Data center-oriented two-phase cold plate liquid cooling leakage current-limiting control method and system

Abstract

The invention discloses a data center-oriented two-phase cold plate liquid cooling leakage current-limiting control method and system, and relates to the technical field of liquid cooling leakage detection. The data center-oriented two-phase cold plate liquid cooling leakage current-limiting control method comprises the following steps: S1, collecting and preprocessing cooling liquid operation state data, control state data and execution feedback data in a liquid cooling process, and constructing a standardized cooling state data set; s2, analyzing the steady-state offset characteristics of the cooling liquid, and adjusting the supply condition of the cooling liquid; s3, evaluating the current-limiting adjusting state and dynamically correcting the working state of the electromagnetic proportional valve; and S4, comprehensively evaluating the liquid supply recovery trend, and regulating and controlling the recovery stage and the regulation path of the cooling loop in real time. The problems that in the two-phase evaporative cooling process of the data center, pipelines are complex, a cooling liquid transmission path is long, leakage control measures are difficult to take in time before cooling liquid losses, and therefore stable operation of a system is threatened are solved.

Description

Two-phase cold plate liquid cooling leakage current limiting control method and system for data centers Technical Field

[0001] This invention relates to the field of liquid cooling leakage detection technology, specifically to a method and system for limiting current control of two-phase cold plate liquid cooling leakage in data centers. Background Technology

[0002] With the continuous increase in computing power density in data centers and the sustained growth in chip heat flux, traditional air cooling methods are no longer sufficient to meet the demand for efficient heat dissipation. Two-phase cold plate liquid cooling, as a new generation of high-performance cooling technology, is widely used in server clusters with high heat loads. It forms an efficient closed-loop heat exchange cycle path by evaporating and absorbing heat in the liquid within the cold plate and then condensing and recirculating it in the condenser. It has advantages such as high energy efficiency and low noise, and is especially suitable for the high-precision temperature control requirements of large, heat-sensitive environments such as data centers.

[0003] For example, invention patent CN112911905B discloses a leak-proof system and control method for rack-mounted server indirect cold plate liquid cooling. The system includes a leak-proof tray unit, an indirect cold plate liquid cooling element leak-proof unit, a liquid cooling pipeline leak-proof unit, a control unit, a drainage pipeline, and a leak collection unit. The leak-proof tray unit monitors the safety of the liquid cooling circulation and can safely drain leaked coolant if leakage occurs. The indirect cold plate liquid cooling element leak-proof unit monitors for leaks in the indirect cold plate liquid cooling elements and prevents leaks from damaging the server. The liquid cooling pipeline leak-proof unit monitors for leaks in the liquid cooling pipelines and drains leaked coolant to the outside of the device. The control unit receives and outputs control signals. The drainage pipeline and leak collection unit collect leaked coolant. The modular design makes the entire leak-proof device more flexible and reduces the operational risks of rack-mounted liquid cooling data center systems.

[0004] For example, invention patent CN114938613B discloses a liquid cooling system and a method for locating and detecting leaks in the liquid cooling system. This liquid cooling system includes a liquid storage tank, a heat exchanger, a circulating pump, a server, a pressure acquisition component, a temperature acquisition component, and a flow acquisition component. The liquid storage tank, a first pipeline, a heat exchanger, a second pipeline, a circulating pump, a third pipeline, the server, and a fourth pipeline together form a circulation loop. The pressure acquisition component is used to collect real-time pressure values ​​of the first, second, third, and fourth pipelines; the temperature acquisition component is used to collect real-time temperature values ​​of the third and fourth pipelines; and the flow acquisition component is used to collect real-time flow values ​​of the third pipeline. This invention can accurately locate leaks, allowing users to address them promptly and preventing coolant from flowing into the server due to worsening leaks, effectively ensuring the normal operation of the server.

[0005] However, in the actual deployment of two-phase liquid cooling systems, due to the structural characteristics such as dense equipment layout, extended path depth, and complex branch nodes, the coolant is prone to pressure fluctuations and local flow abrupt changes when flowing through various branches and cold plate channels. This can induce false leakage signals or mask actual micro-leakage trends, leading to misjudgments and delayed responses. Especially with limited accuracy of flow sensors or pressure sensors, disturbances such as uneven evaporation, unbalanced condensation return flow, or sudden pressure drops in branches can significantly interfere with the stability of monitoring parameters, increasing the difficulty of leakage risk identification and control response. Existing leakage detection methods mainly rely on single-point anomaly threshold judgment, lacking a comprehensive analysis mechanism of coupled indicators in the coolant's operating state. They cannot effectively identify complex anomalies under conditions of coordinated fluctuations in both pressure and flow. At the same time, current flow-limiting control strategies are mostly based on static rule-driven approaches, lacking the ability to dynamically judge and adjust leakage evolution trends. This results in frequent problems such as delayed control responses and uncontrolled fluid replenishment during actual operation, seriously threatening the stable operation and safety of two-phase liquid cooling systems.

[0006] To address the above issues, there is an urgent need for a two-phase cold plate liquid cooling leakage current limiting control method and system for data centers. Summary of the Invention

[0007] Addressing the shortcomings of existing technologies, this invention provides a method and system for limiting leakage in two-phase cold plate liquid cooling for data centers. It solves the problem that in two-phase evaporative cooling of data centers, the pipeline is complex and the coolant transmission path is long, making it difficult to take timely measures to control leakage before coolant loss, thus threatening the stable operation of the system.

[0008] Technical solution

[0009] To achieve the above objectives, the present invention provides the following technical solution: a two-phase cold plate liquid cooling leakage current limiting control method and system for data centers, comprising: S1, collecting coolant operating status data, control status data, and execution feedback data during the liquid cooling process, and preprocessing the collected coolant operating status data, control status data, and execution feedback data to construct a standardized cooling status dataset; S2, based on the standardized cooling status dataset, performing feature analysis on the coolant steady-state deviation characteristics, and dynamically determining the coolant leakage risk and adjusting the coolant supply status based on the feature analysis results; S3, based on the standardized cooling status dataset, evaluating the current limiting adjustment status from the coupling relationship between the valve core opening change rate and the coolant flow response, and dynamically correcting the electromagnetic proportional valve operating status based on the current limiting adjustment status evaluation results; S4, using the coolant steady-state deviation feature analysis results and the current limiting adjustment status evaluation results as input, comprehensively evaluating the liquid supply recovery trend, and adjusting the recovery stage and adjustment path of the cooling circuit in real time based on the comprehensive evaluation results.

[0010] Further, the specific steps for collecting coolant operating status data, control status data, and execution feedback data during the liquid cooling process are as follows: Coolant operating status data is collected using high-precision pressure sensors and flow sensors deployed at key nodes of the two-phase cold plates. This data includes the current static pressure value and instantaneous flow rate value of the coolant. Under stable operating conditions, the average static pressure level of the target monitoring section under steady-state conditions is calculated using an arithmetic average algorithm based on all collected static pressure values, and recorded as the static pressure equilibrium value. The average result of all steady-state flow rates is calculated using a transverse averaging algorithm to obtain the current loop's flow rate level under steady-state conditions, recorded as the flow equilibrium value. Within the same time window, data is collected at each monitoring point. The collected static pressure values ​​are calculated using the standard deviation method to obtain the natural fluctuation level of the static pressure values, which is recorded as the static pressure fluctuation tolerance value. At the same time, based on the set of instantaneous flow values ​​sampled within the time window, the standard deviation is applied to calculate the flow velocity fluctuation amplitude under undisturbed conditions, which is recorded as the flow fluctuation tolerance value. Control status data is collected through a high-speed data acquisition unit integrated into the edge control module. The control status data includes: the leakage velocity value monitored at each sampling time and the initial leakage velocity value when the first leakage is detected, while also recording the number of sampling cycles and the sampling interval time. The system execution feedback data is collected through a linkage execution recording mechanism. The execution feedback data includes: the sampling time interval between two samplings and the duration of the flow limiting adjustment phase.

[0011] Furthermore, the specific steps for preprocessing the collected coolant operating status data, control status data, and execution feedback data to construct a standardized cooling status dataset are as follows: Data cleaning is performed on the collected coolant operating status data, control status data, and execution feedback data. Abnormal fluctuations, missing values, and abrupt changes are removed using sliding window filtering and interpolation completion. After cleaning, steady-state sections are identified for static pressure values ​​and instantaneous flow rates, and stable time windows are extracted. Static pressure equilibrium values ​​and flow equilibrium values ​​are calculated based on the data within the stable time windows. For the control status data, the leakage rate values ​​at each moment are continuously calibrated. The data is verified, and the initial leakage velocity value is combined with the corresponding sampling period and sampling interval for unified labeling and time serialization. For the execution feedback data, the sampling time interval and the duration of the flow-limiting adjustment phase are formatted, and records from different time periods are archived to ensure alignment with the coolant operating status data in the time dimension. The standardized coolant operating status data, control status data, and execution feedback data are then subjected to dimensionless normalization to construct a standardized coolant status dataset. The complete process of synchronously acquiring and preprocessing coolant operating status data, control status data, and execution feedback data is recorded as one sampling period.

