Adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flow collaborative perception

Abstract

The present application relates to the technical field of battery cooling, and discloses an adaptive dual-mode immersed liquid cooling system based on acoustic-heat flow collaborative sensing. The system comprises an acoustic sensor, a temperature sensor, a fuzzy controller, a multi-channel controller, a pump, a cooling coil, a radiator and a water tank. The system can perform single phase change cooling operation in a low rate mode and flow phase change collaborative cooling operation in a high rate mode, and mode switching decisions are made based on frequency domain changes of acoustic characteristics and calculated values of heat flow density. The acoustic sensor collects boiling acoustic wave signals of the cooling liquid, the temperature sensor monitors the battery temperature gradient, the fuzzy controller generates control instructions based on the above signals, and the multi-channel controller controls the pump power, the cooling coil flow and the radiator power. Through collaborative sensing and dual-mode adaptive adjustment, the system improves the cooling efficiency and response speed, adapts to different battery load states, and ensures the safe and stable operation of the battery.

Description

Technical Field

[0001] This invention relates to the field of battery cooling technology, specifically to an adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flow co-sensing. Background Technology

[0002] With the rapid development of the new energy industry, the operating temperature of power batteries, as core components, directly affects their performance and safety. During high-power charging and discharging, batteries generate a large amount of heat. If this heat cannot be dissipated in time, it will lead to excessively high temperatures or uneven temperature distribution, causing problems such as capacity decay, shortened lifespan, and even thermal runaway. Therefore, an efficient and reliable cooling system has become a critical requirement for power battery applications.

[0003] Currently, common battery cooling methods include air cooling, liquid cooling, and phase change cooling. Air cooling systems are simple in structure and low in cost, but their heat dissipation efficiency is limited, making it difficult to meet heat dissipation requirements in high-power scenarios. Liquid cooling systems remove heat through liquid circulation, offering better heat dissipation capabilities than air cooling. However, traditional liquid cooling systems often employ a single flow pattern, making it difficult to dynamically adjust the cooling intensity based on the battery's thermal state, which can easily lead to over-cooling or under-cooling.

[0004] Phase change cooling utilizes the characteristic of coolant absorbing a large amount of latent heat during phase change, offering the advantage of high heat dissipation efficiency. However, traditional phase change cooling systems are mostly passive or operate in a single mode, unable to adaptively adjust according to changes in battery heat flux density. When the battery is under low load, excessive phase change cooling may lead to energy waste; while when the battery is under high load, single phase change cooling may not be able to remove a large amount of heat in time, affecting battery performance and safety.

[0005] Existing cooling systems primarily rely on temperature signals acquired by temperature sensors. Due to the lag in temperature transmission, these signals often fail to reflect changes in the internal thermal state of the battery in a timely manner, resulting in delayed control response. Furthermore, the lack of effective monitoring of coolant flow patterns and phase change processes hinders precise adaptive control, impacting the overall performance of the cooling system. Summary of the Invention

[0006] The purpose of this invention is to provide an adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flow co-sensing to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides an adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flux co-sensing, the system comprising:

[0008] An acoustic sensor, a temperature sensor, a fuzzy controller, a multi-channel controller, a pump, a cooling coil, a radiator, and a water tank are included. The system is configured to perform a single-phase-change cooling operation in a low-rate mode and a flow-phase-change synergistic cooling operation in a high-rate mode, and to make mode switching decisions based on frequency domain changes of acoustic characteristics and calculated heat flux density. Specifically, the acoustic sensor acquires the acoustic wave signal of the coolant boiling, the temperature sensor monitors the battery temperature gradient, the fuzzy controller generates control commands based on the acoustic wave signal and temperature gradient, and the multi-channel controller regulates the pump power, cooling coil flow rate, and radiator power.

[0009] Preferably, the operation in the low-rate mode includes: turning off the pump to stop the liquid flow, the battery generating heat to cause the coolant to absorb heat and evaporate to form nucleate boiling, the gaseous coolant rising to the top and condensing into liquid after contacting the cooling coil, and then falling back to the liquid surface to form an internal circulation; the acoustic sensor collects the high-frequency signal of nucleate boiling, and the temperature sensor monitors the gentle temperature gradient.

[0010] Preferably, the system further includes a signal processor configured to receive the high-frequency signal of nucleated boiling acquired by the acoustic sensor and the gentle temperature gradient monitored by the temperature sensor, analyze the frequency domain characteristics of the acoustic signal and the changing trend of the temperature gradient, and generate a boiling state determination result; the boiling state determination result is used to maintain the low magnification mode.

[0011] Preferably, the operation in the high-rate mode includes: starting the pump to pump part of the hot coolant in the tank into the cooling tower for cooling, while simultaneously replenishing an equal amount of cold coolant from the water tank to enhance convection; the power of the cooling coil and radiator increases synchronously to achieve dual heat dissipation of latent heat and sensible heat.

[0012] Preferably, the fuzzy controller is configured to receive the low-frequency acoustic signal of transition boiling captured by the acoustic sensor and the sudden increase in heat flux density monitored by the temperature sensor, and generate a transition boiling risk assessment result based on the calculated heat flux density value and the frequency domain change of the acoustic signal; the transition boiling risk assessment result triggers pump start-up and coolant replenishment operations.

