Intelligent monitoring system of liquid cooling cabinet and processing equipment thereof

Abstract

The invention provides an intelligent monitoring system of a liquid cooling cabinet and processing equipment thereof, and relates to the technical field of data processing, and the system comprises a cold source self-control module which is used for dynamically adjusting the operation state of a cooling tower of primary side circulation and the flow of a circulating water pump according to cooling water parameters, and achieving the variable flow control of cooling water; the dynamic environment monitoring module is used for monitoring the secondary side pipeline state and the cabinet environment in real time through the cooling capacity distribution unit and the distributed sensor network; and the comprehensive monitoring sub-module is used for realizing leakage positioning through a built-in liquid leakage sensor, a temperature sensor and a pressure sensor, and automatically supplementing liquid by using a liquid storage tank. Through multi-mode sensor fusion, dynamic threshold value self-adaption and multi-parameter collaborative analysis, accurate leakage positioning and graded alarm are achieved, automatic liquid supplementing closed-loop control and variable flow adjustment are combined, the system reliability is improved, and energy consumption is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of heat exchange devices, more particularly to the technical field of monitoring of liquid-cooled cabinets, and particularly to an intelligent monitoring system for a liquid-cooled cabinet and a processing device thereof. BACKGROUND

[0002] Traditional monitoring relies on a small number of scattered sensors, and there are blind spots in the monitoring of the secondary side pipeline and the cabinet environment. For example, if a local overheat occurs due to a blockage in the middle section of the pipeline, it is difficult to discover in time by only setting a temperature sensor at the cabinet inlet. The dynamic ring monitoring module can master the state of each place in real time and comprehensively through a distributed sensor network. Under the traditional mode, leakage detection relies on manual inspection or simple sensing devices, and when leakage is discovered, a large loss has often been caused. For example, because of untimely leakage detection, the leakage of cooling liquid caused the short circuit damage of a server in a computer room. In addition, the intelligent monitoring system does not have the function of automatic liquid supplement, and once the cooling liquid is insufficient, manual intervention is required for supplement, which is low in efficiency and significantly lags behind the automatic positioning and liquid supplement capacity of the comprehensive monitoring sub-module.

[0003] Traditional monitoring sets a fixed threshold value for a single parameter to alarm, and lacks multi-source data fusion analysis. For example, an alarm is only triggered when the temperature exceeds the threshold value, without comprehensive judgment in combination with parameters such as flow and pressure, which is prone to false alarms or missed alarms. There was a false alarm triggered by temporary fluctuation of the ambient temperature, which interfered with the operation and maintenance work, but failed to timely detect the real leakage. SUMMARY

[0004] The technical problem to be solved by the present application is to provide an intelligent monitoring system for a liquid-cooled cabinet and a processing device thereof, which realizes variable flow control of cooling water, and comprehensively monitors the pipeline, the environment and the leakage and automatically processes.

[0005] To solve the above technical problems, the technical solutions of the present application are as follows:

[0006] In a first aspect, the intelligent monitoring system for a liquid-cooled cabinet comprises:

[0007] a cold source self-control module, configured to dynamically adjust the operation state of a cooling tower and the flow of a circulating water pump of a primary side circulation according to cooling water parameters, so as to realize variable flow control of the cooling water;

[0008] a dynamic ring monitoring module, configured to monitor the state of a secondary side pipeline and the environment of a cabinet in real time through a cold distribution unit and a distributed sensor network;

[0009] a comprehensive monitoring sub-module, configured to realize leakage positioning through an internal liquid leakage sensor, a temperature sensor and a pressure sensor, and to automatically supplement liquid by using a liquid storage tank;

[0010] The intelligent analysis and alarm module is used to integrate multi-source data, including cooling water temperature, pressure, flow parameters, leakage signals, humidity gradient and pressure difference data, to calculate the dynamic detection threshold and trigger graded alarms based on the synergistic relationship between continuous negative deviation of flow, asymmetric distortion of temperature field, sudden change of humidity gradient and matching of pressure difference spectrum characteristics.

[0011] Furthermore, the cooling tower operating state and circulating water pump flow of the primary side circulation are dynamically adjusted according to the cooling water parameters to achieve variable flow control of the cooling water, including:

[0012] The temperature, pressure and flow data of cooling water are collected in real time through the temperature sensor, pressure sensor and mass flow meter in the primary side circulation pipeline of cooling water;

[0013] The cooling water temperature, pressure and flow data are input into the PLC controller, and the cooling tower heat exchange efficiency is calculated by the approximation algorithm and the difference between the cooling tower outlet water temperature and the outdoor wet bulb temperature;

[0014] Dynamically adjust the cooling tower fan frequency and number of running units according to the heat exchange efficiency;

[0015] According to the adjusted operating status of the cooling tower, combined with the actual demand for cooling water flow and energy consumption optimization goals, the PLC controller adjusts the speed of the circulating water pump through the frequency converter to achieve variable flow control of cooling water.

[0016] Furthermore, the cooling water temperature, pressure and flow data are input into the PLC controller. The cooling tower heat exchange efficiency is calculated by the approximation algorithm and the difference between the cooling tower outlet water temperature and the outdoor wet bulb temperature, including:

[0017] The PLC controller extracts the cooling tower outlet water temperature and the outdoor wet-bulb temperature data from the storage area, and obtains the difference between the cooling tower outlet water temperature and the outdoor wet-bulb temperature data;

[0018] The PLC controller starts the approximation algorithm, compares the temperature difference with the preset difference range, and obtains the comparison result;

[0019] Based on the comparison results, a comprehensive analysis is performed in combination with the temperature, pressure and flow data of the cooling water. If the temperature difference is within the preset difference range, the potential impact of fluctuations in each parameter on the efficiency is analyzed; if the temperature difference is ≥ the preset difference range, the abnormal parameters are determined, and finally the heat exchange efficiency of the cooling tower is obtained.

[0020] Furthermore, the cooling distribution unit and distributed sensor network monitor the secondary side pipeline status and cabinet environment in real time, including:

[0021] Install an array of optical fiber temperature sensors on the secondary side pipeline, distribute them along the pipeline to form a temperature field monitoring network, and collect temperature data at different locations on the secondary side pipeline in real time;

[0022] Temperature anomaly data is extracted from the temperature data at different locations on the secondary side pipeline. At the same time, the pressure differential sensor in the cooling distribution unit monitors the secondary side inlet and outlet pressure differential data in real time. When the secondary side inlet and outlet pressure differential is ≥ the preset range and the change trend meets the characteristics of pipeline blockage, the pipeline is assessed to have a blockage risk based on the preset threshold and logical judgment rules, and an early warning signal is issued;

[0023] After receiving the pipeline blockage risk warning signal, the humidity sensor array and temperature and humidity sensor inside the cabinet simultaneously collect cabinet environmental parameters; the humidity sensor array detects changes in environmental humidity inside the cabinet, and the temperature and humidity sensor collects environmental temperature and humidity data.

[0024] Furthermore, leak location is achieved through built-in leakage sensors, temperature sensors, and pressure sensors, and automatic liquid replenishment is performed using a liquid storage tank, including:

[0025] A liquid collection tray is set at the bottom of the cooling distribution unit, and a leakage detection cable is laid inside the tray. When the coolant leaks, an impedance change signal is triggered, and the impedance change signal is cross-verified with the signal sent by the leakage sensor;

[0026] After confirming the coolant leak, the leak point is located through the temperature field distortion characteristics of the distributed optical fiber temperature sensor, and the spatial position information of the leak point is spatially matched with the leakage signal;

[0027] When it is detected that the liquid level in the liquid storage tank is less than the preset threshold, the refill pump is immediately started to inject coolant into the cooling circulation loop.

[0028] Furthermore, multi-source data, including cooling water temperature, pressure, flow parameters, leakage signals, humidity gradient, and pressure differential data, are integrated to calculate dynamic detection thresholds. Graded alarms are triggered based on the synergistic relationship between sustained negative flow deviation, asymmetric temperature field distortion, sudden humidity gradient changes, and pressure differential spectrum feature matching, including:

[0029] The temperature, pressure, and flow data of the primary side cooling water, the secondary side leakage signal, the humidity gradient data in the cabinet, and the pressure difference data of the cooling distribution unit are input into the data fusion unit. The data of the data fusion unit is preprocessed through time series to generate a denoised time series data set.

[0030] Based on the denoised time series data set, the covariance between parameters is calculated, periodic fluctuations are detected, and abnormal patterns are identified. Combined with the correlation characteristics of normal operating conditions in historical operating data, an adaptive dynamic threshold for leak detection is calculated.

[0031] The dynamic threshold for leak detection is adjusted based on the real-time data stream. When the cooling water flow rate has a continuous negative deviation ≥ the dynamic threshold for leak detection and the distributed fiber optic temperature sensor network detects asymmetric distortion of the temperature field, it is determined to be an early leakage risk and an early warning signal is triggered. If, on the basis of triggering the early warning signal, it is also detected that the humidity difference between adjacent sensors in the cabinet is ≥ the set threshold and the pressure difference spectrum shows leakage characteristics, it will be upgraded to a high-confidence leak alarm.