[0012] Furthermore, the specific steps for feature analysis of coolant steady-state deviation characteristics based on the standardized cooling state dataset are as follows: Perform a difference operation on the instantaneous flow rate value, subtracting the instantaneous flow rate value of the previous sampling period from the current sampling period's instantaneous flow rate value and dividing by the sampling time interval to obtain the flow rate change rate per unit time, recorded as the flow rate change value; subtract the static pressure equilibrium value from the current static pressure value, divide by the static pressure fluctuation tolerance value, and take the absolute value to obtain the static pressure anomaly value; subtract the flow equilibrium value from the current instantaneous flow rate value, divide by the flow fluctuation tolerance value, and take the absolute value to obtain the flow anomaly value; divide the flow rate change value by the current instantaneous flow rate value and take the absolute value to obtain the flow rate change rate; divide the static pressure fluctuation tolerance value by the static pressure equilibrium value and take the absolute value to obtain the static pressure relative fluctuation value; add the static pressure anomaly value, flow rate anomaly value, flow rate change rate, and static pressure relative fluctuation value to obtain the leakage collaborative judgment value.

[0013] Furthermore, the specific steps for dynamically determining the coolant leakage risk and adjusting the coolant supply based on the feature analysis results are as follows: Real-time comparison of the leakage collaborative judgment value and leakage judgment threshold of the current monitoring path: When the leakage collaborative judgment value is less than or equal to the leakage judgment threshold, the distributor maintains its current flow distribution level, the two-phase cold plates maintain their current heat exchange channel configuration, the condenser maintains its current reflux rhythm, and the monitor takes a single snapshot of the current sampling cycle path status and marks it as a routine record; When the leakage collaborative judgment value is greater than the leakage judgment threshold, the monitor issues a supply restriction command to the distributor, lowers the current branch flow distribution ratio and locks the valve position, increases the reflux speed of the condenser, increases the sampling frequency of the pressure sensor and flow sensor and activates zero-point verification, and applies a duty cycle adjustment mechanism in the branch channel corresponding to the two-phase cold plates: the supply signal is pulse-reconstructed so that the pump output flow only maintains short-term continuous supply at the beginning of each startup, and then quickly enters an intermittent pump stop state, maintaining the pump stop until the pressure sensor detects that the coolant pressure has returned to stability before resuming short-term supply to reduce the input rate, while simultaneously triggering an audible and visual alarm device.

[0014] Furthermore, the specific steps for evaluating the flow-limiting regulation state based on the standardized cooling state dataset and the coupling relationship between the valve core opening change rate and the coolant flow response are as follows: Calculate the difference in leakage velocity values ​​between each adjacent moment and divide it by the sampling time interval; take the absolute value of the ratio of the leakage velocity value difference to the sampling time interval to obtain the instantaneous leakage velocity change rate; divide the leakage velocity value at each sampling moment by the initial leakage velocity value to obtain the relative leakage intensity ratio; multiply the instantaneous leakage velocity change rate by the relative leakage intensity ratio to obtain the coordinated change intensity of the current sampling period; sum the coordinated change intensities of each sampling period and divide by the number of sampling periods to obtain the average coordinated intensity; subtract the duration of the current flow-limiting regulation phase from the average coordinated intensity to obtain the electromagnetic flow-limiting duty cycle.

[0015] Furthermore, the specific steps for dynamically correcting the working state of the electromagnetic proportional valve based on the current limiting adjustment state evaluation results are as follows: The current electromagnetic current limiting duty cycle is acquired in real time, and the following adjustment measures are implemented: When the electromagnetic current limiting duty cycle of the current sampling period shows a decreasing trend compared to the previous sampling period, a lowered pulse signal is continuously sent to the electromagnetic proportional valve to adjust the valve core opening in real time, further limiting the coolant input flow rate; the coolant is introduced into the buffer chamber structure located at the inlet section of the two-phase cold plate to absorb the forward pressure in the main path, and the heat source load information is simultaneously fed back to the upper power supply. The control interface reduces server power consumption. When the electromagnetic current limiting duty cycle of the current sampling period shows an upward trend compared to the previous sampling period, the duty cycle of the electromagnetic proportional valve pulse signal is increased, causing the valve core to gradually open. This triggers the liquid separator to intercept and process the micro-air masses entrained in the restored flow, and simultaneously activates the pressure regulating spring slow-release valve at the rear of the two-phase cold plate. When the electromagnetic current limiting duty cycle of the current sampling period remains unchanged compared to the previous sampling period, the current output signal of the electromagnetic proportional valve remains unchanged, triggering the active self-test program of the differential pressure balancer at the end of the two-phase cold plate.

[0016] Furthermore, the specific steps for comprehensively evaluating the coolant supply recovery trend using the coolant steady-state offset characteristic analysis results and the flow-limiting regulation state evaluation results as input are as follows: Subtract the static pressure equilibrium value from the collected static pressure values ​​at each sampling time and divide by the static pressure fluctuation tolerance value to obtain a standardized pressure offset sequence; subtract the flow equilibrium value from the instantaneous flow rate value and divide by the flow fluctuation tolerance value to obtain a standardized flow rate offset sequence; construct a joint fluctuation vector based on the standardized pressure offset sequence and flow rate offset sequence at each sampling time, and calculate the magnitude of the joint fluctuation vector to obtain the joint offset at the current sampling time. The pressure and flow rate fluctuation amplitude is obtained by applying sliding variance processing to all joint offset amplitudes over the entire time window. The difference between one and the leakage co-judgment value is multiplied by the difference between one and the electromagnetic current limiting duty cycle to obtain the leakage current limiting coupling value. The absolute value of the pressure and flow rate fluctuation amplitude is taken, plus one, and then the logarithm is taken and plus one again to obtain the joint disturbance suppression value. The electromagnetic current limiting duty cycle is multiplied by the corresponding recovery adjustment factor and then plus one to obtain the coupling adjustment amplitude value. The leakage current limiting coupling value is divided by the joint disturbance suppression value and then multiplied by the coupling adjustment amplitude value to obtain the recovery dynamic discrimination value.

[0017] Furthermore, the specific steps for real-time control of the cooling circuit recovery phase and adjustment path based on comprehensive evaluation results are as follows: Real-time comparison of the current recovery dynamic discrimination value with the recovery discrimination threshold, where the recovery discrimination threshold includes a first recovery threshold and a second recovery threshold, with the first recovery threshold being higher than the second recovery threshold; when the recovery dynamic discrimination value is less than or equal to the second recovery threshold, the current flow-limiting state of the electromagnetic proportional valve remains unchanged, prohibiting any form of liquid supply recovery command from being issued, while simultaneously performing high-frequency sampling through the pressure sensor and flow sensor on the path from the condenser to the two-phase cold plate, maintaining the minimum flow input state before leakage is controllable; when the recovery dynamic discrimination value is higher than the second recovery threshold and less than or equal to the first recovery threshold, the path reconstruction observation phase begins: based on the flow sensor and pressure sensor, the trend convergence at each monitoring node is continuously monitored, and when the data collected by the flow sensor and pressure sensor are within two consecutive sampling cycles... When a stable upward trend is observed during the period, the coolant is switched to a branch loop with a shorter length than the current loop via the distributor to help determine whether the recovery is a stable regression rather than a short-term disturbance. When the coolant operating status data collected by the flow sensor and pressure sensor shows a downward trend within two consecutive sampling periods, the preparatory process is exited, and the current path reconstruction observation phase is extended. When the recovery dynamic discrimination value is greater than the first recovery threshold, the coolant supply recovery operation is immediately and automatically initiated: the electromagnetic proportional valve is quickly adjusted to the initial valve state, and the distributor is activated to redistribute the coolant recovered from the condenser to each parallel two-phase cold plate channel. At the same time, the coolant operating status data is recorded by the monitor throughout the entire process of coolant recovery from the flow-limited state to the stable supply state. During the stable period after the coolant supply recovery, the time synchronization mechanism of the flow sensor and pressure sensor is activated to identify potential delayed response segments, and the path response lag time of this round is written into the path diagnosis log.

[0018] The second aspect of this invention provides a two-phase cold plate liquid cooling leakage limiting control system for data centers, comprising: a distributed cold plate status data acquisition module, used to acquire coolant operating status data, control status data, and execution feedback data during the liquid cooling process, and preprocess the acquired coolant operating status data, control status data, and execution feedback data to construct a standardized cooling status dataset; a pressure-flow collaborative judgment and analysis module, used to analyze the coolant steady-state deviation characteristics based on the standardized cooling status dataset, and dynamically determine the coolant leakage risk and adjust the coolant supply status based on the analysis results; a stepped feedback limiting adjustment module, used to evaluate the limiting adjustment status based on the coupling relationship between the valve core opening change rate and the coolant flow response based on the standardized cooling status dataset, and dynamically correct the electromagnetic proportional valve operating status based on the evaluation results; and a leakage recovery dynamic discrimination module, used to comprehensively evaluate the liquid supply recovery trend based on the coolant steady-state deviation characteristic analysis results and the limiting adjustment status evaluation results, and to adjust the recovery stage and adjustment path of the cooling circuit in real time based on the evaluation results.

[0019] Beneficial effects

[0020] The present invention has the following beneficial effects:

[0021] (1) The two-phase cold plate liquid cooling leakage current limiting control method and system for data centers avoids false triggering judgment due to instantaneous disturbance by using the steady-state equilibrium value of pressure and flow characteristics and natural fluctuation tolerance, effectively improving the robustness and reliability of leakage identification under complex working conditions.

