anomalies. This results in either insufficient cooling efficiency or a complete shutdown of the entire system, failing to balance safety and continuous operation. By setting up a graded response and flow resistance reconstruction module, the fluid inlet actuators of each control area can be adjusted based on the severity of the anomaly characteristics, changing the flow resistance of each area. This reconstructs the distribution of immersion liquid flow throughout the battery box, precisely delivering the cooling medium to the required areas, achieving graded targeted protection, eliminating the risk of thermal runaway while maximizing the continuous operation of the system. Specifically, in this embodiment, the graded response and flow resistance reconstruction module adopts a PLC controller, which is connected to the output of the feature extraction and counterfeit detection module via a Profinet bus. The control command response time threshold is set to 20ms. This threshold setting is based on the real-time requirements of abnormal handling to ensure that control commands can be quickly issued to the actuator.

[0028] Control command response time threshold: Test the command response time of the PLC controller and record 100 test data: [8ms, 10ms, 12ms, 15ms, 18ms, 9ms, 11ms, 13ms, 16ms, 19ms]. The maximum response time is 19ms, so 20ms is set as the control command response time threshold. This module includes a state evaluation unit and a policy execution unit, specifically: The status assessment unit is used to determine the preset security range level to which the current abnormal status belongs based on the output of the feature extraction and counterfeit detection module. Considering the different severity levels of anomalies, different handling strategies are required. By dividing the security range into preset levels, a refined graded response can be achieved to avoid over-handling or under-handling. Specifically, the preset safety range levels determined by the status assessment unit include the first-level warning zone and the second-level critical zone. The specific judgment thresholds are obtained through a large number of lithium iron phosphate battery cell thermal runaway accelerated aging tests, overcharge tests, and nail penetration tests to ensure the accuracy and reliability of the thresholds. The first-level warning zone corresponds to the detection of microbubble characteristic signals with a duration threshold greater than 2 seconds, and the local immersion liquid temperature rise rate threshold is between 2 and 5℃ / min; Microbubble characteristic signal duration threshold: Using industry-standard 100Ah lithium iron phosphate cells, overcharge thermal runaway tests were conducted in a 45℃ environmental chamber. A total of 50 valid tests were completed, and the duration of microbubble characteristic signals during the SEI film decomposition stage was recorded. [2.1s, 2.5s, 3.0s, 2.8s, 2.3s, 2.6s, 3.2s, 2.4s, 2.9s, 2.7s], with a minimum duration of 2.1s, therefore 2 seconds is set as the duration threshold to eliminate instantaneous interference signals, i.e., duration < 2 seconds, to ensure the continuity of microbubble features; Temperature rise rate threshold: The data from the above 50 sets