[0013] Preferably, the mode switching decision includes: when the calculated heat flux density value is between the low-magnification mode threshold and the high-magnification mode threshold, and the frequency domain change of the acoustic signal shows signs of over-boiling, the mode switching is triggered in advance; the mode switching decision is based on the acoustic signal analysis results and the calculated heat flux density value output by the signal processor.

[0014] Preferably, the system further includes a fuzzy optimizer configured to receive the boiling state determination result and the temperature deviation value, and dynamically correct the temperature setpoint; the temperature setpoint is output to the multi-channel controller for adjusting the heat dissipation parameters.

[0015] Preferably, the multi-channel controller is configured to receive the temperature setpoint output by the fuzzy optimizer and the control command generated by the fuzzy controller, and precisely regulate the pump power, cooling coil flow rate and radiator power; the regulation ensures that the heat dissipation capacity covers the higher heat flux density range in the high-rate mode.

[0016] Preferably, the signal processor is further configured to fuse the frequency domain features of the acoustic signal acquired by the acoustic sensor and the calculated value of the heat flux density monitored by the temperature sensor to generate a collaborative sensing feature vector; the collaborative sensing feature vector is input to the fuzzy controller to generate control commands.

[0017] Preferably, the calculated heat flux density is obtained by analyzing the battery temperature gradient and coolant temperature change monitored by the temperature sensor; the calculated heat flux density is output to the mode switching decision module as one of the mode switching criteria.

[0018] Compared with the prior art, the beneficial effects of the present invention are:

[0019] By incorporating acoustic and temperature sensors, the system achieves coordinated monitoring of coolant boiling acoustic signals and battery temperature gradients, providing a more comprehensive and timely basis for system mode switching and parameter adjustment. Compared to traditional cooling systems that rely solely on temperature signals, introducing frequency domain changes in acoustic characteristics as a sensing indicator can more quickly reflect the phase change state and flow of the coolant, thereby allowing for earlier detection of changes in battery thermal state and reducing control lag.

[0020] The system is designed to operate in two modes: low-rate and high-rate. In low-rate mode, a single phase-change cooling operation is performed, which can fully utilize the latent heat of phase change for efficient heat dissipation when the battery is under low load, avoiding unnecessary energy consumption and meeting energy-saving requirements. In high-rate mode, a combined flow-phase-change cooling operation is performed. When the battery is under high load and generates a large amount of heat, the pump, cooling coil, and radiator work together to enhance the flow and heat exchange of the coolant, remove more heat in time, and prevent the battery temperature from becoming too high.

[0021] The application of fuzzy controllers enables the system to generate reasonable control commands based on complex changes in acoustic signals and temperature gradients. Compared to traditional linear control methods, it is better able to adapt to the nonlinear and uncertain changes in battery thermal state, improving the flexibility and adaptability of system control. The multi-channel controller's coordinated regulation of pump power, cooling coil flow rate, and radiator power enables coordinated operation of all components in the cooling system, avoiding the limitations of adjusting a single component and improving the overall cooling performance of the system.

[0022] By using frequency domain variations based on acoustic characteristics and calculated heat flux density for mode switching decisions, the system ensures accurate and timely mode switching. This allows the system to automatically select the appropriate cooling mode under different battery load conditions, guaranteeing energy-saving operation under low loads while meeting the high-efficiency heat dissipation requirements under high loads. Simultaneously, the dual-mode switching and adaptive parameter adjustment enable the system to maintain good cooling performance under various operating conditions, which is beneficial for maintaining battery performance stability and extending battery life. Attached Figure Description

[0023] Figure 1 This is a timing diagram of the adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flux co-sensing as described in this invention.

[0024] Figure 2 A flowchart for low-magnification mode operation;

[0025] Figure 3 A flowchart for determining the risk of overboiling in a fuzzy controller;

[0026] Figure 4 A flowchart for dynamically correcting the temperature setpoint for the fuzzy optimizer;

[0027] Figure 5 A flowchart for generating collaborative sensing feature vectors for a signal processor. Detailed Implementation

[0028] 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.

[0029] Please see Figure 1 This invention provides an adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flux co-sensing, the system comprising:

[0030] The system comprises acoustic sensors, temperature sensors, a fuzzy controller, a multi-channel controller, a pump, cooling coils, a radiator, and a water tank. It is configured to perform single-phase-change cooling in low-rate mode and flow-phase-change synergistic cooling in high-rate mode, making mode switching decisions based on frequency domain variations of acoustic characteristics and calculated heat flux density. The acoustic sensors acquire acoustic signals of coolant boiling, the temperature sensors monitor the battery temperature gradient, the fuzzy controller generates control commands based on the acoustic signals and temperature gradient, and the multi-channel controller regulates pump power, cooling coil flow rate, and radiator power. The system achieves dynamic switching of cooling modes through acoustic-heat flux synergistic sensing to adapt to the heat dissipation requirements under different battery heat generation conditions.