[0022] (2) The two-phase cold plate liquid cooling leakage current limiting control method and system for data centers introduces a dynamic control strategy of electromagnetic current limiting duty cycle and recovery dynamic discrimination value to build a closed-loop rhythm control mechanism, which can quickly limit the current during the leakage evolution stage and dynamically restore the liquid supply according to the system stability, so as to ensure the continuous and stable operation of the system.

[0023] (3) The two-phase cold plate liquid cooling leakage current limiting control method and system for data centers enhances the real-time perception of coolant transmission status changes in complex liquid circuit structures by integrating multi-path sensor data and differential calculation algorithms, thereby improving the positioning accuracy and judgment stability in long-path cooling loops.

[0024] (4) The two-phase cold plate liquid cooling leakage current limiting control method and system for data centers reduces the dependence on the central processing unit by integrating edge acquisition and local discrimination processes. It has a fast response capability with high timeliness and low latency and is suitable for adaptive leakage control requirements in large-scale distributed liquid cooling deployment scenarios.

[0025] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0026] Figure 1 is a flowchart of the two-phase cold plate liquid cooling leakage current limiting control method for data centers according to the present invention.

[0027] Figure 2 is a structural diagram of the two-phase cold plate liquid cooling leakage current limiting control system for data centers according to the present invention.

[0028] Figure 3 is a line graph of the recovery dynamic discriminant value involved in this invention;

[0029] Figure 4 is a schematic diagram of the working principle involved in this invention;

[0030] Figure 5 is a control flowchart related to the present invention;

[0031] Figure 6 is a diagram of the liquid cooling system involved in this invention;

[0032] In the diagram, 1 is the monitor; 2 is the flow sensor; 3 is the pressure sensor; 4 is the distributor; 5 is the two-phase cold plate; 6 is the server; 7 is the condenser; and 8 is the circulating pump. Detailed Implementation

[0033] 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.

[0034] Please refer to Figures 1-6. This invention provides a technical solution: a two-phase cold plate liquid cooling leakage current limiting control method for data centers. Figure 1 shows a flowchart of the two-phase cold plate liquid cooling leakage current limiting control method for data centers provided in this application example. The method includes: S1, collecting coolant operating status data, control status data, and execution feedback data during the liquid cooling process, and preprocessing the collected coolant operating status data, control status data, and execution feedback data to construct a standardized cooling status dataset; S2, analyzing the coolant steady-state deviation characteristics based on the standardized cooling status dataset, and dynamically determining the coolant leakage risk and adjusting the coolant supply status based on the analysis results; S3, evaluating the current limiting adjustment status based on the coupling relationship between the valve core opening change rate and the coolant flow response, and dynamically correcting the electromagnetic proportional valve's operating status based on the evaluation results; S4, comprehensively evaluating the coolant supply recovery trend using the coolant steady-state deviation characteristic analysis results and the current limiting adjustment status evaluation results as input, and adjusting the recovery stage and adjustment path of the cooling circuit in real time based on the evaluation results.

[0035] Specifically, the steps for collecting coolant operating status data, control status data, and execution feedback data during the liquid cooling process are as follows: High-precision pressure sensors 3 and flow sensors 2, deployed at key nodes of the two-phase cold plate 5, continuously collect multi-channel coolant operating status data to ensure that the dynamic behavior of each loop node is captured in real time. The coolant operating status data includes: the static pressure value of the coolant under static conditions and the instantaneous flow rate value during flow; both together constitute the basic parameter set for status identification and anomaly detection. Under stable operating conditions, the edge control module extracts historical sampling sequences through a sliding window and calculates the average static pressure level of the target monitoring section under the current steady-state conditions by applying an arithmetic mean algorithm to all static pressure values ​​within the window; this is recorded as the static pressure equilibrium value. Simultaneously, a lateral averaging algorithm is used to aggregate the flow rate sampling points within the stable sections of each cooling loop to obtain the current loop's flow rate level under steady-state conditions, recorded as the flow equilibrium value.

[0036] To further characterize the natural fluctuations of parameters under no significant disturbance, within the same time window, the standard deviation calculation method is applied to the static pressure values ​​collected at each monitoring point to extract their fluctuation intensity, thereby obtaining the natural fluctuation level of the static pressure value, which is recorded as the static pressure fluctuation tolerance value. Simultaneously, based on the set of instantaneous flow values ​​sampled within the time window, the standard deviation algorithm is used to evaluate the magnitude of its change under steady-state conditions, thereby obtaining the fluctuation intensity of the flow velocity under undisturbed conditions, which is recorded as the flow fluctuation tolerance value.

[0037] In addition, control status data is recorded in real time by a high-speed data acquisition unit integrated into the edge control module. This control status data includes the leakage velocity value monitored at each sampling moment, as well as the reference velocity value when the first leakage anomaly is detected. It also records the number of sampling cycles and the sampling time interval between each cycle. Through a linked execution recording mechanism, the system synchronously acquires execution feedback data, which includes the sampling time interval between two samples and the duration of the current current limiting adjustment phase since its triggering.

[0038] This implementation plan establishes a standardized operational status data foundation for two-phase cold plate liquid cooling systems, providing reliable initial support for the calculation of core parameters such as leakage collaborative judgment values, recovery dynamic discrimination values, and electromagnetic current limiting duty cycle. By performing high-precision acquisition and multi-level statistical processing of static pressure and instantaneous flow data at key nodes, the system can dynamically characterize the operational equilibrium level and natural fluctuation tolerance of the coolant under steady-state conditions, enhancing its sensitivity and identification capability to abnormal deviation trends. Simultaneously, the linkage acquisition mechanism of control status and execution feedback data provides periodic and structured data support for the current limiting control logic and coolant supply adjustment process, effectively supporting the edge control module to achieve highly reliable leakage detection, response judgment, and adjustment control in operating environments with complex cooling paths and unstable flow states.

[0039] Specifically, the specific steps for preprocessing the collected coolant operating state data, control state data, and execution feedback data to construct a standardized cooling state data set are as follows: When performing data cleaning operations on the collected coolant operating state data, control state data, and execution feedback data, first perform denoising processing on the static pressure value and instantaneous flow value sequences through a sliding window filter,剔除 outliers caused by instantaneous sensor errors and environmental disturbances, and complete the filling for data missing areas during the sampling process using linear interpolation and adjacent value interpolation methods to ensure the integrity and continuity of the original data sequence. After completing the basic cleaning, further identify the steady-state sections of the static pressure value and instantaneous flow value sequences,联合判断稳定时间窗口,并以该时间窗口内数据为基础分别计算得到目标监测通道的静态压力均衡值与流量均衡值,确保参考参数具有环境适应性与时间稳定性。

[0040] For the control state data, perform sequence coherence tests on the leakage speed values monitored at each moment based on the time stamp,剔除异常跳变和短周期重复值,保证其在时间维度上的一致性。同时将首次检测到泄漏的速度值与对应采样周期编号和间隔时间进行统一标记,生成结构化首发泄漏事件序列,用于后续状态判断参考。对于执行反馈数据,将采样时间间隔与限流调节阶段的已维持时间分别格式化为标准化字段,并按照采样时间顺序构建时序记录结构,确保反馈数据与运行状态数据在时间轴上的精准对齐与联动分析。

[0041] Finally, perform dimensionless normalization operations on the standardized coolant operating state data, control state data, and execution feedback data uniformly,使各类物理量具有统一的数值尺度和可比性,同时将完成一次冷却液运行状态数据、控制状态数据与执行反馈数据同步采集与预处理的完整过程,记为一个采样周期,最终构建用于泄漏识别与限流调节判定的标准化冷却状态数据集,为后续各类判断值计算与状态响应提供高一致性、高可靠度的数据支撑。

[0042] It should be noted that there are some parts in the original text that seem to be incomplete or have incorrect expressions, which may affect the accuracy of the translation. For example, in the part of , "联合判断稳定时间窗口,并以该时间窗口内数据为基础分别计算得到目标监测通道的静态压力均衡值与流量均衡值,确保参考参数具有环境适应性与时间稳定性。" is not very clear in the original Chinese and may need further clarification in the source text for a more accurate translation.In this implementation plan, by systematically cleaning, standardizing, and normalizing coolant operating status data, control status data, and execution feedback data, the analytical basis for various original sampled values ​​in terms of temporal integrity, numerical consistency, and physical comparability is ensured. Its core function is to improve data quality and parameter stability, effectively eliminate unstructured interference caused by pipeline disturbances, signal jitter, and sampling gaps, and simultaneously achieve precise alignment of data from different sources on the time axis. This provides reliable support for calculating key criteria for subsequent leak collaborative judgment values ​​and recovery dynamic discrimination values, and ensures that the response mechanisms for flow limiting adjustment and coolant supply recovery possess high judgment accuracy and dynamic adaptability.

[0043] Specifically, based on a standardized cooling state dataset, the analysis of coolant steady-state deviation characteristics involves the following steps: performing a difference operation on the instantaneous flow rate value; subtracting the instantaneous flow rate value of the previous cycle from the current cycle's instantaneous flow rate value and dividing by the sampling time interval to obtain the flow rate change rate per unit time, recorded as the flow rate change value; subtracting the static pressure equilibrium value from the current static pressure value, dividing by the static pressure fluctuation tolerance value, and taking the absolute value to obtain the static pressure anomaly value; subtracting the flow equilibrium value from the current instantaneous flow rate value, dividing by the flow fluctuation tolerance value, and taking the absolute value to obtain the flow anomaly value; dividing the flow rate change value by the current instantaneous flow rate value and taking the absolute value to obtain the flow rate change rate; dividing the static pressure fluctuation tolerance value by the static pressure equilibrium value and taking the absolute value to obtain the static pressure relative fluctuation value; and adding the static pressure anomaly value, flow rate anomaly value, flow rate change rate, and static pressure relative fluctuation value to obtain the leakage collaborative judgment value.