[0032] Furthermore, based on the denoised time series data set, we calculate the covariance between parameters, detect periodic fluctuations, and identify abnormal patterns. Combined with the correlation characteristics of normal operating conditions in historical operating data, we calculate an adaptive dynamic threshold for leak detection, including:

[0033] Collect multi-source data, including primary-side cooling water temperature, pressure, and flow data, secondary-side leakage signals, cabinet humidity gradient data, and cooling distribution unit pressure difference data;

[0034] For any two parameters in multi-source data, the covariance between the two parameters is calculated to obtain the weighted average of the covariance between the parameters;

[0035] For each parameter in the multi-source data, the periodic characteristics of the parameter are determined by analyzing the value change pattern at different time points. Based on the periodic characteristics, the theoretical value of the corresponding parameter at each time point is predicted to determine the periodic fluctuation amplitude of the corresponding parameter. Based on the periodic fluctuation amplitude of each parameter, the weighted average value of the parameter periodic fluctuation is determined.

[0036] Analyze multi-source data and use preset anomaly judgment rules to identify different anomaly patterns. For each anomaly pattern, count the number of anomaly points in the multi-source data that match the anomaly pattern and record the total number of data points. Determine the severity of the anomaly pattern based on the number of anomaly points and the total number of data points for each anomaly pattern, and obtain a weighted average of the severity of the anomaly pattern.

[0037] Extract the correlation features of each parameter under normal operating conditions from historical operating data and convert the correlation features of each parameter into a single characteristic value;

[0038] An adaptive dynamic threshold for leak detection is determined based on the weighted average of covariances between parameters, the weighted average of periodic fluctuations of parameters, the weighted average of severity of abnormal patterns, and a single characteristic value of historical operating data.

[0039] Secondly, the processing equipment of liquid cooling cabinets includes:

[0040] Customized cooling tower processing equipment for manufacturing closed cooling towers with integrated variable frequency fans and optimized heat exchange efficiency structures;

[0041] Pipeline processing equipment is used to form primary and secondary circulation pipelines and pre-embed distributed optical fiber temperature sensors and leakage detection cables inside the pipe walls to achieve real-time monitoring of pipeline status;

[0042] The cold distribution unit assembly line is used to produce cold distribution units with integrated double-walled liquid storage tanks. Self-cleaning differential pressure sensors are installed at the inlet and outlet of the cold distribution unit, and an automatic liquid replenishment pump based on liquid level feedback control is integrated;

[0043] The automatic sensor deployment device is used to navigate through lidar, install a humidity gradient detection array on the inner wall of the cabinet, attach temperature sensors to the surface of the cold plate, and automatically complete the wiring of sensor signal cables.

[0044] According to a third aspect, a computing device includes:

[0045] one or more processors;

[0046] The storage device is used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the system.

[0047] In a fourth aspect, a computer-readable storage medium stores a program, which implements the system when executed by a processor.

[0048] The above solution of the present invention includes at least the following beneficial effects:

[0049] The cooling source control module dynamically adjusts the cooling tower and circulating water pump according to the real-time cooling water parameters, changes the cooling water flow rate, and avoids high-energy consumption operation under the traditional fixed mode. Under low-load conditions, it reduces the operating intensity of the equipment and reduces power consumption; under high load conditions, it ensures sufficient cooling, improves energy utilization, and reduces operating costs in places such as data centers. The dynamic environment monitoring module uses the cooling capacity distribution unit and distributed sensor network to conduct all-round real-time monitoring of the secondary side pipeline and cabinet environment. The temperature field monitoring network constructed by the fiber optic temperature sensor array, combined with the pressure differential sensor's monitoring of the inlet and outlet pressures, can promptly capture subtle changes such as abnormal pipeline temperature and pressure fluctuations; the humidity sensor array and temperature and humidity sensors in the cabinet can grasp the environmental temperature and humidity status in real time, realize the early detection of hidden dangers, and provide strong guarantees for the stable operation of the system.

[0050] The integrated monitoring submodule utilizes multiple sensors, including leak detection cables and distributed fiber-optic temperature sensors, to quickly and accurately locate coolant leaks. If a leak occurs, the leak point can be quickly pinpointed, and the automatic refill function in the reservoir ensures timely coolant replenishment. This prevents equipment downtime caused by leaks, reduces repair costs and downtime losses, and ensures continuous and stable system operation. The intelligent analysis and alarm module integrates multi-source data to calculate dynamic detection thresholds, transforming the traditional single fixed threshold alarm approach. By analyzing the synergistic relationships between parameters such as flow rate, temperature, humidity, and differential pressure, it triggers graded alarms based on characteristics such as persistent negative flow rate deviation and asymmetric temperature field distortion. This accurately identifies different levels of fault risk, reduces false and missed alarms, and enables operations and maintenance personnel to quickly take appropriate action based on the alarm level, improving troubleshooting efficiency. These modules work seamlessly together to form a complete closed loop from cooling control, environmental monitoring, leak handling, to intelligent alarms, enabling intelligent monitoring and management of the entire liquid-cooled cabinet process. Reduce manual inspections and interventions, improve operation and maintenance efficiency, reduce the risk of human error, provide stable and reliable liquid cooling guarantees for key facilities such as data centers, and promote the development of liquid cooling system operation and maintenance towards intelligence and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 Schematic diagram of an intelligent monitoring system for a liquid cooling cabinet provided by an embodiment of the present invention.

[0052] Figure 2 It is a schematic diagram of processing equipment for a liquid cooling cabinet provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0053] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0054] like Figure 1 As shown, the present invention relates to a heat exchange device. The intelligent monitoring system for a liquid cooling cabinet proposed by the present invention includes:

[0055] The cooling source control module 1 is used to dynamically adjust the cooling tower operating status and circulating water pump flow of the primary side circulation according to the cooling water parameters to achieve variable flow control of the cooling water;

[0056] Dynamic environment monitoring module 2, used to monitor the secondary side pipeline status and cabinet environment in real time through the cooling distribution unit and distributed sensor network;

[0057] Comprehensive monitoring submodule 3, used to locate leaks through built-in leakage sensors, temperature sensors, and pressure sensors, and automatically replenish liquid using a liquid storage tank;

[0058] Intelligent analysis and alarm module 4 is used to integrate multi-source data, including cooling water temperature, pressure, flow parameters, leakage signals, humidity gradient and pressure difference data, to calculate the dynamic detection threshold and trigger graded alarms based on the synergistic relationship between continuous negative deviation of flow, asymmetric distortion of temperature field, sudden change of humidity gradient and pressure difference spectrum feature matching.

[0059] In an embodiment of the present invention, the cooling source control module dynamically adjusts the cooling tower and the circulating water pump according to the real-time parameters of the cooling water to change the cooling water flow rate. Under low-load conditions, the equipment operation intensity is reduced to reduce energy consumption, while sufficient cooling can be guaranteed under high load conditions, avoiding energy waste while ensuring that the equipment in the cabinet is always at a suitable operating temperature, thereby improving equipment stability and service life. The dynamic environment monitoring module uses the cooling distribution unit and the distributed sensor network to carry out blind spot monitoring of the secondary side pipeline status and cabinet environment. Whether it is abnormal pipeline temperature, pressure fluctuations, or changes in temperature and humidity in the cabinet, they can be captured in time to provide comprehensive data support for the stable operation of the system, so that operation and maintenance personnel can grasp the system operation status in real time.

[0060] The integrated monitoring submodule uses multiple sensors to precisely locate leaks and automatically refills the liquid from the reservoir. If a coolant leak occurs, the leak's location is quickly determined, minimizing its impact. Coolant is automatically replenished to maintain normal cooling circulation, preventing equipment downtime caused by leaks and reducing repair costs and downtime losses. The intelligent analysis and alarm module integrates multi-source data to calculate dynamic detection thresholds and trigger graded alarms based on the synergistic relationships between parameters. Compared to traditional single-threshold alarm methods, this module more accurately identifies varying degrees of fault risk, reducing false and missed alarms. This allows operations and maintenance personnel to quickly take appropriate action based on the alarm level, improving troubleshooting efficiency and ensuring the safe operation of the liquid cooling cabinet. The modules work together collaboratively to form an integrated whole. From cooling control to environmental monitoring, to fault handling and early warning, this system enables comprehensive, intelligent monitoring and management of the liquid cooling cabinet, enhancing the reliability, intelligence, and operational efficiency of the entire liquid cooling system.

[0061] In a preferred embodiment of the present invention, the cooling tower operating state and the circulating water pump flow rate of the primary side circulation are dynamically adjusted according to the cooling water parameters to achieve variable flow control of the cooling water, which may include:

[0062] The temperature sensor, pressure sensor and mass flow meter in the primary side circulation pipeline of the cooling water are used to collect the cooling water temperature, pressure and flow data in real time;

[0063] The cooling water temperature, pressure and flow data are input into the PLC controller, and the cooling tower heat exchange efficiency is calculated by the approximation algorithm and the difference between the cooling tower outlet water temperature and the outdoor wet bulb temperature;

[0064] Dynamically adjust the cooling tower fan frequency and number of running units according to the heat exchange efficiency;

[0065] According to the adjusted operating status of the cooling tower, combined with the actual demand for cooling water flow and energy consumption optimization goals, the PLC controller adjusts the speed of the circulating water pump through the frequency converter to achieve variable flow control of cooling water.