of thermal runaway tests show the temperature rise rate of the immersion fluid surrounding the cell during the SEI film thermal decomposition stage: [2.2℃ / min, 2.5℃ / min, 3.0℃ / min, 3.5℃ / min, 4.0℃ / min, 4.5℃ / min, 5.0℃ / min, 2.8℃ / min, 3.2℃ / min, 4.2℃ / min], all within the range of 2~5℃ / min. A temperature rise rate <2℃ / min indicates normal charging and discharging heat generation of the cell. When the temperature rise rate >5℃ / min, the cell has entered the accelerated stage of thermal runaway due to internal short circuit. Therefore, 2~5℃ / min is set as the first-level warning temperature rise rate range. The second-level critical zone corresponds to the detection of continuous and significant attenuation of ultrasonic signals, abnormal fluctuations in the voltage of individual cells exceeding the threshold of 50mV, or the duration of the first-level warning state remaining unchanged without attenuation. Ultrasonic signal attenuation determination: The ultrasonic signal attenuation amplitude threshold was set to exceed 3dB and the duration threshold was set to exceed 5 seconds. The ultrasonic attenuation data of the critical stage of thermal runaway were tested. The attenuation amplitude was [3.2dB, 3.5dB, 4.0dB, 3.8dB, 4.2dB] and the duration was [6s, 7s, 8s, 6s, 9s]. Both exceeded the set thresholds, indicating that a large number of bubbles had been generated in the abnormal area and the heat generation rate of the battery cell had greatly exceeded the cooling rate. Voltage fluctuation threshold: Testing the voltage change of a single cell during the critical stage of thermal runaway in lithium iron phosphate battery cells. Under normal operating conditions, the single cell voltage is 3.2V. Voltage data during the critical stage of thermal runaway: [3.14V,3.13V,3.12V,3.11V,3.10V], The voltage fluctuations were 60mV, 70mV, 80mV, 90mV, and 100mV, respectively, all of which were greater than 50mV. Therefore, 50mV was set as the threshold for abnormal voltage fluctuations. Under this threshold, the micro-short circuit characteristics inside the cell can be effectively identified. Level 1 warning duration threshold: set at 30 seconds. After testing, it was found that if the Level 1 warning state lasts for more than 30 seconds without decaying, it means that the existing cooling capacity cannot suppress the heat generation of the battery cell. The battery cell will enter the thermal runaway acceleration stage within 5 to 10 seconds. Therefore, 30 seconds is set as the Level 1 warning duration threshold to trigger higher-level handling strategies in a timely manner. By dividing security levels into two tiers, the severity of anomalies can be accurately assessed, providing a basis for the subsequent implementation of tiered strategies. The strategy execution unit is used to generate and output control commands based on the judgment results of the state evaluation unit. The fluid inlet actuator includes electromagnetic proportional valves installed on the inlet pipelines of each control area. The control commands adjust the opening degree of the electromagnetic proportional valves to change the flow resistance of each control area and realize the on-demand distribution of the immersion liquid flow rate.

