Method and device for managing a cooling liquid, storage medium and electronic device

Abstract

The application discloses a cooling liquid management method and device, a storage medium and an electronic device, relates to the technical field of immersion liquid cooling, and comprises the following steps: acquiring a current dielectric characteristic and a current spectral characteristic of cooling liquid, wherein the cooling liquid is used for immersing a server and absorbing and transferring heat of the server; calculating a current content of the cooling liquid according to the current dielectric characteristic and the current spectral characteristic, wherein the current content is used for indicating respective proportions of at least one cooling working medium included in the cooling liquid; and in the case that the current content satisfies a preset content threshold condition, adjusting the current content of the cooling liquid to be a target content. The technical problem that the immersion liquid cooling server in the related art has a low cooling efficiency is solved, and the technical effect that the cooling efficiency of the immersion liquid cooling server is improved is achieved.

Description

Technical Field

[0001] This application relates to the field of immersion liquid cooling, and more particularly to a method and apparatus for managing coolant, a storage medium, and electronic equipment. Background Technology

[0002] Immersion liquid cooling completely submerges the server or its components in a coolant, utilizing the coolant's high thermal conductivity to directly contact the heat-generating elements and rapidly remove heat, achieving efficient heat dissipation. Choosing the right coolant is crucial to the performance of a liquid cooling system, and hydrocarbon and organosilicon compound coolants and fluorocarbon compound coolants have become the most widely used coolants due to their excellent thermal conductivity, good electrical insulation, and environmental friendliness.

[0003] The composition of coolant can change during use due to various factors, including evaporation, decomposition, and contamination. This is especially true in phase change immersion liquid cooling systems, where the working fluid repeatedly undergoes liquid-to-gas transitions, making it more susceptible to external environmental influences. However, in immersion liquid-cooled servers using related technologies, the lack of effective monitoring methods means that coolant performance degradation may go undetected, leading to reduced cooling efficiency and impacting server operation. In other words, immersion liquid-cooled servers in related technologies suffer from relatively low cooling efficiency. Summary of the Invention

[0004] This application provides a method and apparatus for managing coolant, a storage medium, and an electronic device to at least solve the problem of low cooling efficiency in immersion liquid-cooled servers in the related art.

[0005] This application provides a method for managing a coolant, including: acquiring the current dielectric characteristics and current spectral characteristics of the coolant, wherein the coolant is used to immerse a server and absorb and transfer heat from the server;

[0006] The current content of the coolant is calculated based on the current dielectric characteristics and the current spectral characteristics, wherein the current content is used to indicate the proportion of each of the at least one cooling working fluid included in the coolant;

[0007] If the current content meets the preset content threshold, adjust the current content of the coolant to the target content.

[0008] This application also provides a coolant management device, including: a feature acquisition module for acquiring the current dielectric characteristics and current spectral characteristics of the coolant, wherein the coolant is used to immerse the server and absorb and transfer the server's heat;

[0009] The content calculation module is used to calculate the current content of the coolant based on the current dielectric characteristics and the current spectral characteristics, wherein the current content is used to indicate the proportion of each of the at least one cooling working fluid included in the coolant.

[0010] The content adjustment module is used to adjust the current content of the coolant to the target content when the current content meets the preset content threshold condition.

[0011] This application also provides an electronic device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of any of the above-described coolant management methods.

[0012] This application also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements the steps of any of the above-described coolant management methods.

[0013] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of any of the above-described coolant management methods.

[0014] This application obtains the current dielectric and spectral characteristics of the coolant, which is used to immerse the server and absorb and transfer its heat. The current content of the coolant is calculated based on these characteristics, indicating the proportion of each of the at least one cooling medium included in the coolant. If the current content meets a preset threshold, the coolant content is adjusted to a target level. By continuously acquiring the dielectric and spectral characteristics of the coolant and calculating its current content, the proportion of mixed cooling media in the coolant is ensured to remain within an optimal range, thereby improving the server's heat exchange efficiency and maintaining stable operation under high loads. The preset threshold can provide early warning of any changes in the cooling medium content that may affect the stability of the cooling system. For example, a low proportion of fluorocarbon cooling media may lead to uneven cooling, or an excessive proportion of hydrocarbon cooling media may affect electrical insulation. Timely adjustments can avoid these problems and maintain long-term system stability. Therefore, this addresses the problem of low cooling efficiency in immersion liquid-cooled servers in related technologies. Attached Figure Description

[0015] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the hardware environment for an optional coolant management method according to an embodiment of this application;