[0066] In an embodiment of the present invention, high-precision temperature sensors, pressure sensors, and mass flowmeters are installed at key nodes in the primary cooling water circulation pipeline. These sensors continuously monitor the temperature, pressure, and flow rate of the cooling water in the pipeline at a set sampling frequency (e.g., once per second). The sensors convert the detected physical signals (e.g., thermal signals corresponding to temperature and mechanical signals corresponding to pressure) into electrical signals, which are then transmitted to subsequent data processing equipment via data transmission lines (e.g., shielded cables or optical fibers). A programmable logic controller (PLC) receives the cooling water temperature, pressure, and flow rate data transmitted by the sensors via its data interface. Simultaneously, the PLC controller extracts cooling tower outlet water temperature data and outdoor wet-bulb temperature data from a connected storage area or other monitoring device. The PLC controller then activates an internally preset approximation algorithm, which compares and analyzes the difference between the cooling tower outlet water temperature and the outdoor wet-bulb temperature within a preset range. During this comparison, a comprehensive calculation is performed based on the input cooling water temperature, pressure, and flow rate data. For example, by analyzing the changing trends of these parameters and the correlations between them, the cooling tower's heat exchange efficiency value is ultimately determined.

[0067] The PLC controller compares the calculated cooling tower heat exchange efficiency with a pre-set efficiency standard. If the heat exchange efficiency exceeds the standard, it indicates that the cooling tower has excess cooling capacity. The PLC controller then sends a control signal to the inverter to reduce the operating frequency of the cooling tower fan or reduce the number of cooling tower fans in operation. Conversely, if the heat exchange efficiency is below the standard, the PLC controller sends a signal to increase the fan frequency or increase the number of fans in operation, thereby dynamically adjusting the cooling tower's heat dissipation capacity to ensure it matches actual cooling needs. After adjusting the cooling tower fan frequency and the number of fans in operation, the PLC controller evaluates the current cooling tower operating status, taking into account the system's actual demand for cooling water flow (for example, determining the required cooling capacity based on the load of the equipment in the cabinet) and energy optimization goals (such as a set energy consumption cap or energy saving ratio). The PLC controller then sends a control command to the inverter connected to the circulating water pump. The inverter changes the frequency of the output power supply based on the command, thereby adjusting the speed of the circulating water pump. Changes in the water pump speed directly affect the flow rate of cooling water. When more cooling water is needed, the water pump speed is increased to increase the flow rate; when the cooling demand decreases, the water pump speed is reduced to reduce the flow rate, thereby achieving precise control of the variable flow rate of cooling water.

[0068] Imagine a data center experiencing peak daytime business hours. The equipment load inside the cabinets increases, generating a significant increase in heat. At this point, temperature sensors installed in the cooling water primary circulation piping detect the rise in water temperature, while pressure sensors and mass flow meters also monitor corresponding changes in pressure and flow. This data is transmitted to a programmable logic controller (PLC). The controller, using an approximation algorithm, combines the extracted cooling tower outlet water temperature and outdoor wet-bulb temperature data to calculate a decrease in the cooling tower's heat exchange efficiency. The PLC controller then increases the cooling tower fan frequency and activates more fans to enhance heat dissipation. Then, taking into account current cooling requirements and energy consumption targets, the PLC controller increases the speed of the circulating water pump via the frequency converter, allowing more cooling water to circulate, effectively removing heat generated by the equipment inside the cabinets and ensuring stable operation at an appropriate temperature. During the nighttime off-peak period, when the equipment load decreases, the PLC controller reduces the cooling tower fan frequency and circulating water pump speed based on the newly calculated parameters, reducing energy consumption.

[0069] By real-time acquisition and analysis of multi-parameter data, the cooling tower and circulating water pump are dynamically adjusted according to the actual cooling demand, which can accurately meet the heat dissipation requirements of the liquid cooling cabinet under different working conditions, effectively avoid equipment overheating damage caused by insufficient cooling, and ensure the stable operation and service life of the equipment. Abandoning the traditional fixed operation mode, the adjustment is based on the optimization target of heat exchange efficiency and energy consumption, which meets the cooling demand while avoiding energy waste. For example, during the low load period, the operation intensity of the cooling tower fan and water pump is reduced, the power consumption is reduced, and the operation cost of the data center is reduced, which meets the development trend of green energy saving. According to the changes of environmental temperature, equipment load and other factors, the running state of the cooling system can be automatically adjusted, so that the liquid cooling cabinet can always maintain good working performance in complex and variable environment, and the stability of the whole cooling system is enhanced. PLC controller and algorithm are used for data processing and decision-making, without frequent manual intervention, realizing the intelligentization of the whole process from data acquisition, analysis to control adjustment of the cooling system, improving the operation and maintenance efficiency, and reducing the labor operation cost and operation error risk.

[0070] In another preferred embodiment of the present application, the cooling water temperature, pressure and flow data are input into the PLC controller, and the cooling tower heat exchange efficiency is calculated by the approximation degree algorithm combined with the difference between the cooling tower outlet water temperature and the outdoor wet bulb temperature, which can include:

[0071] The PLC controller extracts the cooling tower outlet water temperature and outdoor wet bulb temperature data from the storage area to obtain the difference between the cooling tower outlet water temperature and outdoor wet bulb temperature data;

[0072] The PLC controller starts the approximation degree algorithm, compares the temperature difference with the preset difference range, and obtains the comparison result;

[0073] According to the comparison result, and combined with the temperature, pressure and flow data of the cooling water, if the temperature difference is within the preset difference range, the potential influence of parameter fluctuation on the efficiency is analyzed; if the temperature difference is greater than or equal to the preset difference range, the abnormal parameter is determined, and finally the heat exchange efficiency of the cooling tower is obtained.

[0074] In the embodiment of the present application, in the whole liquid cooling system, there are special sensors responsible for real-time acquisition of cooling tower outlet water temperature and outdoor wet bulb temperature. These sensors transmit the collected data to the data storage area, which can be the internal memory of the PLC controller. The PLC controller reads the latest cooling tower outlet water temperature T out and outdoor wet bulb temperature T wb data from the storage area through the preset communication protocol and interface. The calculation unit inside the PLC controller will perform subtraction operation on the two temperature data to obtain the difference ΔT=T out -T wb .

[0075] The approximation algorithm can measure the closeness between an actual value and a preset range. The PLC controller will calculate the temperature difference ΔT and the preset difference range [ΔT min ,ΔT max ] for comparison. The PLC controller will calculate the approximation degree of ΔT relative to the preset difference range according to the rules of the approximation algorithm. For example, using the interval judgment method, when ΔT min ≤ΔT≤ΔT max When ΔT<ΔT min or ΔT>ΔT max When the temperature difference exceeds the preset range, the comparison result can be represented by a logical value (such as "within the range" is true and "out of range" is false).

[0076] When the temperature difference is within the preset range, while the cooling tower is operating normally, fluctuations in cooling water parameters such as temperature, pressure, and flow rate may still potentially impact heat exchange efficiency. The PLC controller monitors and analyzes these parameters in real time, comparing them with historical data to assess the impact of each parameter's fluctuations and changing trends on heat exchange efficiency. For example, a sudden increase in cooling water flow rate may shorten heat exchange time, impacting heat exchange efficiency; an increase in cooling water temperature can also negatively impact heat exchange performance. When the temperature difference exceeds the preset range, it indicates a possible cooling tower anomaly. The PLC controller conducts in-depth analysis of cooling water temperature, pressure, and flow rate data, using data mining and fault diagnosis algorithms to determine which parameters are anomalies. For example, an abnormally high cooling water temperature may indicate insufficient cooling tower capacity or a blockage in the cooling water piping. Abnormal pressure fluctuations could indicate a pump failure or a leak in the piping. After determining the abnormal parameters, the PLC controller first retrieves the operating data of the cooling tower under normal operating conditions from the database, including the cooling water inlet temperature, outlet temperature, flow rate, pressure, and corresponding outdoor wet-bulb temperature in different time periods. At the same time, it calls out the pre-established heat exchange efficiency model under normal conditions. This model is a relationship model between various parameters and heat exchange efficiency based on a large amount of historical data and actual tests. For the identified abnormal parameters, such as an abnormal increase in cooling water temperature, the PLC controller will analyze the impact mechanism of the abnormal parameter on the heat exchange process. For example, if the cooling water temperature increases, it means that the heat entering the cooling tower increases. Under the same heat dissipation conditions, the cooling tower needs to dissipate heat more efficiently to maintain a normal cooling effect. The PLC controller will combine the physical principles of heat exchange and historical experience to evaluate the possible impact of the abnormal parameter on the heat exchange efficiency.

[0077] Based on the impact analysis of abnormal parameters, the PLC controller adjusts the parameters of the heat exchange efficiency model under normal conditions. For example, if an abnormally low cooling water flow rate is detected, resulting in a prolonged heat exchange time and a potential change in the heat transfer coefficient, the PLC controller will adjust the model's parameters related to the heat transfer coefficient accordingly to better reflect the actual conditions under the abnormal operating conditions. The current operating data, including the abnormal parameter values ​​and other normal parameter data, collected in real time, is substituted into the adjusted heat exchange efficiency model. The model's calculation logic gradually calculates the cooling tower's actual heat exchange efficiency under the current abnormal operating conditions. After calculating the preliminary heat exchange efficiency, the PLC controller compares the result with other relevant monitoring data for verification. For example, it compares the operating status data of other related equipment during the same time period, or references historical heat exchange efficiency data from similar abnormal conditions, to determine the validity of the calculated result. If any deviation is detected, the PLC controller will re-examine the abnormal parameter analysis and model adjustment steps, make corrections, and then recalculate until an accurate and reliable cooling tower heat exchange efficiency is obtained.