[0044] The formula for calculating the leakage synergy judgment value is:

[0045] ;

[0046] In the formula:

[0047] P represents the current static pressure value, which reflects the steady-state stress of the coolant at the monitoring point. It is a basic indicator for judging the operating stability of the cold plate and whether there is a risk of liquid leakage. It is derived from the real-time data collected by pressure sensor 3. It represents the static pressure equilibrium value under stable operating conditions, used to describe the average static pressure level when multiple sensing points in the target monitoring section are currently in a stable operating state, and is the benchmark parameter for judging the pressure deviation of the current monitoring point. It represents the static pressure fluctuation tolerance value under stable operating conditions, used to characterize the natural variation range of pressure fluctuation amplitude at each monitoring point in adjacent stable areas, and is a normalized reference factor for determining the degree of single-point pressure anomaly. This represents the current instantaneous flow rate value, which reflects the real-time delivery capacity of coolant through the monitoring point. It is a key indicator for determining whether there is abnormal liquid loss or transmission blockage, and it is derived from the real-time sampling data of flow sensor 2. It represents the flow balance value within the steady-state infusion section, used to indicate the normal flow rate level of coolant flowing through this circuit under the current working conditions, and is the basic quantity for judging instantaneous flow deviation; It represents the flow fluctuation tolerance value under steady-state conditions, which is used to reflect the natural fluctuation range of the cold plate loop flow under the condition of no significant disturbance within the current control cycle. It is an important reference value for judging the significance of flow deviation. It represents the change in flow rate per unit time, used to characterize the current rate of change in flow rate, and is a leading indicator for identifying the trend of increased flow velocity during a sudden leakage process. It is derived from the differential results of continuously sampled flow data.

[0048] In this implementation plan, under the two-phase cold plate liquid cooling environment of a data center, multiple key monitoring parameters are integrated to quantitatively determine the degree of coolant leakage risk. By combining the current static pressure value, instantaneous flow rate value, flow rate change rate per unit time, and corresponding steady-state equilibrium reference value, the system comprehensively assesses whether the current operating conditions deviate from normal steady-state characteristics under the influence of multi-source disturbances, and accurately captures any abnormal transmission states that may exist during coolant delivery. This calculation process improves the sensitivity and anti-interference capability of leakage detection, avoids misjudgment and delayed response due to a single outlier, and helps to trigger subsequent flow restriction adjustment and coolant supply repair measures at the initial stage of leakage.

[0049] Specifically, the steps for dynamically determining the risk of coolant leakage and adjusting the coolant supply based on feature analysis results are as follows: Real-time comparison of the leakage coordination judgment value and leakage judgment threshold of the current monitoring path: When the leakage coordination judgment value is less than or equal to the leakage judgment threshold, it is determined that there is no obvious abnormality in the current path, the established operating mode is maintained, and each component enters a steady-state state. At this time, the distributor 4 maintains its existing flow level and the original coolant distribution ratio remains unchanged; the two-phase cold plate 5 continues to supply coolant according to the current heat exchange channel configuration without triggering channel switching operations; the condenser 7 continues its current reflux rhythm without increasing reflux capacity; the monitor 1 takes a complete snapshot of the coolant static pressure level, instantaneous flow rate, and valve execution status within the current sampling period, and marks the monitoring data of this round as a regular record stored in the buffer as a benchmark reference for subsequent data comparison and trend extraction.

[0050] When the leakage assessment value exceeds the leakage assessment threshold, a potential leakage risk is identified in the current path, and the abnormal channel pressure drop and supply restriction process is immediately initiated. Monitor 1 immediately issues a supply restriction command to distributor 4, forcibly reducing the current branch's diversion ratio to decrease the liquid flow in that branch, while simultaneously locking the diversion valve position to prevent accidental adjustment. Condenser 7 then switches from its normal rhythm to enhanced reflux mode, accelerating the reflux rate by increasing the heat exchange intensity of the condenser surface to release the accumulated pressure within the loop in a short time. Pressure sensor 3 and flow sensor 2 simultaneously switch to high-frequency sampling mode, entering a high-precision status sensing process, and activate a zero-point verification mechanism to dynamically calibrate the collected baseline to eliminate false alarm interference. At the target branch where the two-phase cold plate 5 is located... In the circuit channel, the duty cycle adjustment mechanism is activated, reconstructing the original continuous liquid supply signal into a pulse control signal with intermittent rhythm. This ensures that the pump's output flow only maintains a continuous liquid supply state for a short period at the beginning of each startup, before quickly transitioning to the pump stop phase. The pump stop state will continue until the pressure sensor 3 continuously detects that the circuit pressure has stabilized and returned to the safe operating range, and the instantaneous flow rate fluctuation value returned by the flow sensor 2 is lower than the risk threshold. Then, the next short-term liquid supply process is restarted. Throughout the entire control process, the audible and visual alarm device is automatically activated to notify the on-duty personnel that the current circuit is in the controlled intervention phase.

[0051] In this implementation plan, the leakage collaborative judgment value is compared with the leakage judgment threshold in real time to construct a dynamic control process for anomaly identification and response linkage. Its function is to achieve early detection and rapid intervention of potential leakage events in the cooling circuit. When the judgment value is within the normal range, the operating configuration of each component remains unchanged to ensure data continuity and path stability. When the judgment value exceeds the threshold, a series of emergency response operations are triggered, including supply limitation, accelerated reflux, high-frequency sampling, and pulsed liquid supply. This proactively reduces the operating pressure and liquid supply rate of high-risk branches without human intervention, while maintaining the normal heat exchange function of other paths, effectively controlling the impact range of local leaks and preventing overall instability. This mechanism significantly improves the accuracy of leak identification and the timeliness of response in a two-phase cold plate liquid cooling environment with multiple parallel branches and frequent dynamic load changes, and is a core control link to ensure the safety, stability, and high availability of the liquid cooling system.

[0052] Specifically, based on a standardized cooling state dataset, the evaluation of the flow-limiting regulation state from the coupling relationship between the valve core opening change rate and the coolant flow response involves the following steps: Calculate the difference in leakage velocity values ​​between each adjacent moment and divide it by the sampling time interval; take the absolute value of the ratio of the leakage velocity value difference to the sampling time interval to obtain the instantaneous leakage velocity change rate; divide the leakage velocity value at each sampling moment by the initial leakage velocity value to obtain the relative leakage intensity ratio; multiply the instantaneous leakage velocity change rate by the relative leakage intensity ratio to obtain the coordinated change intensity of the current cycle; sum the coordinated change intensities of each sampling cycle and divide by the number of sampling cycles to obtain the average coordinated intensity; subtract the duration of the current flow-limiting regulation phase from the average coordinated intensity to obtain the electromagnetic flow-limiting duty cycle.

[0053] The formula for calculating the electromagnetic current limiting duty cycle is:

[0054]

[0055] This represents the number of sampling periods, used to count the number of samples participating in the average calculation within the current control period. It is an important parameter for smoothly judging the trend of leakage rate changes and is derived from the window length configuration of the sampling control logic in the edge control module. This represents the time index corresponding to the sampling moment in the current time series. It is used to identify the latest sampling position currently participating in the calculation. The value is derived from the sampling sequence number formed by numbering in chronological order during continuous sampling. It is used to determine the termination position of the sliding statistical window and, together with the number of sampling periods n, limits the continuous sampling segment covered by the summation interval [t-n+1,t]. This represents the leakage velocity value monitored at the i-th sampling time. It is used to characterize the coolant loss rate per unit time at the current monitoring point and is an important basis for determining the adjustment intensity in dynamic control. It is derived from the differential calculation result of the continuous measurement value of the flow sensor. It represents the sampling time interval between two consecutive samples, used to calculate the instantaneous rate of change of leakage velocity, and is the basic time unit for feedback control rhythm; This represents the initial leakage velocity value when a leak is first detected. It is used as a normalization reference to reflect the degree of change in the current leakage velocity relative to the initial state. It is the basic indicator for threshold judgment in each adjustment stage and is derived from the flow rate capture value before triggering the first adjustment. This indicates the duration of the current flow restriction adjustment phase. It is used to assess the time range in which the current adjustment measures have been implemented and is an important parameter for time control in a tiered adjustment strategy. It is derived from the cumulative duration recorded since the start of the current phase.

[0056] In this implementation plan, the severity of flow rate changes during the current coolant leakage process is dynamically assessed as a key control criterion for determining whether the flow-limiting adjustment phase needs to be continued, strengthened, or terminated. By weighted averaging the ratio of the instantaneous rate of change of leakage velocity to the initial leakage velocity over multiple consecutive sampling periods, and combining this with the duration of the current adjustment period, the plan accurately characterizes the magnitude and stability trend of the current leakage evolution. This improves the adaptability of the response adjustment in the later stages of leakage events under complex cooling paths, avoids premature termination or blind maintenance of the flow-limiting response due to misjudgment, and thus achieves fine-grained adjustment of the flow rate control rhythm, enhancing the steady-state repair capability under leakage disturbances.