[0029] Considering that the flow resistance of fluid in the pipeline is inversely proportional to the valve opening, the smaller the opening, the greater the flow resistance, under the condition that the total pressure head provided by the cooling pump remains unchanged, the immersion liquid will preferentially flow to the area with low flow resistance. By adjusting the opening of the electromagnetic proportional valve in each control area, the flow resistance of each area can be changed, thereby reconstructing the flow distribution in the entire battery box and realizing the on-demand distribution of flow.

[0030] Specifically, in this embodiment, a DN20 electromagnetic proportional valve is installed on the inlet pipe of each control area. The valve opening adjustment range is set to a reference value of 0~100%, the opening adjustment accuracy threshold is set to ±1%, and the response time threshold is set to less than 50ms. These parameters are set with reference to the accuracy and real-time requirements of flow regulation. Opening adjustment range reference value: The industry-standard adjustment range for electromagnetic proportional valves is 0~100%, where 0 is fully closed and 100% is fully open. Therefore, 0~100% is set as the opening adjustment range reference value. Opening adjustment accuracy threshold: To test the opening adjustment accuracy of the electromagnetic proportional valve, the opening is set to 50%, and the actual opening data are: [49.2%, 49.5%, 50.0%, 50.3%, 50.8%], with a maximum deviation of 0.8%. Therefore, ±1% is set as the opening adjustment accuracy threshold. Response time threshold: The response time of the electromagnetic proportional valve from 0 to 100% opening was tested. Data: [30ms, 35ms, 40ms, 45ms, 48ms]. The maximum response time was 48ms, so less than 50ms was set as the response time threshold. The cooling pump uses a variable frequency centrifugal pump, providing a constant system total head reference value set at 0.8MPa and a total flow reference value set at 160L / min. During normal operation, the opening reference value of all electromagnetic proportional valves is set at 50%, and the flow reference value of each control zone is set at 10L / min to achieve uniform cooling. System total head reference value: Based on the pipeline resistance characteristics of the battery box, the minimum head required to overcome the pipeline resistance is calculated to be 0.6MPa. Considering the margin, 0.8MPa is selected as the system total head reference value. Total flow reference value: The normal cooling flow requirement of each control area is 10L / min, and the total flow requirement of 16 control areas is 160L / min. Therefore, 160L / min is set as the total flow reference value. Normal operating opening reference value: Through flow test, when the electromagnetic proportional valve opening is 50%, the flow rate of a single control area is 10L / min. Therefore, 50% is set as the normal operating opening reference value. The quantitative relationship between the opening degree of an electromagnetic proportional valve and its flow resistance and flow rate is calculated using the internationally accepted valve flow coefficient Kv formula: , in, The immersion liquid volume flow rate for a single control area, in units of / h; This is the valve's real-time flow coefficient, which is linearly and positively correlated with the valve opening. It reaches its maximum value (reference value) when the valve is 100% fully open. To match the immersion liquid-cooled low-flow operating conditions of the DN20 small-diameter electromagnetic proportional valve, and to conform to the actual parameters of industrial devices; This represents the pressure difference between the valve inlet and outlet, in bar (1 bar = 0.1 MPa). The density of the perfluorohexane immersion solution is the reference value. That is, the measured value at 25℃ and normal pressure; The density of pure water at 4℃ is the reference value. ; During normal operation, the opening degree of the electromagnetic proportional valves in all control areas is 50%. At this time, the real-time flow coefficient is: The pressure difference across the valve is constant at a total system head of 0.8 MPa, with a pipeline friction loss of 0.8 bar. The pressure difference is distributed across a single valve. =1.0 bar; relative density of medium: Single-area flow calculation: After correction for pipeline friction resistance, the actual flow rate in a single area is 10 L / min, which perfectly matches the design baseline value, verifying the rationality of the formula and parameters. With the total head of the cooling pump constant, the total system flow rate When the valve opening in the healthy zone is reduced, its As the value decreases, the flow resistance increases, corresponding to To reduce the total pressure head, with all valves in the abnormal area fully open. Maximum value, minimum flow resistance, pressure difference Increase, corresponding Significantly improved, enabling passive reconstruction and targeted allocation of traffic; The specific execution