[0078] Assume that in a data center’s liquid cooling system, the difference between the cooling tower outlet water temperature and the outdoor wet-bulb temperature is pre-set by the PLC controller to be [5°C, 10°C]. At a certain moment, the PLC controller extracts the cooling tower outlet water temperature T from the storage area. out =28℃, outdoor wet bulb temperature T wb = 20°C, the calculated temperature difference ΔT = 28 - 20 = 8°C. Since 8°C is within the preset differential range [5°C, 10°C], the PLC controller further analyzes the cooling water temperature, pressure, and flow data. It then discovers that the cooling water flow has been increasing over time. Analyzing historical data and the heat exchange efficiency model, it predicts that this increase in flow may reduce heat exchange efficiency by approximately 5%.

[0079] Assume that at another moment, the cooling tower outlet water temperature is T out =35℃, outdoor wet bulb temperature T wb =20°C, the calculated temperature difference ΔT = 35-20 = 15°C, which exceeded the preset range. The PLC controller analyzed the cooling water temperature, pressure, and flow data and discovered an abnormally high temperature and a slight drop in pressure. Further diagnosis determined that a malfunction in the cooling tower's heat dissipation fan was causing insufficient heat dissipation. Based on this abnormal information, the PLC controller modified the heat exchange efficiency model and calculated that the cooling tower's current heat exchange efficiency was only 60% of normal.

[0080] By combining a comprehensive analysis of the difference between the cooling tower's outlet water temperature and the outdoor wet-bulb temperature, along with various data such as cooling water temperature, pressure, and flow, the actual operating status of the cooling tower can be more comprehensively and accurately reflected, leading to a more precise calculation of heat exchange efficiency. This helps operations and maintenance personnel promptly understand the cooling tower's performance and provides a reliable basis for optimizing the cooling system. When the temperature difference exceeds the preset range, abnormal parameters can be quickly located, allowing potential faults in the cooling tower's operation to be promptly identified. For example, by analyzing cooling water parameters, problems such as water pump failure, pipe blockage, and cooling fan failure can be detected in advance, allowing timely maintenance measures to be taken to prevent further escalation of the problem and reduce downtime and repair costs.

[0081] When the temperature difference is within a preset range, analyzing the potential impact of various parameter fluctuations on efficiency helps proactively implement adjustments and optimization measures. For example, the operating frequency of the water pump can be adjusted based on flow fluctuations, and the cooling tower's heat dissipation capacity can be adjusted based on temperature changes. This improves the efficiency and stability of the entire cooling system and reduces energy consumption. The entire process is automatically completed by the PLC controller, which utilizes approximation algorithms and data analysis techniques to achieve intelligent calculation and monitoring of the cooling tower's heat exchange efficiency. This reduces manual intervention, improves the real-time and accuracy of monitoring, and enhances the management level of the data center's liquid cooling system.

[0082] In a preferred embodiment of the present invention, real-time monitoring of the secondary side piping status and cabinet environment through the cooling distribution unit and the distributed sensor network may include:

[0083] Install an array of optical fiber temperature sensors on the secondary side pipeline, distribute them along the pipeline to form a temperature field monitoring network, and collect temperature data at different locations on the secondary side pipeline in real time;

[0084] Temperature anomaly data is extracted from the temperature data at different locations on the secondary side pipeline. At the same time, the pressure differential sensor in the cooling distribution unit monitors the secondary side inlet and outlet pressure differential data in real time. When the secondary side inlet and outlet pressure differential is ≥ the preset range and the change trend meets the characteristics of pipeline blockage, the pipeline is assessed to have a blockage risk based on the preset threshold and logical judgment rules, and an early warning signal is issued;

[0085] After receiving the pipeline blockage risk warning signal, the humidity sensor array and temperature and humidity sensor inside the cabinet simultaneously collect cabinet environmental parameters; the humidity sensor array detects changes in environmental humidity inside the cabinet, and the temperature and humidity sensor collects environmental temperature and humidity data.

[0086] In an embodiment of the present invention, during the planning and design phase of the secondary cooling pipeline, the installation position of the fiber optic temperature sensor array is determined based on the direction, length, and key node distribution of the pipeline. Through the installation process, the fiber optic temperature sensor is tightly attached to the surface of the pipeline to ensure that the sensor can accurately sense the temperature changes of the pipeline. These sensors are distributed along the pipeline at a certain interval (such as one installed every 1 meter) to form a temperature field monitoring network covering the entire secondary pipeline. When the sensor is working, it uses the optical time domain reflectometry (OTDR) principle of optical fiber to monitor and record the temperature information of each position in real time. The sensor converts the temperature signal into an optical signal and transmits it to the data acquisition module via optical fiber. The data acquisition module parses and converts the optical signal, converts it into digital temperature data, and transmits the data to the central processing unit of the monitoring system at a set frequency (such as once per second), thereby realizing real-time acquisition of the secondary pipeline temperature.

[0087] After receiving the temperature data of each position of the secondary side pipeline, the central processing unit will first store the data in a specific cache area for subsequent processing and analysis. The preset normal temperature range (for example, 15℃-25℃) and abnormal judgment rules (such as the temperature suddenly rises or falls by more than 5℃) are stored in the configuration file of the central processing unit. The central processing unit will read the temperature data in the cache in chronological order and compare the real-time temperature of each position with the preset normal range. For each temperature data point, it will be checked whether it falls within the range of 15℃-25℃. If it is not within the range, the central processing unit will further calculate the difference between the temperature and the temperature of the previous monitoring cycle.

[0088] To more accurately determine whether temperature changes are abnormal, the central processing unit maintains a temperature change history, recording temperature changes at each location over a period of time. By analyzing this history and combining it with the current temperature difference, it determines whether the abnormality judgment rule of "a sudden temperature increase or decrease exceeding 5°C" is met. If so, the temperature data for that location is marked as abnormal and stored in a dedicated abnormal data list. The differential pressure sensor installed in the cooling distribution unit uses a high-precision pressure sensing element to accurately measure the pressure difference between the inlet and outlet of the secondary pipe in real time. The differential pressure sensor converts the measured analog pressure difference signal into a digital signal and transmits the data to the central processing unit at regular intervals (for example, once per second) via a specific communication interface (such as RS-485 or CAN bus). After receiving the differential pressure data, the central processing unit performs preliminary verification and processing to ensure its accuracy and integrity. This verification process includes checking the data format for correctness and that the data values ​​are within a reasonable range. If the data passes verification, the central processing unit stores it in a dedicated differential pressure data buffer.

[0089] The central processing unit will periodically read the latest pressure differential data from the pressure differential data cache and compare it with the preset pressure differential range (such as 0.05MPa-0.1MPa). The comparison process uses a simple numerical comparison algorithm to determine whether the current pressure differential data is greater than or equal to the upper limit of the preset range (0.1MPa). At the same time, the central processing unit will use a pattern recognition algorithm to analyze the changing trend of the pressure differential data. The pattern recognition algorithm will match the current pressure differential data sequence with the pre-set pipeline blockage characteristic pattern. The pipeline blockage characteristic pattern can be obtained through the analysis of historical data, such as a pattern in which the pressure differential continues to rise and the rate of increase gradually accelerates. The central processing unit will pre-process the pressure differential data, including data smoothing, differential calculation and other operations to highlight the changing trend of the data. Then, the processed data will be compared with the blockage characteristic pattern to calculate the similarity between them. If the similarity exceeds a certain threshold (for example, 80%), it is considered that the changing trend of the pressure differential data is consistent with the pipeline blockage characteristics.

[0090] When the central processing unit determines that the differential pressure data is greater than or equal to a preset range and its trend matches pre-defined characteristics of pipeline blockage, it activates pre-set thresholds and logical judgment rules for further judgment. These thresholds and logical judgment rules (e.g., if the differential pressure exceeds the range and the trend matches the characteristics of pipeline blockage for three consecutive monitoring cycles) are stored in the central processing unit's rule engine. The central processing unit then checks whether the differential pressure data for the last three consecutive monitoring cycles has exceeded the preset range and whether the trend matches the characteristics of pipeline blockage. If these conditions are met, the central processing unit determines that there is a risk of pipeline blockage. Once a blockage risk is determined, the central processing unit triggers the early warning signal generation module. The early warning signal generation module generates an early warning signal based on a pre-set early warning template, containing detailed information such as pipeline location, current differential pressure data, and temperature anomaly data. The early warning signal is sent to operations and maintenance personnel via various means, such as text messages, emails, and system pop-up notifications, ensuring that they receive timely information and take appropriate measures. Upon receiving the pipeline blockage risk warning signal, the monitoring system's early warning receiving module immediately sends data collection instructions to the humidity sensor array and temperature and humidity sensors within the cabinet. The humidity sensor array consists of multiple humidity sensors located in various locations within the cabinet (e.g., top, bottom, and sides), enabling comprehensive monitoring of humidity changes in different areas within the cabinet. The temperature and humidity sensor simultaneously measures both ambient temperature and humidity. Upon receiving a command, the humidity sensor array and temperature and humidity sensors quickly initiate data collection, collecting humidity, temperature, and humidity parameters within the cabinet at a set frequency (e.g., every 5 seconds).