[0031] Example 1: See Figure 2 In low-rate mode, the system performs a single-phase-change cooling operation, with the pump off and the coolant remaining static within the tank. The battery module is immersed in the coolant, and the heat it generates is transferred to the surrounding liquid medium through natural convection. Upon absorbing heat, the coolant undergoes a phase change, forming nucleated boiling on the battery surface. During boiling, numerous uniformly distributed microbubbles are generated, with bubble diameters maintained below millimeters. The gaseous coolant rises to the top of the tank due to its decreased density, contacting the cooling coils. A low-temperature working fluid flows inside the cooling coils, causing the gaseous coolant to condense into a liquid state on the coil surface, releasing latent heat. The condensed liquid coolant then falls back to the surface under gravity, forming a closed internal circulation heat dissipation path.

[0032] Acoustic sensors are deployed on the side walls and bottom of the enclosure to collect acoustic signals during the coolant boiling process in real time. During nucleate boiling, the acoustic signals exhibit stable high-frequency characteristics, with the dominant frequency range concentrated between 50kHz and 200kHz. The sensors convert acoustic vibrations into electrical signals using piezoelectric transducers, which are then processed by a preamplifier and transmitted to a signal processor. Temperature sensors are distributed, with multiple temperature measurement points placed in key hot areas on the battery surface (such as electrode connections and cell gaps) to simultaneously monitor the temperature distribution at different depths of the coolant. Temperature data is recorded at a sampling frequency of 10 times per second, generating axial and radial temperature gradient maps of the battery surface.

[0033] For the acoustic signal, the processor performs a Fast Fourier Transform to extract frequency domain feature parameters, including: 1) the energy proportion in the frequency band above 200kHz; 2) the peak power spectral density in the 50-100kHz frequency band; and 3) the amplitude fluctuation coefficient of the characteristic frequency. For the temperature data, the processor calculates the rate of change of temperature difference between adjacent temperature measurement points on the battery surface and plots a time-series curve of the temperature gradient. When the proportion of high-frequency acoustic energy is consistently higher than a preset threshold (e.g., 85% of the total energy) and the rate of change of temperature gradient is less than 0.5℃ / s, it is determined that the current state is a stable nucleate boiling state.

[0034] The fuzzy controller receives the boiling state determination result from the signal processor. The determination result is a binary logic quantity: when the high-frequency characteristics continuously meet the conditions and the temperature gradient is stable, it outputs a "maintenance mode" command; if the low-frequency component of the acoustic spectrum shows an abnormal increase or the temperature gradient changes abruptly, it outputs a "ready to switch" warning. According to the "maintenance mode" command, the multi-channel controller keeps the pump in the off state and controls the working fluid flow rate in the cooling coil at a baseline value (e.g., 0.2 m / s), and the radiator fan runs at the lowest speed.

[0035] The system enhances natural circulation efficiency through physical structural design. The top of the housing features a conical gas-collecting structure to accelerate the accumulation of gaseous working fluid in the cooling coils. The coils employ a serpentine, multi-layered layout to increase the condensation surface area. A flow guide channel at the bottom of the housing facilitates the return of condensate to the battery's heat source area. A low-boiling-point fluorinated ether compound is selected as the coolant, whose nucleation boiling characteristic at 40°C matches the battery's low-rate heat generation conditions. Temperature sensor data is synchronously used for heat flux density-assisted calculation. By analyzing the dynamic temperature difference between the battery surface and the coolant, combined with material thermal conductivity parameters, the processor estimates the local heat flux density distribution. This data is cross-validated with acoustic signal characteristics: when the calculated heat flux density stabilizes in the 5-10 W / cm² range and the high-frequency characteristics of the acoustic waves do not attenuate, the applicability of the low-rate mode is further confirmed.

[0036] If a sudden low-frequency component below 10kHz appears in the acoustic signal (indicating bubble aggregation), or if the local temperature gradient of the battery exceeds 3℃ / cm, the mode switching preparation procedure is immediately triggered. At this time, the fuzzy controller initiates adaptive adjustment of the cooling coil flow rate, gradually increasing the working fluid flow rate to improve condensation efficiency while maintaining natural circulation, preparing for a possible high-rate mode switch. Throughout the low-rate operation, energy transfer relies entirely on latent heat exchange due to phase change. The heat absorbed by coolant evaporation and the heat released by condensation are dynamically balanced, and the system power consumption comes only from the basic operating requirements of the cooling coil and radiator, reducing energy consumption by approximately 70% compared to the forced circulation mode. Continuous monitoring of the acoustic signal can identify abnormal operating conditions, such as a sudden decrease in bubble frequency indicating boiling suppression. In this case, the system automatically fine-tunes the cooling coil temperature setpoint to avoid the risk of localized dry burning.

[0037] Example 2: See Figure 3 In high-rate mode, the system performs a flow-phase change synergistic cooling operation. The pump switches from standby to operation, driving a portion of the hot coolant in the tank through the circulation pipeline into the external cooling tower. The cooling tower employs a multi-stage finned tube heat exchanger structure, where the hot coolant exchanges heat with forced convection air as it flows through the heat exchange tubes. Simultaneously, cold coolant in the water tank is injected into the bottom of the tank through the replenishment pipeline to replenish the volume of liquid pumped out, maintaining liquid level balance. The operating power of the cooling coils and radiators increases synchronously according to the real-time heat load, forming a dual heat dissipation mechanism combining latent heat and sensible heat.