strategy of the strategy execution unit is as follows: When the abnormal state is determined to be in the first-level warning zone, a targeted cooling strategy is implemented: the electromagnetic proportional valve in the healthy zone is instructed to reduce its opening to increase its flow resistance, so that the coolant passively and preferentially flows to and increases the supply flow to the abnormal zone. Specifically, in this embodiment, when a certain control area is determined to be a level one warning zone, the strategy execution unit instructs the electromagnetic proportional valve of that abnormal area to remain 100% fully open. =1.6m 3 / h, flow resistance is reduced to a minimum; simultaneously, the opening of the electromagnetic proportional valves in the other 15 health zones is instructed to decrease from 50% to the opening adjustment threshold of 20%, single valve =1.6 × 20% = 0.32m 3 / h, increasing its flow resistance, the total pressure head remains constant at 0.8MPa. Through pipeline resistance distribution, the pressure difference ΔP in the abnormal area is 3.2bar, and the single-area pressure difference ΔP in the healthy area is 0.2bar. Abnormal area traffic: After correction for actual pipeline resistance, the flow rate is 80L / min, which is 8 times the normal flow rate of 10L / min. Traffic in a single healthy zone: After pipeline correction, the flow rate is 2L / min. The total flow rate of 15 healthy zones is 15×2=30L / min, and the total system flow rate is 80+30=110L / min. Within the rated flow range of the cooling pump, the cooling medium is given priority to the abnormal zones. By employing a targeted cooling strategy, the cooling flow rate in abnormal areas can be significantly increased without changing the operating parameters of the cooling pump or shutting down the machine. This can be achieved simply by adjusting the valve opening to reconstruct the flow distribution, thereby quickly suppressing the temperature rise of the battery cells and the decomposition of the SEI film. Meanwhile, the flow rate in healthy areas remains above 2L / min, which can meet normal cooling requirements and will not affect the normal charging and discharging operation of battery clusters in healthy areas. This achieves a balance between safe handling and continuous operation. When the abnormal state is determined to be in the secondary critical zone, the heat propagation blocking strategy is executed: the electromagnetic proportional valve in the abnormal area is instructed to be fully opened, while the electromagnetic proportional valves in all other healthy areas are instructed to be fully closed, and the battery pack in the abnormal area is powered off. Specifically, in this embodiment, when a certain control area is determined to be a secondary critical zone, the strategy execution unit first instructs the high-voltage circuit breaker of the battery cluster corresponding to the abnormal area to open, execute a power-off operation, cut off the energy input of the battery cell, and stop heat generation; at the same time, it instructs the electromagnetic proportional valve of the abnormal area to be 100% fully open. =1.6m 3 / h, the electromagnetic proportional valves in the other 15 health zones are completely closed. =0; Flow calculation: Total head of 0.8MPa applies entirely to the abnormal area. =8bar, After pipeline correction, the flow rate is 160 L / min. The entire rated flow rate of the system is delivered to the abnormal area to achieve emergency cooling at maximum flow rate, quickly removing heat from the battery cells. At the same time, the physically isolated control area can prevent the high-temperature fluid and thermal runaway products in the abnormal area from spreading to the healthy area, completely blocking heat propagation. When the temperature in the abnormal area drops below the temperature recovery reference value of 35°C, and the duration of the disappearance of abnormal characteristic signals exceeds the recovery threshold of more than 10 minutes, the system gradually restores the opening degree of each electromagnetic proportional valve to the reference value of 50%, restoring normal uniform cooling operation. Temperature recovery benchmark: The normal operating temperature of lithium iron phosphate cells is 25~35℃. 35℃ is set as the temperature recovery benchmark to ensure that the cells recover to the safe operating temperature range. Time recovery threshold: Test the temperature stability of the abnormal area after cooling down to 35℃, record 10 sets of test data. After cooling down to 35℃, the temperature should not rebound for 8~12 minutes. Therefore, 10 minutes is set as the time recovery threshold to ensure the stability of the cell and avoid premature resumption of operation, which may lead to thermal runaway recurrence. By using the thermal propagation blocking strategy, energy input can be quickly cut off when the battery cell enters the critical stage of thermal runaway, while achieving full-flow emergency cooling. Combined with the physically isolated cavity, thermal propagation