[0057] Specifically, the steps for dynamically correcting the working state of the electromagnetic proportional valve based on the evaluation results of the current limiting regulation state are as follows: The current electromagnetic current limiting duty cycle is acquired in real time, and the following adjustment measures are implemented: When the electromagnetic current limiting duty cycle of the current sampling period shows a downward trend compared to the previous sampling period, a pulse control signal with a reduced duty cycle is continuously sent to the electromagnetic proportional valve to adjust the displacement of the valve core in real time, causing the cross-section of the coolant input channel to gradually shrink, thereby further limiting the instantaneous input rate of coolant in the main channel; simultaneously, a portion of the coolant is temporarily introduced into the buffer chamber structure located at the front end of the inlet section of the two-phase cold plate 5, allowing it to absorb the forward hydraulic fluctuations in the main path, suppressing the disturbance of the front-end impact of the coolant on the heat exchange stability of the cold plate, and simultaneously transmitting the heat source load information of the current server 6 operating state back to the upper power control interface, thereby reducing the power output level of the server 6 in conjunction with the reduction of the heat load growth rate at the heat exchange end.

[0058] When the electromagnetic current limiting duty cycle of the current sampling period shows an upward trend compared to the previous sampling period, the edge control module increases the duty cycle of the output signal, driving the valve core in the electromagnetic proportional valve to slide in the opening direction, gradually increasing the effective diameter of the coolant input channel and realizing flow recovery; at the same time, the liquid distributor 4 is linked to operate in a preset interception mode, intercepting small gas masses and unstable impurities mixed in the coolant during the process of increasing the flow rate in the coolant, so as to reduce the interference on the stable heat exchange capacity of the subsequent channel; in addition, the pressure regulating spring slow release valve located at the outlet of the rear section of the two-phase cold plate 5 is activated simultaneously, so that the structure releases part of the retained gas-liquid mixture when the back pressure of the channel increases, mitigating the heat exchange lag and abnormal fluctuations caused by the instantaneous high back pressure.

[0059] When the electromagnetic current limiting duty cycle of the current sampling period remains unchanged compared to the previous sampling period, it is determined that the current path state is in the stable operating range. The current output signal of the electromagnetic proportional valve remains unchanged to ensure that the coolant flow rate matches the main circuit demand. At the same time, the differential pressure balancer installed at the end of the two-phase cold plate 5 is triggered to perform an active self-check program to monitor for any slight deviation trend and to assess whether the path is still in the high-efficiency heat exchange range, providing auxiliary criteria for subsequent adjustment.

[0060] In this implementation scheme, based on the dynamic change trend of the electromagnetic current-limiting duty cycle, the system actively adjusts the coolant input flow rate and path response behavior to achieve refined current-limiting control of the two-phase cold plate liquid cooling path. When the duty cycle decreases, the system reduces the valve core opening, guides the coolant into the buffer chamber, and reduces the server power supply, thereby weakening the supply shock and heat source load and preventing leakage expansion and heat exchange imbalance. When the duty cycle increases, the system increases the valve core diameter, intercepts the backflow air mass, and simultaneously releases the cold plate outlet back pressure, effectively ensuring the stability and safety of the supply recovery process. When the duty cycle remains constant, the system triggers the self-checking mechanism of the terminal differential pressure balancer to verify whether the stable state of the path continues, providing a precise criterion for subsequent adjustments. Overall, this step achieves dynamic feedback adjustment under current-limiting conditions, enhancing the adaptive control capability in abnormal scenarios such as coolant fluctuations and micro-leakage evolution.

[0061] Specifically, using the coolant steady-state offset characteristic analysis results and the flow-limiting regulation state assessment results as inputs, the comprehensive evaluation of the coolant supply recovery trend is carried out through the following steps: Subtract the static pressure equilibrium value from the collected static pressure values ​​at each sampling time and divide by the static pressure fluctuation tolerance value to obtain a standardized pressure offset sequence; subtract the flow equilibrium value from the instantaneous flow rate value and divide by the flow fluctuation tolerance value to obtain a standardized flow rate offset sequence; construct a joint fluctuation vector based on the standardized pressure offset sequence and flow rate offset sequence at each sampling time, and calculate the magnitude of the joint fluctuation vector to obtain the joint offset at the current sampling time. The intensity is calculated by applying sliding variance to all joint offset intensities over the entire time window to obtain the joint pressure-flow fluctuation amplitude in the current period. The difference between one and the leakage coordination judgment value is multiplied by the difference between one and the electromagnetic current limiting duty cycle to obtain the leakage current limiting coupling value. The absolute value of the joint pressure-flow fluctuation amplitude is taken, plus one, and then the logarithm is taken and plus one again to obtain the joint disturbance suppression value. The electromagnetic current limiting duty cycle is multiplied by the corresponding recovery adjustment factor and then plus one to obtain the coupling adjustment amplitude value. The leakage current limiting coupling value is divided by the joint disturbance suppression value and then multiplied by the coupling adjustment amplitude value to obtain the recovery dynamic discrimination value.

[0062] The formula for calculating the dynamic discriminant value is as follows:

[0063] ;

[0064] The leakage synergy judgment value reflects the overall leakage risk level at the current monitoring point and serves as a discriminant factor for continuously assessing whether a leakage trend still exists. It indicates the electromagnetic current limiting duty cycle, which reflects the current control state of the electromagnetic proportional valve and is an execution signal parameter that identifies the current degree of liquid supply suppression. It represents the combined fluctuation amplitude of pressure and flow rate, used to quantify the severity of changes in the cooling state of the monitoring point, and is a basic physical characteristic value for determining whether it has recovered to a steady state. It is derived from the weighted sum of the pressure offset and flow rate offset within the current sampling period. The recovery adjustment factor represents the amplification effect of the electromagnetic current limiting control ratio on the overall recovery trend in the calculation of the dynamic discrimination value of recovery. Its value ranges from 1 to 3, and the specific value is dynamically updated based on the intensity of fluctuations in the current coolant operating state, the rhythm of the fluctuation response, and the similarity changes in disturbance intervention characteristics. In the specific calculation, firstly, the electromagnetic execution signal sequence and instantaneous flow rate change sequence of the cooling channel within the current sliding time window are extracted, and their change slope, rhythm fluctuation frequency density, and peak amplitude of the current limiting action are calculated to characterize the intensity and frequency of dynamic disturbances in the current recovery adjustment process. Then, the current limiting callback characteristics of the same channel are extracted from historical stable adjustment states to construct a benchmark control template, including the electromagnetic execution amplitude sequence under typical recovery paths, the rhythm of adjustment fluctuations, and the duration and intensity distribution of the suppression segment. Based on this, the current channel's control fluctuation characteristics are aligned with the historical template using multi-dimensional features, calculating the recovery rhythm similarity, current limiting amplitude deviation rate, and adjustment fluctuation interference overlap, and combining this with... The dynamic stability score and disturbance callback tolerance range of the current channel in different data center deployment environments are used to weight and aggregate all alignment results to generate a recovery adjustment factor. This factor reflects the matching degree of the current recovery rhythm in disturbance-sensitive scenarios and the adjustment intensity required for flow limiting intervention. When the current adjustment state of the cooling channel fluctuates drastically, the response rhythm is unstable, and the electromagnetic control execution amplitude and recovery trend deviate significantly from the historical stable callback pattern, the value of the recovery adjustment factor increases. This enhances the amplification of the weight of the flow limiting strategy in the dynamic judgment value of recovery, prompting a stronger limiting response to the flow rate control rhythm under severe disturbance conditions. Conversely, when the flow rate recovery trend is gentle, the electromagnetic execution amplitude is stable, and the overall recovery path is highly consistent with the historical steady-state callback characteristics, the value of the recovery adjustment factor decreases. This moderately weakens the constraint of the flow limiting segment execution suppression on the overall recovery trend, avoiding system rigid jitter caused by overreaction, thereby improving the adaptive callback capability and rhythm coordination accuracy of the dynamic judgment value of recovery in multi-fluctuation linkage scenarios.

[0065] In this implementation example, the leakage coordination judgment value for Example 1 is set to 0.25, the electromagnetic current limiting duty cycle is 0.30, the combined pressure and flow rate fluctuation amplitude is 0.12, and the recovery adjustment factor is 1.5; the leakage coordination judgment value for Example 2 is set to 0.40, the electromagnetic current limiting duty cycle is 0.55, the combined pressure and flow rate fluctuation amplitude is 0.08, and the recovery adjustment factor is 2.0; the leakage coordination judgment value for Example 3 is set to 0.10, the electromagnetic current limiting duty cycle is 0.20, the combined pressure and flow rate fluctuation amplitude is 0.20, and the recovery adjustment factor is 1.8; the leakage coordination judgment value for Example 4 is set to 0.60, the electromagnetic current limiting duty cycle is... The leakage synergy judgment value for Example 5 was set to 0.15, the electromagnetic current limiting duty cycle to 0.70, the combined pressure and flow fluctuation amplitude to 0.30, and the recovery adjustment factor to 2.5. For Example 6, the leakage synergy judgment value was set to 0.35, the electromagnetic current limiting duty cycle to 0.25, the combined pressure and flow fluctuation amplitude to 0.10, and the recovery adjustment factor to 1.0. For Example 7, the leakage synergy judgment value was set to 0.05, the electromagnetic current limiting duty cycle to 0.15, the combined pressure and flow fluctuation amplitude to 0.25, and the recovery adjustment factor to 2.8. The recovery dynamic discrimination value for each example was calculated, as shown in Table 1.