[0091] In a large data center's liquid cooling system, the secondary cooling pipes are crucial for dissipating heat for multiple cabinets. One afternoon, due to long-term use, impurities accumulated in a section of the pipe. The fiber optic temperature sensor array detected a rapid rise in temperature at that location from a normal 20°C to 28°C, exceeding the preset normal temperature range. The system then detected this abnormal temperature.

[0092] At the same time, the pressure differential sensor within the cooling distribution unit detected that the secondary inlet and outlet pressure differential began to rise continuously from the normal 0.08 MPa, rising to 0.15 MPa within 30 minutes, exceeding the preset pressure differential range of 0.1 MPa. The pressure differential rise was also gradually accelerating, consistent with the trend of pipeline blockage. Based on preset thresholds and logical judgment rules, the system determined that there was a risk of pipeline blockage and issued a warning signal.

[0093] Upon receiving the warning signal, the monitoring system immediately activated the humidity sensor array and temperature and humidity sensors inside the cabinet. Data collected revealed that the previously relatively stable humidity inside the cabinet had rapidly risen from 45% RH to 60% RH, and the temperature had also increased from 22°C to 25°C. This data indicated that the pipe blockage had impacted the cabinet environment, providing critical information for operations and maintenance personnel to further assess the impact of the fault and develop a repair plan.

[0094] A fiber optic temperature sensor array forms a temperature field monitoring network, enabling comprehensive, high-precision temperature monitoring of the secondary piping. This allows for timely detection of localized temperature anomalies, enabling earlier detection of potential faults compared to traditional single-point temperature monitoring. Furthermore, combined with real-time monitoring of the inlet and outlet pressure differential by a differential pressure sensor, the system comprehensively assesses the pipeline status from both temperature and pressure perspectives, enhancing the comprehensiveness and accuracy of monitoring. Based on preset thresholds and logical judgment rules, it issues an early warning signal at the first sign of a pipeline blockage, enabling operators to take pre-emptive action to prevent serious issues such as insufficient coolant flow and overheating of cabinet equipment caused by a blockage, effectively reducing the risk of equipment damage and data center downtime. Upon detecting a blockage risk, timely collection of internal cabinet humidity and temperature data helps assess the impact of the fault on the cabinet environment. By monitoring environmental parameters in real time, operators can implement targeted measures (such as increasing ventilation and adjusting the load on other cooling pipes) to maintain stable temperature and humidity within the cabinet, ensuring a favorable environment for normal equipment operation and extending equipment life. Rich monitoring data (temperature, pressure difference, humidity, temperature and humidity) provides multi-dimensional information for fault diagnosis, helping operation and maintenance personnel to more accurately determine the cause and location of the fault, formulate scientific and reasonable maintenance plans, improve fault handling efficiency, and reduce fault handling time and cost.

[0095] In a preferred embodiment of the present invention, leak location is achieved by using a built-in leak sensor, a temperature sensor, and a pressure sensor, and automatic liquid replenishment is performed using a liquid storage tank, which may include:

[0096] A liquid collection tray is set at the bottom of the cooling distribution unit, and a leakage detection cable is laid inside the tray. When the coolant leaks, an impedance change signal is triggered, and the impedance change signal is cross-verified with the signal sent by the leakage sensor;

[0097] After confirming the coolant leak, the leak point is located through the temperature field distortion characteristics of the distributed optical fiber temperature sensor, and the spatial position information of the leak point is spatially matched with the leakage signal;

[0098] When it is detected that the liquid level in the liquid storage tank is less than the preset threshold, the refill pump is immediately started to inject coolant into the cooling circulation loop.

[0099] In an embodiment of the present invention, during the design and installation phase of a cooling distribution unit (CDU), a collection pan is installed at its base. The pan is made of corrosion-resistant, high-strength materials to ensure it can collect leaked coolant without being damaged by corrosion. Leak detection cables are arranged within the pan in a specific layout. The cables are typically made of a conductive material with a special coating. Under normal conditions, the cables are dry and their impedance remains stable at its initial value. When coolant leaks and drips into the pan and comes into contact with the leak detection cables, the special coating on the cables undergoes physical or chemical changes upon contact with the coolant, causing the cable's impedance to rapidly change, generating an impedance change signal. This signal is transmitted to a signal processing module via a dedicated data transmission line (e.g., a shielded cable). Simultaneously, independent leak sensors (e.g., electronic or optical) installed at key locations within the CDU (e.g., at pipe connections or near valves) also generate and transmit a leak signal to the signal processing module upon detecting a coolant leak. After receiving the impedance change signal and the leak sensor signal, the signal processing module compares and analyzes the two signals' characteristics, such as their occurrence time and change trends. A coolant leak is only confirmed when the two signals are close in time and their change characteristics meet preset cross-validation rules (for example, when the impedance change amplitude reaches a certain threshold and the leak sensor output signal strength exceeds a standard value), thereby reducing false positives.

[0100] In the secondary cooling pipeline system, distributed fiber optic temperature sensors are pre-installed along the pipeline. Fiber optic sensing technology enables continuous monitoring of the temperature of the entire pipeline. When a coolant leak is confirmed, the central processing unit retrieves the real-time temperature data collected by the distributed fiber optic temperature sensors to construct a temperature field distribution model for the pipeline. Since coolant leakage will carry away the surrounding heat, the temperature near the leak point will change abnormally, resulting in a temperature field distortion compared to the temperature field during normal operation. The central processing unit processes the temperature field distribution model through data analysis algorithms (such as temperature gradient analysis algorithms and thermal imaging inversion algorithms), identifies areas where the temperature has dropped abnormally, and preliminarily determines the approximate location of the leak point based on the severity and range of the temperature change. In order to further accurately locate the leak point, the central processing unit combines the spatial position information of the fiber optic temperature sensor (the coordinates of each sensor node have been accurately recorded during installation) to spatially match the determined leak point location information with the previously confirmed leakage signal. By comparing the spatial coordinates of the two, we ensure that the located leak point corresponds to the actual location of the leakage, and ultimately obtain the accurate spatial location information of the leak point, which is displayed on the monitoring system interface to provide clear guidance for maintenance personnel.

[0101] A liquid level sensor (such as an ultrasonic liquid level sensor or a float-type liquid level sensor) is installed inside the liquid storage tank. The liquid level sensor monitors the liquid level of the coolant in the liquid storage tank in real time and transmits the liquid level data in the form of an electrical signal or a digital signal through a communication line (such as an RS-485 bus or Ethernet) to the liquid level monitoring and control module. A preset threshold is pre-set in the liquid level monitoring and control module (the threshold is determined based on the normal operating requirements and safety margin of the cooling cycle), and the real-time liquid level data received is continuously compared with the preset threshold. When the liquid level in the liquid storage tank is detected to be lower than the preset threshold, the liquid level monitoring and control module immediately issues a start command, which is transmitted to the drive device of the refill pump through the control line. After receiving the start command, the refill pump drives the motor to start extracting coolant from the liquid storage tank and injecting the coolant into the cooling cycle through a dedicated refill pipeline. During the refill process, the liquid level monitoring and control module continuously monitors the liquid level in the storage tank. When the liquid level rises to or exceeds the preset stop refill threshold (usually slightly higher than the preset threshold to ensure sufficient coolant reserves), the liquid level monitoring and control module sends a stop command, turns off the refill pump, and stops the refill operation, thereby maintaining the normal liquid level of the coolant in the cooling circulation loop and ensuring stable operation of the system.

[0102] During operation of a liquid cooling system in a large data center, a microcrack developed at a weld point in a secondary pipe, causing a slow coolant leak. The leaked coolant dripped into the collection tray at the bottom of the cooling distribution unit and came into contact with a leak detection cable installed inside the tray. This caused a change in the cable's impedance, generating an impedance change signal that was transmitted to the signal processing module. Simultaneously, a leak sensor located near the pipe weld also detected the coolant, generating a leak signal and transmitting it. The signal processing module cross-validated the two signals and confirmed a coolant leak. The central processing unit then retrieved data from the distributed fiber optic temperature sensors and discovered an abnormal temperature field in the corresponding pipe area. The temperature near the weld point was significantly lower than at other locations. Using algorithmic analysis and spatial matching, the leak point was precisely located and marked on the monitoring system interface. As the coolant continued to leak, the liquid level in the reservoir tank continued to drop. When the level sensor detected that the level fell below a preset threshold, the level monitoring and control module quickly activated the refill pump to inject coolant from the reservoir into the cooling circuit. During the refill process, the liquid level continues to rise. When the refill stop threshold is reached, the refill pump stops working, effectively avoiding cooling system failures caused by insufficient coolant and ensuring the normal operation of data center equipment.