[0038] When the battery enters high-rate operation, the coolant boiling characteristics gradually shift from nucleate boiling to transitional boiling. Sensors detect significant changes in the acoustic signal spectrum, primarily manifested as a gradual increase in the energy proportion of low-frequency components from 10kHz to 50kHz, while high-frequency components above 200kHz attenuate accordingly. The temperature sensor network detects a step increase in the heat flux density on the battery surface, with the local temperature gradient expanding to 3 to 5 times the normal value within a short period. These changes are transmitted to the signal processing unit via a high-speed data acquisition channel, with the sampling frequency increased to 50 times per second to adapt to dynamic operating conditions.

[0039] In addition to conventional spectrum analysis, a time-frequency joint analysis method is employed, using continuous wavelet transform to track the instantaneous frequency changes of the acoustic signal. The processor identifies three key characteristics: 1) the energy proportion in the low-frequency band (10-50kHz) exceeds 40% of the total energy; 2) the fluctuation amplitude of the characteristic frequency increases to twice the baseline value; and 3) intermittent pulse signals appear in the acoustic envelope. The temperature data analysis module simultaneously calculates the spatial distribution of heat flux density. When it detects that more than 20% of the monitoring points have heat flux density values ​​exceeding a preset threshold, a heat flux anomaly event marker is generated.

[0040] The fuzzy controller receives transitional boiling risk assessment parameters from the signal processor. These parameters include: the proportion of low-frequency acoustic energy, the proportion of the area exceeding the heat flux density limit, and the acceleration value of the temperature gradient change. The controller's internal knowledge base stores 256 fuzzy rules, mapping the input parameters to three risk levels. When the following conditions are simultaneously met: low-frequency energy proportion > 45%, heat flux exceeding the limit area > 30%, and gradient acceleration > 0.8℃ / s², a high-level risk assessment result is output. This result triggers three parallel control commands: pump power is increased to 80% of the rated value, cooling coil working fluid flow rate is increased by 50%, and radiator fan is switched to high speed.

[0041] The multi-channel controller executes the instruction set of the fuzzy controller. The pump uses a variable frequency drive, and its speed smoothly increases from 0 to 3000 rpm within 5 seconds, with the flow rate controlled within the range of 15-20 L / min. The working fluid flow rate of the cooling coil is precisely controlled by an electric regulating valve. The valve opening is linked to the radiator fan speed to ensure a match between sensible heat dissipation and latent heat dissipation capabilities. The forced convection velocity of the coolant inside the casing is maintained in the range of 0.3-0.5 m / s, ensuring sufficient heat exchange efficiency while avoiding instability in the gas-liquid two-phase flow due to excessively high flow velocity.

[0042] When the temperature of the return liquid is detected to be higher than the set value, the auxiliary cooling fan unit is automatically activated, increasing the airflow to 120% of the design maximum. Multiple temperature monitoring points are installed inside the cooling tower to calculate the heat transfer efficiency of each section of the heat exchange tubes in real time. When the efficiency of a section drops by more than 15%, a descaling alarm is automatically triggered. The replenishment system adopts a dual-channel design: the main channel replenishes the liquid at a normal flow rate, while the backup channel activates emergency replenishment when a rapid drop in liquid level is detected. When the acoustic sensor detects a continuous presence of ultra-low frequency components below 5kHz, it is determined to be a precursor to film boiling, and a three-level emergency response is immediately executed: pump power is increased to 100%, the working fluid temperature of the cooling coil is reduced by 5°C, and the replenishment flow rate is increased by 30%. Simultaneously, a bubble breaking device is activated, using an ultrasonic oscillator to generate high-frequency mechanical waves in the liquid, causing large bubbles to break into smaller bubbles, restoring nucleate boiling characteristics. The temperature field reconstruction module generates a three-dimensional temperature field distribution map every 10 seconds based on distributed sensor data. This module uses an inversion algorithm to reconstruct a continuous temperature field from discrete point temperature measurement data, identifying potential heat accumulation areas. When a hot zone exceeding 5cm in diameter is detected, the pump's flow distribution is automatically adjusted to increase the coolant delivery to that area. The coolant flow path is optimized through a guide vane design, creating a spiral upward flow pattern to enhance comprehensive coverage of the battery module.

[0043] The acoustic monitoring system, based on high-speed sampling, adds acoustic emission source localization functionality. Using four acoustic sensors positioned at different locations within the enclosure, a time-difference localization method is employed to pinpoint the precise location of abnormal sound sources. When an abnormal acoustic signal is detected continuously near a specific battery cell, the system automatically designates that cell as a key monitoring target, increases its temperature sampling frequency to 100 times per second, and assigns a dedicated coolant branch to that area.

[0044] When the main pump fails, the backup pump switches over within 200ms; the cooling coil adopts a dual-loop design, maintaining 60% of its heat dissipation capacity even if a single loop fails; the radiator is equipped with an emergency air-cooling module, maintaining basic heat dissipation function through backup power during power outages. All key parameters are subject to a triple verification mechanism, and any sensor failure will be immediately identified and isolated to avoid misjudgments leading to incorrect mode switching.