[0066] Table 1. Recovery Dynamic Discriminant Value Data Table

[0067]

[0068] Figure 3 shows a line graph of the recovery dynamic discrimination value data provided in this application. As can be seen from Table 1 and Figure 3, Instance 3 has the highest recovery dynamic discrimination value. Its leakage coordination judgment value and electromagnetic current limiting duty cycle are both at a low level, but the combined fluctuation amplitude of pressure and flow is relatively high, and the recovery adjustment factor is in the medium-high range. This indicates that this instance has strong stable recovery potential and control feedback capability in the current monitoring cycle, and is suitable as a priority adjustment target for quickly suppressing abnormal disturbances and realizing cooling state reconstruction. Instance 4 has the lowest recovery dynamic discrimination value. Its leakage coordination judgment value is high, and its electromagnetic current limiting duty cycle is medium, but the fluctuation amplitude is extremely small, and the recovery adjustment factor is low. This reflects its weak recovery trend and lagging control response, making it unsuitable as the first choice for active adjustment. It can be retained as a boundary observation parameter to assist in the overall stability assessment. The line graph of recovery dynamic discrimination value clearly shows the differences in recovery control capability of each instance during abnormal leakage. The higher the evaluation value, the easier it is to achieve efficient recovery and state reconstruction, and it should be prioritized in the fine control process to ensure the steady-state operation of the system.

[0069] Specifically, the steps for real-time adjustment of the cooling circuit's recovery phase and adjustment path based on comprehensive evaluation results are as follows: Real-time comparison of the current recovery dynamic discrimination value with the recovery discrimination threshold, where the recovery discrimination threshold includes a first recovery threshold and a second recovery threshold, and the first recovery threshold is higher than the second recovery threshold.

[0070] When the recovery dynamic discrimination value is less than or equal to the second recovery threshold, the current flow limiting state of the electromagnetic proportional valve remains unchanged, and any form of liquid supply recovery command is prohibited from being issued, maintaining the flow suppression state of the passage under the lowest power consumption control; at this time, the pressure sensor 3 and flow sensor 2 on the path from condenser 7 to two-phase cold plate 5 switch to high-frequency sampling mode, keep responding to and capturing small changes, and upload the sampled data to the edge control node in real time. At the same time, all additional heat dissipation auxiliary paths of heat load are closed to ensure that the operation is in the minimum input and maximum suppression standby stage;

[0071] When the recovery dynamic discrimination value is higher than the second recovery threshold and less than or equal to the first recovery threshold, the path reconstruction observation stage begins: instead of directly executing the recovery action, the observation state is entered. Based on the flow sensor 2 and pressure sensor 3, the current coolant operating status data is continuously collected, and the trend convergence at each monitoring node is analyzed. When the flow and pressure data of the key path nodes show a gradual upward trend in two consecutive sampling cycles, it indicates that there may be signs of stable recovery. At this time, the coolant flow direction is switched to a shorter loop with a more sensitive path response through the distributor 4 to help determine whether it is in a natural recovery process. If the sampled data is found to decrease twice in a row during the observation period, it is determined to be an unsteady disturbance. The preparatory process is exited and the current path reconstruction observation stage time is appropriately extended. At the same time, the recovery failure characteristics of the attempts are recorded for subsequent pattern recognition.

[0072] When the dynamic recovery threshold value is greater than the first recovery threshold, it is determined that the liquid supply restart condition has been met, and an automatic liquid supply recovery operation is immediately initiated. The electromagnetic proportional valve is quickly adjusted to the initial valve opening state, the flow restriction mode is released, and the distributor 4 is activated to redistribute the recovered coolant in the condenser 7 to the heat exchange channels of multiple parallel two-phase cold plates 5. The monitor 1 records the data stream from the flow restriction state to the stable liquid supply state throughout the process, forming a complete coolant operation behavior sequence. On this basis, the system enters the liquid supply recovery stabilization period, and the time calibration mechanism of the flow sensor 2 and the pressure sensor 3 is activated to compare the signal response delay differences, identify potential slow response segments and evaluate the path fluctuation delay effect. The response lag time of each branch is written into the path diagnosis log to provide a basis for subsequent optimization of path structure and control parameters.

[0073] Figure 4 shows the working principle diagram of the present invention, illustrating the core structure and signal linkage path of a typical two-phase liquid cooling system in a fault response scenario. The cooling pump, located at the beginning of the entire circulation chain, is responsible for driving the coolant into the two-phase cold plate 5 to absorb heat from the target being cooled. During heat exchange in the two-phase cold plate 5, some of the coolant vaporizes, producing a gas-liquid mixture. This mixture flows through a gas-liquid separator, separating the gas and liquid phases. The liquid portion is then guided into the condenser 7 for condensation and recovery, while the gaseous portion can be reliquefied through the condensation loop. The condenser 7 reduces the refrigerant temperature, improving cooling efficiency, and reintroduces the treated liquid coolant into the cooling pump, completing a closed-loop cycle. A monitoring system is installed at the rear of the gas-liquid separator to collect operating parameters such as coolant pressure, flow rate, and temperature, and to calculate key status parameters such as leakage assessment values ​​and electromagnetic current limiting duty cycle in real time. Once an abnormal operating state is detected, an alarm / automatic shutdown module is triggered via a linkage signal, issuing audible and visual alarm information to prevent the fault from escalating and equipment damage. The overall structure demonstrates a complete liquid cooling loop from cooling drive—heat exchange—gas-liquid separation—condensation recovery—monitoring and early warning, as well as the signal interaction mechanism between key nodes. It particularly emphasizes the linkage response chain that is automatically triggered in case of abnormality, which is an important basic architecture for achieving safe, stable and efficient operation of the liquid cooling system.

[0074] Figure 5 shows the control flow diagram involved in this invention, illustrating the fault alarm information processing control process based on pressure monitoring during the cooling process of the two-phase cold plate 5, presenting the entire processing chain from physical acquisition to remote linkage. The top two-phase cold plate 5 is the core heat exchange component of the cooling structure, used to absorb heat from the cooled device. Simultaneously, due to dynamic fluctuations in the heat exchange process, its internal pressure changes become a key indicator for judging whether the operating status is abnormal. This pressure information is sensed in real time by the pressure sensor 3, generating a continuous raw pressure signal. The pressure signal is then input to the data acquisition module, which preprocesses the signal, constructs a standardized data frame, and transmits it to the computing unit. Based on the built-in recovery dynamic discrimination formula and leakage collaborative identification mechanism, the computing unit determines whether there is an abnormal state in the current cooling path, including sudden changes in flow resistance, leakage trends, and backflow instability. When the computing unit identifies an anomaly and reaches the warning level, it immediately activates the alarm execution mechanism, triggering an audible and visual alarm and linking self-protection actions including power failure, automatic pump shutdown, and automatic channel switching. Simultaneously, all structured data and alarm events are synchronously uploaded to the cloud monitoring platform, enabling remote access, centralized management, and historical record retrieval. The entire process demonstrates a complete closed-loop logic from physical sensing to signal processing to discrimination and calculation to local execution to cloud synchronization, which enhances the sensitivity of liquid cooling equipment to minor fluctuations and sudden anomalies and the efficiency of alarm linkage in actual operation.

[0075] Figure 6 shows the composition of the liquid cooling system involved in this invention, constructing a closed-loop operation path with coolant circulation as the main line and heat conduction and fault monitoring linkage control as the core. In the system, the circulation pump 8 is responsible for driving the coolant circulation, providing power for the entire cooling loop. After completing the gas-liquid phase change heat process through the condenser 7, the liquid is distributed as needed by the distributor 4 to multiple parallel two-phase cold plate channels 5. After absorbing the heat generated by the server core components in the cold plates, it flows back to the condenser 7 to achieve continuous heat exchange. During this process, the pressure sensor 3 and the flow sensor 2 are respectively deployed at key liquid flow nodes to monitor the changes in coolant flow and pressure fluctuations in real time, providing basic data for judging leakage trends, abnormal pressure drops, and gas stagnation. The monitor 1, as the central control unit, summarizes the data from each monitoring point in real time, performs leak collaborative analysis, calculates dynamic judgment values ​​for recovery, and makes flow limiting control logic judgments. It also dynamically issues diversion adjustment and branch locking commands to the distributor to achieve precise intervention in suspected leakage paths. When the operating status is stable, monitor 1 coordinates with distributor 4 to resume valve operation and activates the reflux compensation process of condenser 7 to ensure a balance between liquid supply safety and cold plate heat exchange efficiency. This structure is suitable for high-density server deployment scenarios, and through a sensing-discrimination-adjustment linkage mechanism, it achieves rapid identification and response control of abnormal states in the liquid cooling path.

[0076] This implementation scheme uses multi-level recovery thresholds to dynamically adjust the coolant supply strategy, ensuring a scientific assessment of the current path's stable recovery trend after a leak in the two-phase liquid cooling system. This avoids secondary disturbances and cooling runaway caused by blind recovery. When the dynamic recovery threshold is below the second recovery threshold, the system is in a suppressed state, limiting coolant supply to the maximum extent to ensure safety. When the threshold is between the two thresholds, the system enters the path reconstruction observation phase. Through continuous sampling trend analysis and branch switching verification, it determines whether recovery conditions are met, thereby enhancing the ability to identify short-term disturbances and stable regressions. When the threshold is above the first recovery threshold, the system automatically executes a recovery operation, recording the entire process and verifying delayed responses to establish a closed-loop monitoring and tracking mechanism. Overall, this improves the system's decision-making accuracy, response rationality, and path adaptability control capabilities during the recovery phase, effectively mitigating the fluctuation risks caused by coolant leaks.