[0103] Through cross-validation of the leakage detection cable and the leakage sensor, the accuracy of leakage detection has been effectively improved and false alarms have been reduced. By locating the leakage point using the temperature field distortion characteristics of the distributed optical fiber temperature sensor, it is possible to quickly and accurately locate the leakage point. Compared with the traditional manual troubleshooting method, it greatly shortens the fault location time, improves the fault handling efficiency, and reduces the damage to the equipment caused by coolant leakage. The liquid level monitoring and automatic refilling function of the liquid storage tank realize the automatic adjustment of the coolant level in the cooling circulation loop. When the liquid level is lower than the preset threshold, the refill pump is automatically started for refilling to ensure that the system always maintains a sufficient amount of coolant, avoiding problems such as reduced cooling effect and equipment overheating due to insufficient coolant, improving the stability and reliability of the liquid cooling system operation, reducing the workload of manual inspections and manual refilling, and reducing operation and maintenance costs. The leak location and automatic refilling mechanism forms a complete protection chain from leak detection and location to coolant replenishment. It can promptly detect and deal with coolant leaks, prevent secondary failures such as short circuits and corrosion caused by leaks, protect the safety of liquid-cooled cabinets and internal equipment, provide strong guarantees for the stable operation of key facilities such as data centers, and enhance the safety and risk resistance of the entire liquid cooling system.

[0104] In a preferred embodiment of the present invention, multi-source data, including cooling water temperature, pressure, flow parameters, leakage signals, humidity gradient, and pressure differential data, are integrated to calculate dynamic detection thresholds. Graded alarms are triggered based on the synergistic relationship between persistent negative flow deviation, asymmetric temperature field distortion, sudden humidity gradient changes, and pressure differential spectrum feature matching. These may include:

[0105] The temperature, pressure, and flow data of the primary side cooling water, the secondary side leakage signal, the humidity gradient data in the cabinet, and the pressure difference data of the cooling distribution unit are input into the data fusion unit. The data of the data fusion unit is preprocessed through time series to generate a denoised time series data set.

[0106] Based on the denoised time series data set, the covariance between parameters is calculated, periodic fluctuations are detected, and abnormal patterns are identified. Combined with the correlation characteristics of normal operating conditions in historical operating data, an adaptive dynamic threshold for leak detection is calculated.

[0107] The dynamic threshold for leak detection is adjusted based on the real-time data stream. When the cooling water flow rate has a continuous negative deviation ≥ the dynamic threshold for leak detection and the distributed fiber optic temperature sensor network detects asymmetric distortion of the temperature field, it is determined to be an early leakage risk and an early warning signal is triggered. If, on the basis of triggering the early warning signal, it is also detected that the humidity difference between adjacent sensors in the cabinet is ≥ the set threshold and the pressure difference spectrum shows leakage characteristics, it will be upgraded to a high-confidence leak alarm.

[0108] In an embodiment of the present invention, in an intelligent monitoring system for liquid-cooled cabinets, different types of sensors are responsible for collecting various data. The primary-side cooling water temperature sensor, pressure sensor, and flow sensor respectively capture temperature, pressure, and flow data; the secondary-side leakage sensor generates a leakage signal; multiple humidity sensors distributed throughout the cabinet monitor humidity gradient data; and the differential pressure sensor in the cooling distribution unit collects differential pressure data. This data is input into a data fusion unit via various data transmission channels, such as an RS-485 bus or Ethernet. After receiving the multi-source data, the data fusion unit arranges the data in chronological order to form a time series. Because the collected data may contain noise, such as sensor measurement errors and electromagnetic interference, denoising is required. Common denoising methods include moving average and wavelet transform. For example, for time series data of a parameter, the average value of the data within a certain time window is taken as the new data value at that time point, thereby smoothing the data and reducing the impact of noise. After denoising, a denoised time series dataset is generated.

[0109] Based on the denoised time series dataset, the covariance between each parameter is calculated. Covariance measures the strength and direction of the linear relationship between two parameters. For example, there may be a certain correlation between cooling water temperature and flow rate. By calculating the covariance, we can understand whether their changing trends are consistent. At the same time, methods such as spectral analysis are used to detect periodic fluctuations in the data. For example, the operation of certain equipment may cause periodic changes in the data. In addition, machine learning algorithms (such as cluster analysis) are used to identify abnormal patterns in the data, such as sudden parameter jumps. Historical operating data is retrieved to analyze the correlation characteristics between various parameters under normal operating conditions. For example, during normal operation, there is a certain proportional relationship between cooling water temperature, pressure, and flow rate. The analysis results of the current data are compared with the correlation characteristics of historical normal operating conditions. Taking into account the changes in various parameters, an adaptive dynamic threshold for leak detection is calculated. This threshold is not fixed but dynamically adjusted based on the current system operating status and data characteristics to adapt to different operating conditions and environmental changes.

[0110] The data fusion unit will continuously receive real-time data streams and adjust the dynamic threshold for leak detection in real time based on these data. When the cooling water flow rate is monitored to have a continuous negative deviation ≥ the dynamic threshold for leak detection, it indicates that the cooling water flow rate has decreased abnormally. At the same time, the distributed fiber optic temperature sensor network will monitor the distribution of the temperature field in real time. If asymmetric distortion of the temperature field is detected, that is, the temperature changes in certain areas are significantly different from normal, it may be due to a coolant leak that causes a change in the local cooling effect. When these two conditions are met at the same time, it is determined to be an early leakage risk and an early warning signal is triggered. The early warning signal can be sent to operation and maintenance personnel through sound and light alarms, SMS notifications, etc., to remind them to pay attention to the operating status of the system.

[0111] If, after triggering an early warning signal, the humidity difference between adjacent sensors within the cabinet is greater than or equal to the set threshold, this indicates a significant humidity anomaly within the cabinet, possibly due to coolant leaking into the cabinet and causing the humidity to increase. Simultaneously, the differential pressure spectrum is analyzed. If the spectral characteristics match pre-defined leak signatures, such as fluctuations in specific frequencies or changes in the spectral shape, this indicates a high-confidence leak in the system. At this point, the alarm level is upgraded to a high-confidence leak alert, alerting operations and maintenance personnel to take prompt action.

[0112] Assume that the following situation occurs during the operation of the liquid cooling cabinet intelligent monitoring system in a data center:

[0113] The temperature sensor, pressure sensor and flow sensor of the primary side cooling water respectively collected the temperature of 25℃, pressure of 0.5MPa and flow of 10m 3 / h data; the secondary side leakage sensor did not detect a leakage signal; the humidity sensor in the cabinet detected humidity gradient data; the pressure difference sensor of the cooling distribution unit collected a pressure difference of 0.1MPa. These data were input into the data fusion unit, and after denoising by the moving average method, a denoised time series data set was generated. By calculating the covariance between the parameters, it was found that there was a certain negative correlation between the temperature and flow of the cooling water. At the same time, some periodic fluctuations were detected in the data, but no obvious abnormal patterns were found. Combined with the historical operation data, the current dynamic threshold for leakage detection is calculated to be 1m 3 / h. As time went on, the cooling water flow rate gradually decreased. After a period of monitoring, it was found that the flow rate continued to have a negative deviation of 1.2m 3 / h, exceeding the dynamic threshold for leak detection. Simultaneously, the distributed fiber optic temperature sensor network detected asymmetric distortion in the temperature field, with temperatures in some areas significantly elevated. This identified an early leak risk, triggering an early warning signal and sending a text message notification to operations and maintenance personnel.

[0114] After the early warning signal was triggered, continued monitoring revealed that the humidity difference between adjacent sensors within the cabinet reached 10% RH, exceeding the set threshold of 5% RH. Analysis of the differential pressure spectrum revealed spectral characteristics consistent with a leak, with specific frequency fluctuations. Consequently, the alarm level was upgraded to a high-confidence leak alert, and operations and maintenance personnel quickly arrived at the site for inspection. They discovered a leak in the secondary-side piping and promptly repaired it.

[0115] By integrating multi-source data and comprehensively considering multiple parameters such as cooling water temperature, pressure, flow, leakage signal, humidity gradient, and pressure difference, the operating status of the liquid cooling system can be reflected more comprehensively and accurately. By using a dynamic threshold calculation method, combined with historical data and real-time data streams, the detection threshold can adapt to changes in the system's operation, avoiding the misjudgment and missed judgment problems that may occur with fixed thresholds under different operating conditions, and improving the accuracy of leakage fault detection. Based on the collaborative judgment of multiple features such as continuous negative deviation of flow and asymmetric distortion of temperature field, it is possible to detect abnormal conditions in the early stages of leakage faults and trigger early warning signals. This allows operation and maintenance personnel to take measures before the fault expands, reduce the damage to equipment caused by coolant leakage, reduce maintenance costs and downtime, and ensure the stable operation of critical facilities such as data centers.

[0116] The hierarchical alarm mechanism categorizes faults according to their severity and confidence. Early warning signals alert operators to potential system issues, while high-confidence leak alarms indicate a more severe fault requiring immediate attention. This tiered approach helps operators rationally allocate resources, prioritize urgent and critical faults, and improve the efficiency and effectiveness of O&M management. Dynamic threshold calculation and adjustment adapt to varying operating conditions and environmental changes, ensuring the monitoring system maintains optimal performance in all situations. Furthermore, the hierarchical alarm mechanism enables timely detection and resolution of faults, minimizing their impact on the system. This enhances the adaptability and stability of the liquid cooling cabinet intelligent monitoring system, providing strong support for the long-term stable operation of facilities such as data centers.