[0045] Throughout high-rate operation, the system achieves efficient heat dissipation through the synergistic effect of flow and phase change. Forced convection removes most of the sensible heat, while the phase change process absorbs concentrated heat in key hot zones; these two mechanisms complement each other. Real-time monitoring of acoustic signals provides early warning of changes in the boiling state, enabling control strategies to intervene proactively and prevent temperature runaway due to insufficient heat dissipation. A multi-parameter fusion decision-making mechanism ensures the accuracy of mode switching timing, preventing premature switching that wastes energy or premature switching that results in insufficient heat dissipation.

[0046] Example 3: See Figure 4 The mode switching decision module continuously receives two types of key data from the signal processor: the frequency domain feature matrix of the acoustic signal and the heat flux density distribution map. Acoustic signal analysis employs an improved Mel frequency cepstral coefficient extraction method to convert the original acoustic signal into a 40-dimensional feature vector, where dimensions 12 to 18 are specifically used to identify signs of excessive boiling. Heat flux density calculation introduces a spatial weighting algorithm, applying confidence weights to measurements from different regions of the battery surface. The calculation formula is as follows:

[0047]

[0048] in: Indicates the effective heat flux density. Let i be the heat flux density value at the i-th measurement point. The weighting coefficient for this measurement point (ranging from 0.5 to 1.5) is given by n, where n is the total number of valid measurement points. The weighting coefficient is dynamically adjusted based on the distance between the sensor and the heat source and the stability of historical data. Sensors that are closer to the heat source and have smaller data fluctuations are given higher weights.

[0049] The initial judgment is based on the absolute value and trend of heat flux density. When the value remains within 80%-120% of the mode switching threshold range for 3 consecutive minutes, a secondary judgment is triggered. The secondary judgment analyzes the variability of the Mel coefficients of the acoustic signal, focusing on monitoring the fluctuation of the 15th dimension coefficient (corresponding to the 35kHz frequency band). This coefficient remains relatively stable during the nucleus boiling phase, with its standard deviation typically less than 0.15; when the standard deviation exceeds 0.25 for 10 consecutive samples, a risk of over-boiling is identified. The final judgment integrates the results of the first two levels, generating a mode switching command when both the heat flux density threshold condition and the acoustic variation condition are simultaneously met.

[0050] The acoustic signal is first subjected to adaptive bandwidth filtering to eliminate environmental noise interference; then phase correction processing is performed to compensate for the transmission delay between different sensors. Temperature data is filtered using a sliding window mean filter, with the window width automatically adjusted according to the sampling frequency. A 5-point window is used for high-speed sampling (>50Hz), and the window is expanded to 11 points for low-speed sampling (≤10Hz). The processed data is then aligned using a timestamp module to ensure strict synchronization between acoustic and thermal data, with the time deviation controlled within 1ms.

[0051] The fuzzy optimizer employs an incremental parameter adjustment strategy, and its core function is to dynamically correct the temperature setpoint based on the real-time system status. This value is initially set to the lower limit of the coolant saturation temperature. When a sustained stable boiling state is detected, it is gradually increased at a rate of 0.2°C per minute; when signs of overboiling appear, it is rapidly decreased at a rate of 0.5°C per minute. The optimizer maintains a historical state queue, recording the results of the most recent 20 boiling state evaluations. When the proportion of "stable" states in the queue exceeds 75%, the setpoint is increased; otherwise, it is decreased.

[0052] The instruction mapping mechanism of the multi-channel controller enables coordinated adjustment of control parameters and receives the output of the fuzzy optimizer. After the value is calculated, the controller decomposes it into three sub-instructions: the pump power adjustment instruction corresponds to the flow control loop, the cooling coil adjustment instruction corresponds to the temperature control loop, and the radiator adjustment instruction corresponds to the airflow control loop. Cross-coupling coefficients are set between each control loop, so that when a parameter in one loop is adjusted, the parameters of other loops are automatically adjusted according to a preset ratio. For example, for every 10% increase in pump flow rate, the working fluid velocity in the cooling coil increases by 8%, and the radiator airflow increases by 6%, maintaining the dynamic balance of all parts of the system.

[0053] When switching from low-rate mode to high-rate mode, the pump does not immediately reach the target speed, but accelerates gradually over 30 seconds following an S-curve. The working fluid flow rate in the cooling coil is also gradually adjusted to avoid hydraulic shock. During this period, the acoustic sensor switches to high-speed sampling mode, increasing the sampling frequency from the usual 10kHz to 50kHz to promptly capture subtle changes in the boiling state. The temperature sensor's data upload interval is reduced from 1 second to 0.1 seconds, providing more timely data support for control decisions.

[0054] Level 1 response addresses minor parameter fluctuations by logging the event without taking any control action. Level 2 response initiates fine-tuning of parameters, making minor adjustments within the current mode framework. Level 3 response triggers mode switching preparation, with all actuators entering standby mode. Level 4 response immediately executes a forced mode switch and activates all cooling resources. Response level is determined based on a duration threshold; if the same anomaly persists for 30 seconds without resolution, the response level is automatically escalated.

[0055] The data storage module employs a circular buffer structure, always retaining the system's operational data for the most recent 30 minutes. This data includes raw acoustic signal segments, temperature sampling values, control command sequences, etc., maintaining the original data resolution during storage. When a mode switching event occurs, a 5-minute snapshot of data before and after the event is automatically saved for subsequent analysis. Data compression combines lossy and lossless compression methods; μ-law compression is used for acoustic signals, and differential coding compression is used for temperature data.