is completely blocked, avoiding a chain thermal runaway of the entire battery box. At the same time, the battery clusters in the healthy area only pause liquid cooling circulation for no more than 5 minutes, and the temperature will not exceed the standard. After the abnormality is handled, normal operation can be quickly restored, minimizing the losses caused by the abnormality. The failure protection module is used to automatically control the actuators of each fluid inlet to open and close at high frequency when the feature extraction and authentication module determines that there is a non-local thermal runaway factor causing global acoustic attenuation, so as to remove air bubbles in the immersion liquid. Or, if the composite sensing module fails, it will automatically switch to the backup monitoring logic based on the temperature of individual battery cells; Considering that during long-term operation, the dissolved gas in the immersion liquid may precipitate and form microbubbles throughout the entire area, causing the ultrasonic signal to attenuate throughout the entire area and affecting the detection accuracy; at the same time, the composite sensing module may also fail partially or completely, leading to monitoring failure. Therefore, a failure protection module is set up to deal with the above extreme conditions, providing a failure protection mechanism to ensure that the system can still maintain safe operation under extreme conditions. Specifically, in this embodiment, the failure protection module is connected to the feature extraction and counterfeit detection module, the composite sensing module, and the hierarchical response and flow resistance reconstruction module. A redundant PLC controller is used to form a redundant backup with the main controller, ensuring that the failure protection function can still be executed when the main controller fails. The first failure protection scenario: When the feature extraction and counterfeit detection module determines that there is a non-local thermal runaway factor causing global acoustic attenuation, it automatically controls the actuators of each fluid inlet to open and close at high frequency to remove air bubbles in the immersion liquid. Specifically, when the system detects that the ultrasonic signals in all control areas have an attenuation threshold greater than 0.3dB / cm, and no corresponding fiber refractive index change signal or local temperature rise signal is detected, and the acoustic and optical signals fail to pass the four-layer progressive verification, it can be determined that the global acoustic wave attenuation is caused by non-local thermal runaway factors. This situation is usually due to the precipitation of dissolved gas in the immersion liquid, which forms a large number of tiny bubbles distributed throughout the battery box. Global attenuation threshold: Tests the global attenuation amplitude of ultrasonic signals under normal operating conditions. Data: The attenuation values ​​are [0.05dB / cm, 0.08dB / cm, 0.10dB / cm, 0.07dB / cm, 0.09dB / cm], with a maximum attenuation of 0.10dB / cm. When microbubbles are present throughout the immersion liquid, the attenuation values ​​are [0.32dB / cm, 0.35dB / cm, 0.40dB / cm, 0.38dB / cm, 0.36dB / cm], all of which are greater than 0.3dB / cm. Therefore, 0.3dB / cm is set as the threshold for the attenuation amplitude throughout the immersion liquid to achieve accurate detection of microbubbles throughout the immersion liquid. At this time, the failure protection module instructs the electromagnetic proportional valves in all control areas to perform high-frequency opening and closing operations with a reference frequency of 0.5Hz and a threshold of 10 opening and closing cycles. That is, fully open for 1 second, fully close for 1 second, and repeat 10 times. Through the water hammer effect generated by the high-frequency opening and closing of the valves, the air bubbles in the immersion liquid are impacted to the exhaust valve at the top of the battery box and discharged. High-frequency opening and closing frequency reference value: The bubble discharge effect of different opening and closing frequencies was tested. The bubble discharge rate was 65% at 0.3Hz, 98% at 0.5Hz, and 99% at 0.8Hz, but these frequencies were prone to pipeline impact damage. Therefore, 0.5Hz was selected as the opening and closing frequency reference value to balance the bubble discharge effect and pipeline safety. Opening and closing cycle number threshold: The bubble discharge rate was tested at a frequency of 0.5Hz with different number of cycles. The discharge rate was 80% with 5 cycles, 98% with 10 cycles, and 99% with 15 cycles, with no significant improvement. Therefore, 10 cycles were set as the cycle number threshold to achieve efficient exhaust in the shortest time. After the operation is completed, the system re-acquires acoustic and optical signals for baseline calibration. When the ultrasonic attenuation amplitude drops below 0.1dB / cm, the normal four-layer progressive verification and detection logic is restored.