[0017] Figure 2This is a flowchart of an optional coolant management method according to an embodiment of this application;

[0018] Figure 3 This is a schematic diagram of an optional immersion server according to an embodiment of this application;

[0019] Figure 4 This is a schematic diagram of an optional coolant management method according to an embodiment of this application;

[0020] Figure 5 This is a schematic diagram of another optional immersion server according to an embodiment of this application;

[0021] Figure 6 This is a schematic diagram of another optional coolant management method according to an embodiment of this application;

[0022] Figure 7 This is a structural block diagram of an optional coolant management device according to an embodiment of this application. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0024] It should be noted that, in the description of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. The terms "first," "second," etc., in this application are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0025] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] According to one aspect of the embodiments of this application, a method for managing coolant is provided. As an optional implementation, the above-described method for managing coolant can be applied to, but is not limited to, [examples of other methods]. Figure 1 The hardware environment shown includes a coolant management system. This coolant management system may include, but is not limited to, terminal device 102, immersion server 110, and coolant 112. Terminal device 102 runs a target client (e.g., ...). Figure 1As shown, taking a target client as an example (a client capable of monitoring and managing the status of an immersion server system), the terminal device 102 includes a display 108, a processor 106, and a memory 104. The display 108 can be used to display the system interface, and also provides a human-machine interface to receive operations on the interface and touch operations on different controls. The processor generates interaction instructions in response to the aforementioned human-machine interaction operations and sends these instructions to the server. The memory stores coolant characteristics.

[0027] Specifically, the immersion server 110 may include a server, a liquid cooling tank, a coolant 112, and a cooling system, with the server immersed in the coolant 112.

[0028] Servers include CPUs, GPUs, memory, storage devices (such as SSDs or HDDs), network cards, power modules, and more. These components generate a lot of heat during operation, requiring efficient heat dissipation to ensure performance and prevent overheating damage.

[0029] A liquid cooling tank is a container in which server hardware components are completely or partially immersed, filled with coolant to provide a closed liquid cooling environment. The material of the liquid cooling tank must have good corrosion resistance and thermal stability to meet the requirements of different types of coolants.

[0030] Coolant 112 is a medium used to absorb and transfer heat from server hardware components, and can be hydrocarbon, silicone, or fluorocarbon, etc. The selection of coolant is based on its thermal conductivity, dielectric properties, environmental friendliness, and long-term stability.

[0031] The cooling system may include: A heat exchanger: a component responsible for transferring heat from the coolant to an external cooling medium (such as chilled water or air), typically comprising a series of pipes and fins, which can be water-cooled, air-cooled, or refrigerant-cooled. A circulating pump: a key device used to drive the coolant to circulate within the system, ensuring uniform heat distribution and effective transfer. It can be a centrifugal pump, gear pump, or magnetically driven pump. A filtration system: used to remove impurities, particles, and potential chemical contaminants from the coolant, maintaining its cleanliness and extending its service life. The filter can be a physical filter screen or a chemical adsorption material. A pressure and level regulator: a device that ensures the pressure and level inside the liquid cooling tank remain stable within the target range, preventing cooling efficiency from being affected by air bubble buildup or level fluctuations.

[0032] Assumption Figure 1The terminal device 102 runs a client for managing the coolant of the immersion server. The specific process in this embodiment is as follows: As in step S102, the immersion server 110 sends the current dielectric characteristics and current spectral characteristics of the coolant to the terminal device 102. The terminal device 102 executes steps S104-S108, calculating the current content of the coolant based on the current dielectric characteristics and current spectral characteristics. The current content indicates the proportion of each of the at least one cooling medium included in the coolant. If the current content meets a preset content threshold condition, step S110 is executed to adjust the current content of the coolant to the target content.

[0033] In an optional implementation, the overall operation flow of the coolant management system can be as follows: An online monitoring device continuously monitors the characteristics of the coolant and transmits the data to the data acquisition unit of the terminal equipment. The data processing and analysis module of the terminal equipment analyzes the dielectric and spectral characteristics to identify the type of coolant and classifies it through an intelligent classification management system. The current content of the coolant is calculated based on the dielectric and spectral characteristics, and its content is monitored to ensure it meets preset content threshold conditions. If the current content does not meet the preset conditions, an early warning and control module is triggered, adjusting the coolant content to the target content, and simultaneously adjusting the air pressure and liquid level as needed to maintain system stability. The user interface displays the system status, and the remote monitoring and communication system allows maintenance personnel to remotely monitor data. Upon receiving early warning information, maintenance measures can be taken remotely or on-site to ensure the continuous and efficient operation of the system.