[0056] The human-computer interface displays key parameters for mode switching decisions in real time, including heat flux density trend curves, acoustic characteristic spectrum diagrams, and boiling state assessment indicator lights. Operators can view the history of mode switching, including the timestamp of each switch, the triggering reason, and parameter comparisons before and after the switch. The interface provides manual intervention functionality, allowing experienced operators to fine-tune parameters based on automated decisions; however, all manual interventions are recorded and incorporated into the subsequent learning process of automated decision-making.

[0057] The system maintains a self-diagnostic function, periodically checking the health status of each sensor. It verifies the sensitivity of acoustic sensors by injecting test signals and calibrates the measurement accuracy of temperature sensors using a known heat source. When a sensor's performance degrades by more than 20%, its weight in decision-making is automatically reduced, and redundant sensors compensate for data loss. All diagnostic results are compiled into an equipment health report for predictive maintenance planning.

[0058] Example 4: See Figure 5 The multi-channel controller operates based on the temperature setpoint output by the fuzzy optimizer and the control commands generated by the fuzzy controller. It achieves precise regulation of pump power, cooling coil flow, and radiator power through a multivariable coordinated control algorithm. The system employs a hierarchical control architecture: the upper-level decision unit generates target parameters, the lower-level execution unit implements closed-loop regulation, and real-time information exchange is achieved through a data bus. In a typical operating case, the system records that the battery pack enters a high-load state, at which point the cooling demand changes significantly, and the multi-channel controller initiates the corresponding adjustment program.

[0059] The cooperative sensing feature vector generated by the signal processor contains parameters in 12 dimensions, which are categorized and stored according to type and function. Table 1 shows the changes in key parameters of the cooperative sensing feature vector recorded during a mode switch.

[0060] Table 1: Examples of changes in collaborative sensing feature vector parameters under high magnification mode.

[0061]

[0062] The pump power regulation employs a graded speed-changing strategy, with six power levels based on coolant circulation requirements. When the collaborative sensing feature vector indicates that parameters FV-03 and FV-05 simultaneously exceed the threshold, the controller gradually increases the pump power from the initial level 2 (40% of rated power) to level 4 (70% of rated power). An inertial delay algorithm is incorporated into the power regulation process to prevent large power fluctuations in a short period from impacting the motor system. Each power increase is spaced at least 15 seconds apart, during which the motor winding temperature is monitored. If the temperature rise rate exceeds the safety limit, further power increases are paused.

[0063] The cooling coil flow control employs a pressure-flow composite regulation mode. The controller compares the inlet and outlet water pressure differences in real time and cross-verifies this pressure difference signal with the target flow rate value. When the pressure difference signal deviates from the flow meter reading by more than 5%, the flow calibration program is initiated, compensating for the measurement error by adjusting the opening of the electric valve. In the case shown in Table 1, a decrease in the FV-07 parameter triggers a flow redistribution command. The controller increases the valve opening of the corresponding branch in the hot zone by 3%-5%, while simultaneously reducing the flow rate in the cold zone branch, restoring the flow uniformity index to above 0.9.

[0064] The controller establishes a mapping table between radiator fan speed and cooling coil outlet temperature, predicting the required heat dissipation capacity based on temperature change trends. When the FV-11 parameter continues to rise, the system determines that the radiator efficiency has decreased. At this point, it not only increases the fan speed but also simultaneously initiates the radiator self-cleaning program, using a short-term reverse airflow to blow away accumulated dust. Under extreme conditions, the controller can temporarily exceed the rated power limit, increasing the radiator power to 115% of the design value for a duration not exceeding 10 minutes.

[0065] The raw acoustic wave signals acquired by the acoustic sensor array are preprocessed to extract six feature parameters each in the time and frequency domains. Temperature measurement data acquired by the temperature sensor network undergoes spatial interpolation to generate temperature field distribution feature parameters. These parameters are then selected using a feature selection algorithm to identify the 12 most representative dimensions, forming the final collaborative sensing feature vector. During operation, the system updates the feature vector every 30 seconds, triggering a real-time update when significant parameters change by more than 5%.

[0066] The rule base contains 128 production rules, covering control strategies for various typical operating conditions. The case library stores historically successful control cases, prioritizing validated control parameters when encountering similar conditions. In the cases in Table 1, the controller retrieved three historical cases with a matching degree exceeding 80% with the current feature vector. Therefore, it generated initial instructions by integrating the control parameters from these cases, and then fine-tuned them based on real-time feedback. Operators can observe real-time data such as pump power percentage, flow rate histograms for each branch of the cooling coil, and radiator power curves. The interface uses different colors to indicate parameter status: green indicates normal operating range, yellow indicates parameters requiring attention, and red indicates abnormal values ​​requiring immediate intervention. In the cases in Table 1, parameter FV-03 turned yellow (warning), and parameter FV-11 showed a red alarm; the system automatically popped up the corresponding handling suggestion dialog box.