[0031] This failure protection mechanism can quickly remove air bubbles from the immersion liquid, restore the accuracy of ultrasonic detection, avoid detection failure or false alarms caused by air bubbles, and improve the system's automated operation capability without manual intervention.

[0032] The second failure protection scenario: When the composite sensing module fails, it automatically switches to the backup monitoring logic based on the temperature of the individual battery cells; Specifically, when the system detects a continuous fault condition such as communication interruption, signal loss, or self-test failure in the ultrasonic transducer array or fiber optic sensor network, which exceeds the fault determination time threshold of 5 seconds, it determines that the composite sensing module has failed. Fault determination time threshold: The duration of instantaneous faults in the sensor modules of the test system. Data: [0.5s, 1.0s, 1.5s, 2.0s, 0.8s], all less than 5 seconds, and can recover automatically; when a real fault occurs in the module, the duration of the fault state is greater than 5 seconds. Therefore, 5 seconds is set as the fault determination time threshold to avoid instantaneous faults from falsely triggering the backup logic.

[0033] At this time, the failure protection module automatically switches the system's monitoring logic to the backup monitoring logic. The backup monitoring logic is based on the battery cell temperature and voltage signals collected by the battery management system (BMS). The backup monitoring temperature threshold is set to 60℃ and the backup monitoring temperature rise rate threshold is 5℃ / min. When the temperature of any cell exceeds 60℃ or the temperature rise rate exceeds 5℃ / min, the system directly executes the thermal propagation blocking strategy of the secondary critical zone and issues a fault alarm to notify the maintenance personnel to carry out maintenance. Backup monitoring temperature threshold: The thermal runaway trigger temperature of lithium iron phosphate cells is 80℃, and 60℃ is the critical safe temperature before thermal runaway. At this time, the cell has not yet suffered irreversible damage. Setting 60℃ as the temperature threshold can trigger the protection strategy in advance. Backup monitoring temperature rise rate threshold: The maximum temperature rise rate of the battery cell during normal charging and discharging is 2℃ / min. When the temperature rise rate exceeds 5℃ / min, it is determined that the battery cell is generating abnormal heat and is about to enter the thermal runaway stage. Therefore, 5℃ / min is set as the temperature rise rate threshold. When the fault of the composite sensing module is cleared and the self-test returns to normal, the system automatically switches back from the backup monitoring logic to the original sound and light dual-mode composite sensing and four-layer counterfeit verification logic, restoring the system's high-precision early warning capability. By using backup monitoring logic, redundant safety protection can be provided for the system when the composite sensing module fails, avoiding safety accidents caused by monitoring failure. At the same time, the threshold setting of the backup logic takes into account both the timeliness and accuracy of early warning, ensuring that the system can still achieve thermal runaway protection in the fault state, which greatly improves the system's operational reliability and fault tolerance.

[0034] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0035] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A liquid-based immersion protection distribution control system for electrochemical energy storage, characterized in that, include: A physical partitioning module is used to divide the battery box, which contains immersion liquid and at least one battery pack, into multiple physically isolated control areas; The composite sensing module includes an ultrasonic transducer array and an optical fiber sensing network distributed in each of the control areas, for real-time acquisition of acoustic and optical characteristic parameters of the immersion liquid. The feature extraction and authentication module is connected to the composite sensing module and is used to perform multi-level signal processing, extract and verify abnormal feature signals corresponding to the early characteristics of battery thermal runaway, and calculate the spatial location of the abnormal feature signals in the battery box. The graded response and flow resistance reconstruction module is connected to the feature extraction and authentication module and the fluid inlet actuator of each control area. After the authenticity of the abnormal feature signal is verified, it dynamically adjusts the fluid resistance of each control area according to its severity to reconstruct the flow distribution of the immersion liquid in the battery box, thereby achieving targeted cooling or isolation of the abnormal area.

2. The liquid-based immersion protection distribution control system for electrochemical energy storage according to claim 1, characterized in that, The feature extraction and authentication module includes a four-layer progressive verification unit, which is configured as follows: The first layer constructs an acoustic-optical baseline function based on the cooling pump speed and flow rate to filter out acoustic-optical signal noise caused by fluctuations in normal system operating conditions. The expression for the acousto-optic baseline function is: ， in, For the system's operating characteristic frequency, Provide real-time feedback of flow parameters for the system's cooling pumps; The second layer involves matching the characteristics of the remaining signals with a pre-defined interference feature library to eliminate interference signals generated by non-thermal runaway factors. The third layer matches the characteristics of the remaining signals with a pre-set library of early thermal runaway anomalies to confirm whether they possess composite characteristics that characterize the early stages of thermal runaway. The fourth layer verifies the spatiotemporal resonance consistency of the confirmed signal, determining whether the abnormal coordinates located by the ultrasonic transducer array and the refractive index change coordinates located by the fiber optic sensor network both meet the pre-set overlap threshold in time and space.