[0117] In another preferred embodiment of the present invention, based on the denoised time series data set, the covariance between parameters is calculated, periodic fluctuations are detected, and abnormal patterns are identified. In combination with the correlation characteristics of normal operating conditions in the historical operating data, an adaptive dynamic threshold for leak detection is calculated, which may include:

[0118] Collect multi-source data, including primary-side cooling water temperature, pressure, and flow data, secondary-side leakage signals, cabinet humidity gradient data, and cooling distribution unit pressure difference data;

[0119] For any two parameters in multi-source data, the covariance between the two parameters is calculated to obtain the weighted average of the covariance between the parameters;

[0120] For each parameter in the multi-source data, the periodic characteristics of the parameter are determined by analyzing the value change pattern at different time points. Based on the periodic characteristics, the theoretical value of the corresponding parameter at each time point is predicted to determine the periodic fluctuation amplitude of the corresponding parameter. Based on the periodic fluctuation amplitude of each parameter, the weighted average value of the parameter periodic fluctuation is determined.

[0121] Analyze multi-source data and use preset anomaly judgment rules to identify different anomaly patterns. For each anomaly pattern, count the number of anomaly points in the multi-source data that match the anomaly pattern and record the total number of data points. Determine the severity of the anomaly pattern based on the number of anomaly points and the total number of data points for each anomaly pattern, and obtain a weighted average of the severity of the anomaly pattern.

[0122] Extract the correlation features of each parameter under normal operating conditions from historical operating data and convert the correlation features of each parameter into a single characteristic value;

[0123] An adaptive dynamic threshold for leak detection is determined based on the weighted average of covariances between parameters, the weighted average of periodic fluctuations of parameters, the weighted average of severity of abnormal patterns, and a single characteristic value of historical operating data.

[0124] In this embodiment, we first collect multi-source data, including primary cooling water temperature, pressure, and flow rate data, secondary leakage signals, cabinet humidity gradient data, and differential pressure data for the cooling distribution unit. To ensure the accuracy of subsequent calculations, this data needs to be denoised to form a denoised time series dataset.

[0125] For any two parameters i and j in multi-source data, Calculate the covariance between them. Where n is the number of parameters, m is the length of each parameter time series data, is the average value of the time series data of the i-th parameter, x jk is the value of the jth parameter at the kth time point, is the average value of the time series data of the jth parameter, x ik is the value of the i-th parameter at the k-th time point, is the average value of the time series data of the i-th parameter. For each parameter, assign weight w to i and j. ij , the weight range is between 0 and 1. The weight is determined based on the correlation and importance between the parameters. For example, if the relationship between temperature and flow is very critical for leak detection, then their corresponding w ij It can be set higher, such as 0.8-1; if the relationship between some parameters is relatively minor, the weight can be set to 0.1-0.3. Divide by the total weight Get the weighted average of the covariances between parameters.

[0126] For each parameter in the multi-source data, analyze its value change pattern at different time points to determine the periodic characteristics of the parameter. Methods such as Fourier transform can be used to identify periodicity. Based on the periodic characteristics, the theoretical value of the corresponding parameter at each time point is predicted. Calculate the maximum absolute value of the difference between the actual value and the theoretical value of each parameter This is used as the periodic fluctuation amplitude of the parameter, x it is the actual value of the i-th parameter at the t-th time point. Give each parameter a weight v i , the weight range is between 0 and 1. The weight is determined according to the degree of influence of the parameter on leak detection. For example, if the periodic fluctuation of the flow parameter has a greater impact on leak detection, v i It can be set to 0.7-0.9; for parameters with less influence, the weight can be set to 0.1-0.3. Divide by the total weight Get the weighted average of the periodic fluctuations of the parameters.

[0127] Using the preset anomaly judgment rules, we analyze the multi-source data and identify different anomaly patterns. For example, when the temperature suddenly rises above a certain threshold, or when the flow rate continues to drop, it can be defined as an anomaly pattern. For each anomaly pattern i, we count the number of anomalies a in the multi-source data that match the anomaly pattern. i , while recording the total number of data points N. The severity of the abnormal pattern is calculated as Assign a weight u to each abnormal pattern i The weight range is between 0 and 1. The weight is determined based on the importance of the abnormal pattern to leak detection. For example, if an abnormal pattern is likely to indicate a leak, its weight can be set to 0.8-1; for abnormal patterns that are less certain to be related to leaks, the weight can be set to 0.1-0.3. Divide by the total weight The weighted average of the severity of the abnormal patterns is obtained, where s is the number of types of abnormal patterns.

[0128] Filter data from massive amounts of historical data to identify data from liquid-cooled cabinets operating under normal conditions. This requires excluding data from abnormal operating conditions, such as equipment failures, maintenance, and debugging, based on equipment operation records and fault logs. Analyze the correlations between parameters within this filtered data. For example, observe whether pressure and flow change in specific proportions as the primary cooling water temperature increases over a period of time. Also, examine the changing trends of secondary leakage signals, humidity gradient data within the cabinet, and differential pressure data across the cooling capacity distribution unit over different time periods to determine whether these trends change in the same or opposite directions. Graphs (such as line charts and scatter plots) provide a visual representation of the correlations between parameters and aid analysis. Principal component analysis (PCA) aims to transform numerous interrelated parameters into a small number of uncorrelated principal components while preserving as much information as possible from the original data. This process is like untangling a tangled web of lines and simplifying complex parameter relationships. First, perform data preprocessing to standardize the data dimensions (for example, converting data with different units, such as temperature, pressure, and flow, into comparable units). PCA then identifies the most dominant directions of variation in the data, which are known as principal components. Each principal component is a linear combination of the original parameters, explaining a portion of the information in the original data. Based on the principal components obtained by principal component analysis and their degree of explanation of the original data, these principal components are combined according to specific rules, much like combining gemstones of varying value into a piece of jewelry, ultimately yielding a single eigenvalue, H.

[0129] The calculation results of each part are weighted and summed according to the weight coefficient to obtain the adaptive leakage detection dynamic threshold T d , Among them, α, β, γ, and δ are weight coefficients, and α+β+γ+δ=1.

[0130] The α range is 0.2-0.4, because the covariance between parameters reflects the relationship between parameters and has a certain importance for the overall judgment, but it is not the only determining factor.

[0131] The β range is 0.2-0.3. The periodic fluctuation of the parameter can reflect the stability of the system, but its importance is slightly lower than the covariance.

[0132] The γ range is 0.3-0.5. The severity of the abnormal pattern directly reflects the severity of the abnormal situation in the current data and is critical for leak detection.

[0133] The δ range is 0.1-0.2. The historical operating data provides a reference for normal operating conditions, but its weight is relatively small because the system may change.

[0134] This system comprehensively considers multiple aspects of multi-source data, including inter-parameter covariance, cyclical fluctuations in parameters, the severity of abnormal patterns, and the correlation characteristics of historical operating data. Through comprehensive analysis and weighted summation of these factors, it can more comprehensively and accurately reflect the system's operating status, thereby improving the accuracy of leak detection and reducing false positives and missed detections. Because the threshold is dynamically calculated based on real-time multi-source data, it can adapt to various changes during system operation, such as changes in ambient temperature and equipment load. As the system's operating status changes, the dynamic threshold adjusts accordingly, ensuring effective leak detection under different operating conditions.

[0135] By weight coefficient w ij 、v i 、u i α, β, γ, and δ take into account the impact of different parameters, abnormal patterns, and components on the final threshold. These weights can be adjusted based on actual conditions to ensure the calculated threshold better meets the system's actual needs, improving the relevance and effectiveness of detection. The associated features of historical operating data are converted into single feature values ​​and incorporated into the threshold calculation. Historical data contains information about the system's operation under normal conditions. By referring to historical data, we can better determine whether the current system's operating status is normal, further improving the reliability of leak detection.

[0136] like Figure 2 As shown, an embodiment of the present invention also provides a processing device for a liquid cooling cabinet, including:

[0137] Customized cooling tower processing equipment for manufacturing closed cooling towers with integrated variable frequency fans and optimized heat exchange efficiency structures;

[0138] Pipeline processing equipment is used to form primary and secondary circulation pipelines and pre-embed distributed optical fiber temperature sensors and leakage detection cables inside the pipe walls to achieve real-time monitoring of pipeline status;

[0139] The cold distribution unit assembly line is used to produce cold distribution units with integrated double-walled liquid storage tanks. Self-cleaning differential pressure sensors are installed at the inlet and outlet of the cold distribution unit, and an automatic liquid replenishment pump based on liquid level feedback control is integrated;

[0140] The automatic sensor deployment device is used to navigate through lidar, install a humidity gradient detection array on the inner wall of the cabinet, attach temperature sensors to the surface of the cold plate, and automatically complete the wiring of sensor signal cables.

[0141] It should be noted that the processing equipment of the liquid cooling cabinet is the processing equipment of the liquid cooling cabinet corresponding to the above-mentioned system. All implementation methods in the above-mentioned system embodiment are applicable to this embodiment and can achieve the same technical effect.

[0142] An embodiment of the present invention further provides a computing device comprising: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, executes the system described above. All implementations in the above system embodiments are applicable to this embodiment and can achieve the same technical effects.