[0067] When a delay in the response of an actuator exceeds a set time, the controller automatically switches to the backup control channel. For example, if an abnormality occurs in the flow regulation of the cooling coil, the system immediately switches to the pressure-dominated control mode, indirectly controlling the flow by maintaining a stable differential pressure. All fault events generate detailed diagnostic reports, including the time of occurrence, duration, measures taken by the system, and the final recovery status. The system stores complete operating data for the past three months, allowing playback of control processes for any time period. Analysts can retrieve complete data from the case studies in Table 1 to view the entire process from parameter anomalies to system recovery, including the issuance time of each control command, the response of the actuators, and the final regulation effect. This data is used to optimize control parameters and adjust the weight allocation of eigenvectors. Each successful control case is evaluated for its effectiveness, and excellent cases are added to the case library for future reference. The system reviews and analyzes all control decisions monthly, statistically analyzing the success rate of various decisions and adjusting the confidence level of each rule in the rule base. For recurring problem conditions, the controller generates specific optimization tasks, using simulation to find better control strategies.

[0068] Example 5: The heat flux density calculation module acquires raw temperature data through a distributed temperature sensor network. This network contains 48 temperature measurement points arranged in a matrix on the surface of the battery pack, with 36 points located at the cell gaps and 12 points located at the busbar connections. Each temperature measurement point uses a surface-mount platinum resistance temperature sensor, with measurement accuracy controlled within ±0.1℃, and the sampling frequency is uniformly set to 50Hz. Simultaneously, eight floating temperature measurement units are arranged inside the coolant, located at different heights within the housing, to monitor the spatial distribution of the liquid temperature.

[0069] The system divides the battery surface into 6×8 grid cells, with each cell containing 4 temperature measurement points. The processor calculates the temperature difference within each cell in real time, recording three key gradient parameters: the maximum temperature difference within the cell, the temperature difference between adjacent cells, and the ratio of axial to radial temperature differences. These parameters are updated every 5 seconds, forming a spatiotemporal distribution map of the temperature gradient. When a cell experiences a temperature abrupt change exceeding a set threshold, the sampling frequency for that region is automatically increased to 100Hz.

[0070] The processor analyzes the thermophysical parameters of the battery materials and, combined with the spatiotemporal characteristics of temperature changes, derives the heat transfer process. The calculation process consists of two stages: the first stage establishes the thermal balance equation between the battery surface and the coolant, considering the influence of material heat capacity and contact thermal resistance; the second stage analyzes the relationship between the rate of temperature change and heat flux density, and determines the direction and intensity of heat flux through time series analysis. The calculation cycle is synchronized with temperature sampling, outputting a local heat flux density value every 20 milliseconds.

[0071] The raw temperature data is first filtered by an outlier filtering algorithm to remove anomalies caused by transient sensor malfunctions. Next, a temporal consistency check is performed, comparing the logical correlation between adjacent sensor data. Finally, a spatial rationality check is conducted to ensure that the temperature distribution conforms to the basic laws of heat conduction. The verified data is labeled with a confidence level; high-confidence data is used directly for calculations, while low-confidence data requires further verification. The image resolution matches the temperature sensor grid, with each pixel corresponding to an actual 5mm × 5mm area. The heat map uses pseudo-color encoding, with blue areas representing heat flux densities below 5W / cm² and red areas representing above 20W / cm². The system also generates a heat flux density change rate curve, reflecting the dynamic trend of heat load changes in different areas. After receiving the heat flux density calculation results, the mode switching decision module performs multi-dimensional analysis. The analysis includes: the migration trajectory of the peak heat flux density location, the area expansion rate of high-heat regions, and the heat flux direction consistency index. When more than 30% of the area's heat flux density remains within the 12-18 W / cm² range for 3 minutes, and the area of ​​the high-heat region expands by more than 5% per minute, the mode switching evaluation procedure is triggered. The evaluation procedure combines changes in acoustic characteristics for joint judgment.

[0072] The system adds a timestamp to each frame of heat flux density data, and the acoustic signal acquisition system inserts audio marker pulses at the same time points. During data processing, by identifying these synchronization markers, the heat flux density distribution map is matched with the corresponding acoustic spectrum map to establish a thermo-acoustic coupling feature database. This database stores a complete snapshot every 10 minutes, retaining records from the most recent 24 hours.

[0073] The system accuracy is periodically verified by injecting a standard heat source, which is a thin-film heating element of known power, attached to the non-operating area of ​​the battery pack. During calibration, the deviation between the calculated heat flux density and the actual input power is compared to generate correction coefficients for each region. These coefficients are applied to subsequent calculations to keep the system measurement error within ±7%. The calibration cycle is dynamically adjusted according to the system's operating status: calibrated daily during high-load periods and weekly during low-load periods.

[0074] Historical heat flux density data is used to build a battery thermal behavior model, systematically analyze heat flux distribution patterns under different operating conditions, and identify trends in the thermal characteristics of the battery pack. When a persistent abnormal heat flux concentration is detected in a specific area, a battery health status report is automatically generated, indicating potential localized aging or poor contact issues. This report is displayed through a human-machine interface and includes a map locating the abnormal heat flux area, statistics on the duration of the abnormality, and comparative analysis with historical data.