[0034] Through the coordinated operation of the aforementioned components, the immersion liquid-cooled server system can achieve precise classification and content management of the coolant working fluid, maintaining stable operation of the system in complex environments. At the same time, the intelligent control of terminal equipment further improves maintenance efficiency and system response speed.

[0035] Embodiments of this application provide a method for managing coolant. Figure 2 This is a flowchart of an optional coolant management method according to an embodiment of this application; as shown... Figure 2 As shown, the method for managing the coolant includes:

[0036] Step S202: Obtain the current dielectric characteristics and current spectral characteristics of the coolant, wherein the coolant is used to immerse the server and absorb and transfer the server's heat.

[0037] It should be noted that coolant is a liquid medium used in immersion liquid cooling systems to directly contact server components and absorb and transfer the heat they generate. Coolant must have good thermal conductivity, low electrical conductivity (or a sufficiently low dielectric constant), high dielectric strength, and good environmental and operational safety.

[0038] Dielectric characteristics, including dielectric constant (ε) and dielectric loss (tanδ), reflect a material's ability to store and dissipate electrical energy. In coolant monitoring, changes in dielectric constant can indicate dynamic changes in component proportions, while dielectric loss reflects fluctuations in the coolant's dielectric properties, such as those caused by contamination, aging, or component imbalances. Spectral characteristics refer to the specific absorption or scattering spectra exhibited by the coolant under spectral analysis (such as Raman spectroscopy or infrared spectroscopy). Each compound has its unique spectral fingerprint, and these spectral characteristics can be used to identify the various components and their concentrations in the coolant.

[0039] In an optional implementation, monitoring the dynamic changes in the dielectric and spectral characteristics of the coolant as it directly contacts server components and absorbs heat while transferring heat to the cooling system is crucial for maintaining the performance of the coolant and the stability of server operation.

[0040] In optional implementations, real-time acquisition of the dielectric and spectral characteristics of the coolant is a crucial step in ensuring the effective operation of the cooling system during the classification and content detection of the immersion liquid cooling working fluid. The dielectric and spectral characteristics of the coolant are directly related to its chemical composition and physical state; therefore, any changes in component ratios, liquid degradation, or contamination will be reflected in these characteristics. Continuous monitoring of these characteristics allows for timely detection of coolant state changes, enabling appropriate maintenance or adjustment measures, such as replenishing or replacing the coolant, to maintain optimal coolant performance and stable server operation.

[0041] It should be noted that dielectric characteristics can be obtained by installing electrochemical impedance spectroscopy array sensors and high-frequency dielectric sensors (such as sensors with interdigitated electrodes). These sensors should be designed to withstand the environmental conditions of the coolant and ensure long-term stable operation. Data on the dielectric constant and dielectric loss of the coolant should be continuously or periodically collected.

[0042] It should be noted that spectral characteristics can be acquired by installing a miniature fiber optic spectrometer capable of spanning the entire spectral range of the coolant (e.g., ultraviolet to infrared) to capture all possible spectral features. Spectral data of the coolant, such as Raman or infrared spectra, should be acquired in real-time or periodically. This data should include the absorption and reflectance spectral characteristics of the coolant for identifying specific components. Hydrocarbons, organosilicones, and fluorocarbons in the coolant can be identified by analyzing the position, intensity, and shape of spectral characteristic peaks. The relative abundance of coolant components can be assessed using the analysis of spectral data.

[0043] Step S204: Calculate the current content of the coolant based on the current dielectric characteristics and the current spectral characteristics, wherein the current content is used to indicate the proportion of each of the at least one cooling working fluid included in the coolant.

[0044] It should be noted that the current content indicates the relative proportion of each component in the coolant at the current moment, which is used to indicate the composition of the coolant and is a key parameter for managing and optimizing the cooling system.

[0045] In an optional implementation, the system needs to simultaneously consider the dielectric properties of the coolant (such as dielectric constant and dielectric loss) and spectral analysis data (such as Raman spectroscopy and infrared spectral characteristics) to calculate the real-time content ratio of each type of cooling medium (at least one) in the coolant.