3. The liquid-integrated immersion protection distribution control system for electrochemical energy storage according to claim 2, characterized in that, The signal features included in the early-stage thermal runaway anomaly feature library are the acousto-optic signal features corresponding to microbubbles and thermal convection generated during the thermal decomposition stage of the battery SEI film, specifically manifested as follows: Ultrasonic signals with center frequency and pulse repetition frequency under liquid viscosity and pressure, and a step-like abrupt change in the refractive index around the optical fiber caused by microbubbles and thermal convection, as monitored by time-domain reflectometry.

4. The liquid-integrated immersion protection distribution control system for electrochemical energy storage according to claim 2, characterized in that, The criterion for verifying spatiotemporal resonance consistency is: ， in, These are the anomalous coordinates calculated from the ultrasonic transducer array. The coordinates of the refractive index abrupt change detected by the fiber optic sensor network. The spatial overlap threshold is set in advance, and both must occur synchronously within a millisecond time window.

5. The liquid-integrated immersion protection distribution control system for electrochemical energy storage according to claim 1, characterized in that, The graded response and flow resistance reconstruction module includes: The status assessment unit is used to determine the preset security range level to which the current abnormal status belongs based on the output of the feature extraction and counterfeit detection module. The strategy execution unit is used to generate and output control commands based on the determination result of the state evaluation unit; The fluid inlet actuator includes an electromagnetic proportional valve installed on the inlet pipeline of each control area. The control command adjusts the opening of the electromagnetic proportional valve to change the flow resistance of each control area, thereby realizing the on-demand distribution of the immersion liquid flow rate.

6. The liquid-integrated immersion protection distribution control system for electrochemical energy storage according to claim 5, characterized in that, The preset security range levels determined by the status assessment unit include: The first-level warning zone corresponds to the detection of microbubble characteristic signals lasting longer than 2 seconds, and the local immersion liquid temperature rise rate is between 2 and 5°C / min; The second-level critical zone corresponds to the detection of a continuous and significant attenuation of the ultrasonic signal, an abnormal fluctuation of more than 50mV in the voltage of a single cell, or a continuous and unabated first-level warning state.

7. The liquid-integrated immersion protection distribution control system for electrochemical energy storage according to claim 6, characterized in that, When the strategy execution unit determines that the abnormal state belongs to the first-level warning zone, it executes a targeted cooling strategy: instructs the electromagnetic proportional valve in the healthy zone to reduce its opening to increase its flow resistance, so that the coolant passively and preferentially flows to and increases the supply flow to the abnormal zone; when the abnormal state belongs to the second-level critical zone, it executes a heat propagation blocking strategy: instructs the electromagnetic proportional valve in the abnormal zone to fully open, while instructing the electromagnetic proportional valves in all other healthy zones to fully close, and performs a power-off operation on the battery pack in the abnormal zone.

8. The liquid-integrated immersion protection distribution control system for electrochemical energy storage according to claim 1, characterized in that, The fiber optic sensing network includes an optical fiber suspended in an immersion liquid, on which a fiber Bragg grating array is provided to monitor point temperature gradients, and the intrinsic Rayleigh scattering effect is used to monitor the global distribution of refractive index around the optical fiber.

9. The liquid-integrated immersion protection distribution control system for electrochemical energy storage according to claim 1, characterized in that, The ultrasonic transducer array is a reflection-transmission composite array, installed at relative positions on the inner wall of each control area. It adopts a detection method combining pulse echo and transmission methods, and is equipped with at least two ultrasonic transducers of different frequencies.

10. The liquid-based immersion protection distribution control system for electrochemical energy storage according to claim 1, characterized in that, Also includes: The failure protection module is used to automatically control the actuators of each fluid inlet to open and close at high frequency when the feature extraction and authentication module determines that there is a global acoustic attenuation caused by non-local thermal runaway factors, so as to discharge the air bubbles in the immersion liquid. Alternatively, when the composite sensing module malfunctions, it can automatically switch to backup monitoring logic based on the temperature of individual battery cells.

Images