[0077] The second aspect of this invention provides a two-phase cold plate liquid cooling leakage current limiting control system for data centers. Figure 2 shows a structural diagram of the two-phase cold plate liquid cooling leakage current limiting control system for data centers provided in this application example. It includes: a distributed cold plate status data acquisition module, used to collect coolant operating status data, control status data and execution feedback data during the liquid cooling process, and preprocess the collected coolant operating status data, control status data and execution feedback data to construct a standardized cooling status dataset. The coolant operating status data includes flow rate, pressure and temperature information in each cold plate channel. The control status data includes electromagnetic proportional valve signal output, power control interface status and current gear command of distributor 4. The execution feedback data includes valve core response delay, adjustment result error and path switching delay information. The preprocessing process includes time alignment, missing data removal and fluctuation noise reduction processing of the sampled data.

[0078] The pressure-flow coordination judgment and analysis module is used to analyze the steady-state deviation characteristics of coolant based on a standardized cooling state dataset, and dynamically determine the risk of coolant leakage and adjust the coolant supply status based on the analysis results. The steady-state deviation characteristics include the synchronicity of pressure drop and flow decay in the supply path, and the coupling trend of pressure drop rate and pullback magnitude. The leakage risk judgment includes the comparison of the coordination deviation score of the current monitoring path with the steady-state tolerance threshold. The supply status adjustment includes locking the diversion level, reducing the target supply rate, and switching the condenser recirculation mode.

[0079] The stepped feedback flow limiting regulation module is used to evaluate the flow limiting regulation state based on the coupling relationship between the valve core opening change rate and the coolant flow response, using a standardized cooling state dataset. Based on the evaluation results, the module dynamically corrects the operating state of the electromagnetic proportional valve. The coupling relationship evaluation includes a comparison of the flow response slope and a response hysteresis measurement at different opening stages. The correction of the electromagnetic proportional valve operating state includes fine-tuning of the output signal duty cycle, correction of the adjustment step size, and dynamic adjustment of the adjustment trigger frequency.

[0080] The leakage recovery dynamic discrimination module is used to comprehensively evaluate the coolant supply recovery trend based on the coolant steady-state offset characteristic analysis results and the flow restriction adjustment state evaluation results. Based on the evaluation results, it adjusts the recovery stage and adjustment path of the cooling circuit in real time. The recovery trend evaluation includes the fitting and convergence judgment of the joint trend of pressure and flow recovery rate. The recovery stage control includes whether to switch to the preparatory branch, whether to trigger the initial coolant supply state, and whether to activate the coolant redistribution logic of distributor 4. The adjustment path selection includes the path switching sequence, control granularity, and control circuit execution strategy configuration.

[0081] In this implementation scheme, the distributed cold plate status data acquisition module provides a high-quality, structured cooling operation data foundation for subsequent analysis. This module is responsible not only for collecting the real-time operating status of the coolant in the multi-channel two-phase cold plate 5, but also for collecting the control output status and feedback results of the electromagnetic proportional valve and distributor. A standardized cooling status dataset is constructed through a preprocessing process to ensure that downstream modules have a consistent and reliable data source.

[0082] The pressure-flow coordinated judgment and analysis module identifies abnormal deviations in the cooling path and provides early warnings of potential leakage risks. This module analyzes the coupling characteristics between coolant pressure drop and flow rate decay, combining the rate and magnitude of change to assess in real time whether the coolant deviates from the steady-state supply pattern, determine if there is a significant leakage trend, and adjust the coolant supply strategy based on the risk level, including downgrading the supply level and switching the recirculation mode.

[0083] The stepped feedback flow limiting adjustment module optimizes the response strategy of the electromagnetic proportional valve in actual operation, improving the sensitivity and stability of flow limiting adjustment. This module analyzes the flow response trend caused by each change in valve core opening to determine the effectiveness of the adjustment. It further refines the feedback characteristics by stage, enabling dynamic fine-tuning of the duty cycle, adaptive adjustment step size, and adaptive correction of feedback delay. This ensures a high degree of coupling between the flow limiting adjustment process and actual cooling requirements.

[0084] The leak recovery dynamic judgment module, upon detecting a coolant leak and entering a flow-limited state, determines whether the cooling circuit has the conditions to restore coolant supply and drives path reconstruction and coolant supply recovery operations accordingly. This module uses steady-state offset characteristics and flow-limiting assessment results as input, and by fitting the recovery trends of pressure and flow, identifies whether the recovery is truly effective. It then controls whether to switch to a branch circuit, whether to activate condenser recirculation distribution, and verifies sensor response consistency, providing a decision-making basis for the safety and efficiency of the recovery phase.

[0085] 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.

[0086] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for limiting leakage current in two-phase cold plate liquid cooling systems for data centers, characterized in that: include: S1. Collect coolant operating status data, control status data, and execution feedback data during the liquid cooling process, and preprocess the collected coolant operating status data, control status data, and execution feedback data to construct a standardized cooling status dataset. S2. Based on the standardized cooling status dataset, perform feature analysis on the coolant steady-state deviation characteristics, and dynamically determine the coolant leakage risk and adjust the coolant supply status based on the feature analysis results. S3. Based on the standardized cooling status dataset, evaluate the flow-limiting regulation status from the coupling relationship between the valve core opening change rate and the coolant flow response, and dynamically correct the electromagnetic proportional valve's operating status based on the flow-limiting regulation status evaluation results. S4. Using the coolant steady-state deviation feature analysis results and the flow-limiting regulation status evaluation results as input, comprehensively evaluate the coolant supply recovery trend, and adjust the recovery stage and regulation path of the cooling circuit in real time based on the comprehensive evaluation results.

2. The two-phase cold plate liquid cooling leakage current limiting control method for data centers according to claim 1, characterized in that: The specific steps for collecting coolant operating status data, control status data, and execution feedback data during the liquid cooling process are as follows: Coolant operating status data is collected using high-precision pressure sensors (3) and flow sensors (2) installed at key nodes of the two-phase cold plate (5). The coolant operating status data includes the current static pressure value and instantaneous flow rate value of the coolant. Under stable operating conditions, the average static pressure level of the target monitoring section under steady-state conditions is calculated using an arithmetic average algorithm based on all collected static pressure values, and recorded as the static pressure equilibrium value. The average result of all steady-state flow rates is calculated using a transverse average algorithm to obtain the current loop's flow rate level under steady-state conditions, recorded as the flow equilibrium value. Within the same time window, each monitoring... The static pressure values ​​collected at each sampling point are calculated using the standard deviation method to obtain the natural fluctuation level of the static pressure value, which is recorded as the static pressure fluctuation tolerance value. Simultaneously, based on the set of instantaneous flow values ​​sampled within the time window, the standard deviation is applied to calculate the flow velocity fluctuation amplitude under undisturbed conditions, which is recorded as the flow fluctuation tolerance value. Control status data is collected through a high-speed data acquisition unit integrated into the edge control module. The control status data includes: the leakage velocity value monitored at each sampling time and the initial leakage velocity value when the first leakage is detected, while also recording the number of sampling cycles and the sampling interval time. System execution feedback data is collected through a linkage execution recording mechanism. The execution feedback data includes: the sampling time interval between two samplings and the duration of the flow limiting adjustment phase.

3. The two-phase cold plate liquid cooling leakage current limiting control method for data centers according to claim 1, characterized in that: The specific steps for preprocessing the collected coolant operating status data, control status data, and execution feedback data to construct a standardized cooling status dataset are as follows: Data cleaning is performed on the collected coolant operating status data, control status data, and execution feedback data. Abnormal fluctuations, missing values, and abrupt changes are removed using sliding window filtering and interpolation completion. After cleaning, steady-state segments are identified for static pressure values ​​and instantaneous flow rates, and stable time windows are extracted. Static pressure equilibrium values ​​and flow equilibrium values ​​are calculated based on the data within the stable time windows. For control status data, the leakage rate values ​​at each moment are continuously verified, and the initial leakage rate value is combined with the corresponding sampling period and sampling interval for unified labeling and time serialization. For execution feedback data, the sampling time interval and the duration of the flow-limiting adjustment phase are formatted, and records from different time periods are archived to ensure alignment with the coolant operating status data in the time dimension. The standardized coolant operating status data, control status data, and execution feedback data are subjected to dimensionless normalization to construct a standardized coolant status dataset. The complete process of synchronously collecting and preprocessing coolant operating status data, control status data, and execution feedback data is recorded as one sampling cycle.

4. The two-phase cold plate liquid cooling leakage current limiting control method for data centers according to claim 1, characterized in that: The specific steps for feature analysis of coolant steady-state deviation characteristics based on the standardized cooling state dataset are as follows: Perform a difference operation on the instantaneous flow rate value, subtracting the instantaneous flow rate value of the previous sampling period from the current sampling period's instantaneous flow rate value and dividing by the sampling time interval to obtain the flow rate change rate per unit time, recorded as the flow rate change value; subtract the static pressure equilibrium value from the current static pressure value, divide by the static pressure fluctuation tolerance value, and take the absolute value to obtain the static pressure anomaly value; subtract the flow equilibrium value from the current instantaneous flow rate value, divide by the flow fluctuation tolerance value, and take the absolute value to obtain the flow anomaly value; divide the flow rate change value by the current instantaneous flow rate value and take the absolute value to obtain the flow rate change rate; divide the static pressure fluctuation tolerance value by the static pressure equilibrium value and take the absolute value to obtain the static pressure relative fluctuation value; add the static pressure anomaly value, flow rate anomaly value, flow rate change rate, and static pressure relative fluctuation value to obtain the leakage collaborative judgment value.