[0143] The embodiment of the present invention further provides a computer-readable storage medium storing instructions, which, when executed on a computer, causes the computer to execute the system described above. All implementations in the above system embodiments are applicable to this embodiment and can achieve the same technical effects.

[0144] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. Intelligent monitoring system for liquid cooling cabinet, characterized by: include: The cooling source control module is used to dynamically adjust the cooling tower operating status and circulating water pump flow of the primary side circulation according to the cooling water parameters to achieve variable flow control of the cooling water; Dynamic environment monitoring module, used to monitor the secondary side pipeline status and cabinet environment in real time through the cooling distribution unit and distributed sensor network; Comprehensive monitoring submodule, used to locate leaks through built-in leak sensors, temperature sensors, and pressure sensors, and automatically replenish liquid using a liquid storage tank; The intelligent analysis and alarm module is used to integrate multi-source data, including cooling water temperature, pressure, flow parameters, leakage signals, humidity gradient and pressure difference data, to calculate the dynamic detection threshold and trigger graded alarms based on the synergistic relationship between continuous negative deviation of flow, asymmetric distortion of temperature field, sudden change of humidity gradient and matching of pressure difference spectrum characteristics.

2. The intelligent monitoring system for liquid cooling cabinet according to claim 1, characterized in that: Dynamically adjust the cooling tower operating status and circulating water pump flow of the primary side circulation according to the cooling water parameters to achieve variable flow control of cooling water, including: The temperature, pressure and flow data of cooling water are collected in real time through the temperature sensor, pressure sensor and mass flow meter in the primary side circulation pipeline of cooling water; The cooling water temperature, pressure and flow data are input into the PLC controller, and the cooling tower heat exchange efficiency is calculated by the approximation algorithm and the difference between the cooling tower outlet water temperature and the outdoor wet bulb temperature; Dynamically adjust the cooling tower fan frequency and number of running units according to the heat exchange efficiency; According to the adjusted operating status of the cooling tower, combined with the actual demand for cooling water flow and energy consumption optimization goals, the PLC controller adjusts the speed of the circulating water pump through the frequency converter to achieve variable flow control of cooling water.

3. The intelligent monitoring system for liquid cooling cabinet according to claim 2, characterized in that: The cooling water temperature, pressure and flow data are input into the PLC controller. The cooling tower heat exchange efficiency is calculated by the approximation algorithm and the difference between the cooling tower outlet water temperature and the outdoor wet bulb temperature, including: The PLC controller extracts the cooling tower outlet water temperature and the outdoor wet-bulb temperature data from the storage area, and obtains the difference between the cooling tower outlet water temperature and the outdoor wet-bulb temperature data; The PLC controller starts the approximation algorithm, compares the temperature difference with the preset difference range, and obtains the comparison result; Based on the comparison results, a comprehensive analysis is performed in combination with the temperature, pressure and flow data of the cooling water. If the temperature difference is within the preset difference range, the potential impact of fluctuations in each parameter on the efficiency is analyzed; if the temperature difference is ≥ the preset difference range, the abnormal parameters are determined, and finally the heat exchange efficiency of the cooling tower is obtained.

4. The intelligent monitoring system for a liquid cooling cabinet according to claim 3, characterized in that: Real-time monitoring of secondary side piping status and cabinet environment through cooling distribution units and distributed sensor networks, including: Install an array of optical fiber temperature sensors on the secondary side pipeline, distribute them along the pipeline to form a temperature field monitoring network, and collect temperature data at different locations on the secondary side pipeline in real time; Temperature anomaly data is extracted from the temperature data at different locations on the secondary side pipeline. At the same time, the pressure differential sensor in the cooling distribution unit monitors the secondary side inlet and outlet pressure differential data in real time. When the secondary side inlet and outlet pressure differential is ≥ the preset range and the change trend meets the characteristics of pipeline blockage, the pipeline is assessed to have a blockage risk based on the preset threshold and logical judgment rules, and an early warning signal is issued; After receiving the pipeline blockage risk warning signal, the humidity sensor array and temperature and humidity sensor inside the cabinet simultaneously collect cabinet environmental parameters; the humidity sensor array detects changes in environmental humidity inside the cabinet, and the temperature and humidity sensor collects environmental temperature and humidity data.

5. The intelligent monitoring system for liquid cooling cabinet according to claim 4, characterized in that: Leak location is achieved through built-in leak sensors, temperature sensors, and pressure sensors, and automatic refilling using a liquid storage tank, including: A liquid collection tray is set at the bottom of the cooling distribution unit, and a leakage detection cable is laid inside the tray. When the coolant leaks, an impedance change signal is triggered, and the impedance change signal is cross-verified with the signal sent by the leakage sensor; After confirming the coolant leak, the leak point is located through the temperature field distortion characteristics of the distributed optical fiber temperature sensor, and the spatial position information of the leak point is spatially matched with the leakage signal; When it is detected that the liquid level in the liquid storage tank is less than the preset threshold, the refill pump is immediately started to inject coolant into the cooling circulation loop.

6. The intelligent monitoring system for a liquid cooling cabinet according to claim 5, characterized in that: The system integrates multi-source data, including cooling water temperature, pressure, flow parameters, leakage signals, humidity gradient, and pressure differential data, to calculate dynamic detection thresholds. It then triggers graded alarms based on the synergistic relationship between persistent negative flow deviations, asymmetric temperature field distortion, sudden humidity gradient changes, and pressure differential spectrum feature matching, including: The temperature, pressure, and flow data of the primary side cooling water, the secondary side leakage signal, the humidity gradient data in the cabinet, and the pressure difference data of the cooling distribution unit are input into the data fusion unit. The data of the data fusion unit is preprocessed through time series to generate a denoised time series data set. Based on the denoised time series data set, the covariance between parameters is calculated, periodic fluctuations are detected, and abnormal patterns are identified. Combined with the correlation characteristics of normal operating conditions in historical operating data, an adaptive dynamic threshold for leak detection is calculated. The dynamic threshold for leak detection is adjusted based on the real-time data stream. When the cooling water flow rate has a continuous negative deviation ≥ the dynamic threshold for leak detection and the distributed fiber optic temperature sensor network detects asymmetric distortion of the temperature field, it is determined to be an early leakage risk and an early warning signal is triggered. If, on the basis of triggering the early warning signal, it is also detected that the humidity difference between adjacent sensors in the cabinet is ≥ the set threshold and the pressure difference spectrum shows leakage characteristics, it will be upgraded to a high-confidence leak alarm.

7. The intelligent monitoring system for a liquid cooling cabinet according to claim 6, characterized in that: Based on the denoised time series data set, the covariance between parameters is calculated, periodic fluctuations are detected, and abnormal patterns are identified. Combined with the correlation characteristics of normal operating conditions in historical operating data, an adaptive dynamic threshold for leak detection is calculated, including: Collect multi-source data, including primary-side cooling water temperature, pressure, and flow data, secondary-side leakage signals, cabinet humidity gradient data, and cooling distribution unit pressure difference data; For any two parameters in multi-source data, the covariance between the two parameters is calculated to obtain the weighted average of the covariance between the parameters; For each parameter in the multi-source data, the periodic characteristics of the parameter are determined by analyzing the value change pattern at different time points. Based on the periodic characteristics, the theoretical value of the corresponding parameter at each time point is predicted to determine the periodic fluctuation amplitude of the corresponding parameter. Based on the periodic fluctuation amplitude of each parameter, the weighted average value of the parameter periodic fluctuation is determined. Analyze multi-source data and use preset anomaly judgment rules to identify different anomaly patterns. For each anomaly pattern, count the number of anomaly points in the multi-source data that match the anomaly pattern and record the total number of data points. Determine the severity of the anomaly pattern based on the number of anomaly points and the total number of data points for each anomaly pattern, and obtain a weighted average of the severity of the anomaly pattern. Extract the correlation features of each parameter under normal operating conditions from historical operating data and convert the correlation features of each parameter into a single characteristic value; An adaptive dynamic threshold for leak detection is determined based on the weighted average of covariances between parameters, the weighted average of periodic fluctuations of parameters, the weighted average of severity of abnormal patterns, and a single characteristic value of historical operating data.

8. A processing device for a liquid cooling cabinet, which implements the system according to any one of claims 1 to 7, characterized in that: include: Customized cooling tower processing equipment for manufacturing closed cooling towers with integrated variable frequency fans and optimized heat exchange efficiency structures; Pipeline processing equipment is used to form primary and secondary circulation pipelines and pre-embed distributed optical fiber temperature sensors and leakage detection cables inside the pipe walls to achieve real-time monitoring of pipeline status; The cold distribution unit assembly line is used to produce cold distribution units with integrated double-walled liquid storage tanks. Self-cleaning differential pressure sensors are installed at the inlet and outlet of the cold distribution unit, and an automatic liquid replenishment pump based on liquid level feedback control is integrated; The automatic sensor deployment device is used to navigate through lidar, install a humidity gradient detection array on the inner wall of the cabinet, attach temperature sensors to the surface of the cold plate, and automatically complete the wiring of sensor signal cables.

9. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the system according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program, which, when executed by a processor, implements the system according to any one of claims 1 to 7.

Images