[0075] Operators can observe dynamically updated heat maps and view real-time heat flux density values ​​at any location. The system provides historical data playback functionality, allowing users to select a specific time period to view the evolution of heat flux density. The interface features a two-level alarm mechanism: a yellow alert is triggered when the local heat flux density exceeds 25 W / cm², and a red alarm is triggered and an emergency cooling procedure is automatically initiated when it exceeds 30 W / cm².

[0076] The heat flux density calculation results are shared in real time with the fuzzy controller to optimize control parameters; transmitted to the signal processor to assist in acoustic feature analysis; and provided to the mode switching decision module as one of the core criteria. Data transmission employs a redundant channel design, with the primary channel based on the real-time Ethernet protocol and the backup channel using a CAN bus to ensure communication reliability. The system maintains a heat flux density calculation log, recording in detail the input parameters, processing procedures, and output results for each calculation. Log data includes complete information such as raw temperature values, gradient calculation results, heat flux density distribution maps, and confidence level assessments. Log files are stored in hourly segments with a retention period of 30 days for post-analysis and system optimization. The log management interface supports conditional query functions, allowing retrieval of historical records by time range, battery region, heat flux threshold, and other criteria.

[0077] 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.

[0078] 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. Adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flow co-sensing, comprising acoustic sensor, temperature sensor, fuzzy controller, multi-channel controller, pump, cooling coil, radiator and water tank, characterized in that: The system is configured to perform single-phase change cooling in low-rate mode and flow-phase change synergistic cooling in high-rate mode, and to make mode switching decisions based on frequency domain changes of acoustic characteristics and calculated heat flux density; wherein, the acoustic sensor collects the boiling acoustic wave signal of the coolant, the temperature sensor monitors the battery temperature gradient, the fuzzy controller generates control commands based on the acoustic wave signal and temperature gradient, and the multi-channel controller regulates the pump power, cooling coil flow rate and radiator power.

2. The adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flow co-sensing of claim 1, wherein, The operation in the low-rate mode includes: turning off the pump to stop the liquid flow, the battery generating heat to cause the coolant to absorb heat and evaporate to form nucleate boiling, the gaseous coolant rising to the top and condensing into liquid after contacting the cooling coil, and then falling back to the liquid surface to form an internal circulation; the acoustic sensor collects the high-frequency signal of nucleate boiling, and the temperature sensor monitors the gentle temperature gradient.

3. The adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flux co-sensing as described in claim 2, characterized in that, The system also includes a signal processor configured to receive the high-frequency signal of nucleated boiling acquired by the acoustic sensor and the gentle temperature gradient monitored by the temperature sensor, analyze the frequency domain characteristics of the acoustic signal and the changing trend of the temperature gradient, and generate a boiling state determination result; the boiling state determination result is used to maintain the low magnification mode.

4. The adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flux co-sensing as described in claim 3, characterized in that, The operation in the high-rate mode includes: starting the pump to pump part of the hot coolant in the tank into the cooling tower for cooling, while simultaneously replenishing an equal amount of cold coolant from the water tank to enhance convection; the power of the cooling coil and radiator increases synchronously to achieve dual heat dissipation of latent heat and sensible heat.

5. The adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flux co-sensing as described in claim 4, characterized in that, The fuzzy controller is configured to receive the low-frequency acoustic signal of transition boiling captured by the acoustic sensor and the sudden increase in heat flux density monitored by the temperature sensor, and generate a transition boiling risk assessment result based on the calculated heat flux density value and the frequency domain change of the acoustic signal; the transition boiling risk assessment result triggers pump start-up and coolant replenishment operations.

6. The adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flux co-sensing as described in claim 5, characterized in that, The mode switching decision includes: when the calculated heat flux density value is between the low-magnification mode threshold and the high-magnification mode threshold, and the frequency domain change of the acoustic signal shows signs of over-boiling, the mode switching is triggered in advance; the mode switching decision is based on the acoustic signal analysis results and the calculated heat flux density value output by the signal processor.

7. The adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flux co-sensing as described in claim 6, characterized in that, The system also includes a fuzzy optimizer configured to receive the boiling state determination result and temperature deviation value, and dynamically correct the temperature setpoint; the temperature setpoint is output to the multi-channel controller for adjusting the heat dissipation parameters.

8. The adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flux co-sensing as described in claim 7, characterized in that, The multi-channel controller is configured to receive the temperature setpoint output by the fuzzy optimizer and the control command generated by the fuzzy controller, and to precisely regulate the pump power, cooling coil flow rate and radiator power; the regulation ensures that the heat dissipation capacity covers the higher heat flux density range in the high-rate mode.

9. The adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flux co-sensing as described in claim 8, characterized in that, The signal processor is further configured to fuse the frequency domain features of the acoustic signal collected by the acoustic sensor and the calculated value of the heat flux density monitored by the temperature sensor to generate a collaborative sensing feature vector; the collaborative sensing feature vector is input to the fuzzy controller to generate control commands.

10. The adaptive dual-mode immersion liquid cooling system based on acoustic-thermal flux co-sensing as described in claim 9, characterized in that, The calculated heat flux density is obtained by analyzing the battery temperature gradient and coolant temperature changes monitored by the temperature sensor; the calculated heat flux density is output to the mode switching decision module as one of the mode switching criteria.

Images