5. The two-phase cold plate liquid cooling leakage current limiting control method for data centers according to claim 1, characterized in that: The specific steps for dynamically determining the risk of coolant leakage and adjusting the coolant supply based on the feature analysis results are as follows: Real-time comparison of the leakage collaborative judgment value and the leakage judgment threshold of the current monitoring channel: When the leakage collaborative judgment value is less than or equal to the leakage judgment threshold, the distributor (4) maintains the existing diversion level, the two-phase cold plate (5) maintains the current heat exchange channel configuration, the condenser (7) maintains the current reflux rhythm, and the monitor (1) takes a single snapshot of the current sampling cycle channel status and marks it as a regular record; When the leakage collaborative judgment value is greater than the leakage judgment threshold, the monitor (1) sends a signal to the distributor (4) The system issues a supply restriction command, lowers the current branch flow ratio and locks the valve position, increases the return flow rate of the condenser (7), increases the sampling frequency of the pressure sensor (3) and the flow sensor (2) and opens the zero-point verification, and applies a duty cycle adjustment mechanism in the branch channel corresponding to the two-phase cold plate (5): the liquid supply signal is pulse reconstructed so that the pump output flow only maintains short-term continuous liquid supply at the beginning of each start-up, and then quickly enters the intermittent pump stop state, keeping the pump stop until the pressure sensor (3) detects that the coolant pressure has recovered to a stable state and then resumes short-term liquid supply to reduce the input rate, while triggering the audible and visual alarm device.

6. The two-phase cold plate liquid cooling leakage current limiting control method for data centers according to claim 1, characterized in that: The specific steps for evaluating the flow-limiting regulation state based on the standardized cooling state dataset and the coupling relationship between the valve core opening change rate and the coolant flow response are as follows: Calculate the difference in leakage velocity values ​​between each adjacent moment and divide it by the sampling time interval; take the absolute value of the ratio of the leakage velocity value difference to the sampling time interval to obtain the instantaneous leakage velocity change rate; divide the leakage velocity value at each sampling moment by the initial leakage velocity value to obtain the relative leakage intensity ratio; multiply the instantaneous leakage velocity change rate by the relative leakage intensity ratio to obtain the coordinated change intensity of the current sampling period; sum the coordinated change intensities of each sampling period and divide by the number of sampling periods to obtain the average coordinated intensity; subtract the duration of the current flow-limiting regulation phase from the average coordinated intensity to obtain the electromagnetic flow-limiting duty cycle.

7. The two-phase cold plate liquid cooling leakage current limiting control method for data centers according to claim 1, characterized in that: The specific steps for dynamically correcting the working state of the electromagnetic proportional valve based on the evaluation results of the current limiting adjustment state are as follows: real-time acquisition of the current electromagnetic current limiting duty cycle and execution of the following adjustment measures: when the electromagnetic current limiting duty cycle of the current sampling period shows a downward trend compared to the previous sampling period, continuously send the reduced pulse signal to the electromagnetic proportional valve, adjust the valve core opening in real time, and further limit the coolant input flow rate; introduce the coolant into the buffer cavity structure located in the inlet section of the two-phase cold plate (5) to absorb the forward pressure in the main path, and simultaneously feed back the heat source load information to the upper power control interface to reduce the power supply power of the server (6); When the electromagnetic current limiting duty cycle of the current sampling period shows an upward trend compared to the previous sampling period, the duty cycle of the electromagnetic proportional valve pulse signal is increased, so that the valve core is gradually opened, and the liquid separator (4) is opened in conjunction to intercept the micro air particles entrained in the recovery flow. At the same time, the pressure regulating spring slow release valve of the two-phase cold plate (5) is started simultaneously. When the electromagnetic current limiting duty cycle of the current sampling period remains unchanged compared to the previous sampling period, the current output signal of the electromagnetic proportional valve is kept unchanged, and the active self-test program of the differential pressure balancer at the end of the two-phase cold plate (5) is triggered.

8. The two-phase cold plate liquid cooling leakage current limiting control method for data centers according to claim 1, characterized in that: The specific steps for comprehensively evaluating the coolant supply recovery trend using the coolant steady-state offset characteristic analysis results and the flow-limiting regulation state evaluation results as input are as follows: Subtract the static pressure equilibrium value from the collected static pressure values ​​at each sampling time and divide by the static pressure fluctuation tolerance value to obtain a standardized pressure offset sequence; subtract the flow equilibrium value from the instantaneous flow rate value and divide by the flow fluctuation tolerance value to obtain a standardized flow rate offset sequence; construct a joint fluctuation vector based on the standardized pressure offset sequence and flow rate offset sequence at each sampling time, and calculate the magnitude of the joint fluctuation vector to obtain the joint offset intensity at the current sampling time. Within the entire time window, the sliding variance processing of all joint offset intensities is performed to obtain the joint pressure-flow fluctuation amplitude within the current sampling period. The difference between one and the leakage co-judgment value is multiplied by the difference between one and the electromagnetic current limiting duty cycle to obtain the leakage current limiting coupling value. The absolute value of the joint pressure-flow fluctuation amplitude is taken, plus one, and then the logarithm is taken and plus one again to obtain the joint disturbance suppression value. The electromagnetic current limiting duty cycle is multiplied by the corresponding recovery adjustment factor and then plus one to obtain the coupling adjustment amplitude value. The leakage current limiting coupling value is divided by the joint disturbance suppression value and then multiplied by the coupling adjustment amplitude value to obtain the recovery dynamic discrimination value.

9. The two-phase cold plate liquid cooling leakage current limiting control method for data centers according to claim 1, characterized in that: The specific steps of the recovery stage and adjustment path of the cooling circuit in real time based on the comprehensive evaluation results are as follows: compare the current recovery dynamic discrimination value with the recovery discrimination threshold in real time. The recovery discrimination threshold includes the first recovery threshold and the second recovery threshold, wherein the first recovery threshold is higher than the second recovery threshold. When the recovery dynamic discrimination value is less than or equal to the second recovery threshold, keep the current flow limiting state of the electromagnetic proportional valve unchanged, prohibit any form of liquid supply recovery command from being issued, and at the same time perform high-frequency sampling through the pressure sensor (3) and flow sensor (2) on the path from the condenser (7) to the two-phase cold plate (5) to maintain the minimum flow input state before the leakage is controllable. When the recovery dynamic discrimination value is higher than the second recovery threshold and less than or equal to the first recovery threshold, the path reconstruction observation stage begins: Based on the flow sensor (2) and pressure sensor (3), the trend convergence at each monitoring node is continuously monitored. When the data collected by the flow sensor (2) and pressure sensor (3) shows a stable upward trend in two consecutive sampling periods, the branch loop with a length shorter than the current loop is switched through the distributor (4) to help determine whether the recovery is a stable regression rather than a short-term disturbance. When the coolant operating status data collected by the flow sensor (2) and pressure sensor (3) shows a downward trend in two consecutive sampling periods, the preparation stage ends. The process extends the current path reconstruction observation phase; when the recovery dynamic discrimination value is greater than the first recovery threshold, the liquid supply recovery operation is initiated immediately: the electromagnetic proportional valve is quickly adjusted to the initial valve state, and the distributor (4) is activated to redistribute the coolant recovered in the condenser (7) to each parallel two-phase cold plate (5) channel, and the coolant operation status data is recorded by the monitor (1) throughout the process of the coolant recovering from the flow-limited state to the stable liquid supply state; during the stable period after the liquid supply recovery, the time verification mechanism of the flow sensor (2) and the pressure sensor (3) is activated to identify potential delayed response segments, and the path response lag time of this round is written into the path diagnosis log.

10. A two-phase cold plate liquid cooling leakage current limiting control system for data centers, employing the two-phase cold plate liquid cooling leakage current limiting control method for data centers as described in any one of claims 1-9, characterized in that: include: The distributed cold plate status data acquisition module is used to collect coolant operating status data, control status data, and execution feedback data during the liquid cooling process, and to preprocess the collected coolant operating status data, control status data, and execution feedback data to construct a standardized cooling status dataset. The pressure-flow collaborative judgment and analysis module is used to analyze the steady-state deviation characteristics of coolant based on a standardized cooling state dataset, and dynamically determine the risk of coolant leakage and adjust the coolant supply status based on the analysis results. The stepped feedback flow limiting regulation module is used to evaluate the flow limiting regulation state based on the coupling relationship between the valve core opening change rate and the coolant flow response using a standardized cooling state dataset, and dynamically corrects the working state of the electromagnetic proportional valve based on the evaluation results. The leakage recovery dynamic discrimination module is used to comprehensively evaluate the coolant supply recovery trend using the coolant steady-state deviation characteristic analysis results and the flow limiting regulation state evaluation results as input, and adjusts the recovery stage and regulation path of the cooling circuit in real time based on the evaluation results.

Images