[0046] In optional implementations, a database containing the dielectric and spectral characteristics of various cooling media (hydrocarbons, silicones, and fluorocarbons) can be established for comparison and identification. Alternatively, a model can be developed that can predict the proportion of each component in a mixed coolant using dielectric properties and spectral characteristics. The model can be based on linear regression, machine learning algorithms (such as support vector machines and neural networks), or more complex mathematical calculation methods. The dielectric and spectral characteristics collected in real time are input into the model to calculate the proportion of each cooling media at the current moment. The model's output can be compared with standard features in the database to calibrate and verify the accuracy of the calculation results. By comparing with historical data or preset thresholds, the system can automatically detect abnormal changes in the proportions and trigger corresponding warning or adjustment mechanisms. The specific calculation methods will be described in detail later and will not be repeated here.

[0047] It should be noted that coolant can be a mixture of various coolants. In immersion liquid cooling technology, selecting a suitable coolant is a key factor determining system efficiency, safety, economy, and environmental impact. The following are types of coolants: hydrocarbons, fluorocarbons, water-based coolants, lubricating oils, and specialty coolants.

[0048] Step S206: If the current content meets the preset content threshold condition, adjust the current content of the coolant to the target content.

[0049] It should be noted that preset content thresholds can be a series of predefined content ranges used to determine whether the content of coolant components is at the desired or safe level. Preset content thresholds are set based on coolant type, system requirements, and operating conditions. Target content is the specific component content required for optimal system operation, typically consistent with or within an acceptable range of the preset content thresholds. Adjustment involves modifying the coolant component content to approach or reach the target content to ensure cooling system performance or safety.

[0050] When the detected coolant component content is within a preset allowable range, the system will take measures to adjust the content to precisely achieve the target content level. This includes, but is not limited to, adding or removing specific components, or adjusting the component ratio through chemical treatment.

[0051] In an optional implementation, once the current content of components in the coolant is detected, the system evaluates it to see if it meets preset content threshold conditions. These threshold conditions ensure coolant performance and system operational safety. If the current content is within acceptable limits, but to maintain optimal cooling efficiency or avoid potential problems during long-term operation, the system will further adjust the content to a more specific target content. This adjustment may be accomplished through automated equipment or manual intervention, depending on the specific design and operating strategy of the system.

[0052] In an optional implementation, when the content of a component is detected to be below the target content, that component can be added to the system until the content reaches the target value. For example, if the content of a fluorocarbon coolant is too low, fluorocarbon compounds can be added automatically or manually. When the content of a component exceeds the target content, it may be necessary to remove the excess component or replace part of the mixture. This can be achieved by setting up specialized separation equipment or replacing part of the coolant. In some cases, especially when components undergo chemical reactions or quality changes, it may be necessary to adjust the component ratio through chemical treatment to restore it to the target content.

[0053] Example 1:

[0054] Suppose an immersion liquid cooling system uses a mixture of hydrocarbon and organosilicon compounds and fluorocarbon compounds as coolant. The target content of the system is set as 30% for hydrocarbon and organosilicon coolant and 70% for fluorocarbon coolant.

[0055] Step S202: Real-time acquisition of dielectric properties and spectral characteristics of hydrocarbon, organosilicon, and fluorocarbon cooling media in the coolant using a miniature fiber optic spectral probe and an electrochemical impedance array sensor.

[0056] Step S204: Compare the detection data with the coolant database to identify the specific components in the coolant and preliminarily calculate their content ratio.

[0057] Step S206: Based on the analysis results in step S204, assume the currently detected content ratio is 35% for hydrocarbon and organosilicon cooling media and 65% for fluorocarbon cooling media. This exceeds the preset threshold (±5%) and requires adjustment. The system automatically initiates the fluorocarbon cooling media addition procedure, adding fluorocarbon compounds to the system through a precisely metered feed pump until the content ratio is adjusted to the target range (30% for hydrocarbon and organosilicon, 70% for fluorocarbon).

[0058] After adjustment, the system performs another content test to verify whether the target content has been successfully reached. If the content is still not within the target range, the system will repeat step S206 until the content meets the requirements. Throughout the process, the intelligent classification management system records all detection and adjustment data for subsequent analysis and maintenance. Simultaneously, the system can also adjust preset thresholds and target contents based on dynamic changes in the coolant to adapt to coolant aging or changes in system requirements.

[0059] This application obtains the current dielectric and spectral characteristics of the coolant, which is used to immerse the server and absorb and transfer its heat. The current content of the coolant is calculated based on these characteristics, indicating the proportion of each of the at least one cooling medium included in the coolant. If the current content meets a preset threshold, the coolant content is adjusted to a target level. By continuously acquiring the dielectric and spectral characteristics of the coolant and calculating its current content, the proportion of mixed cooling media in the coolant is ensured to remain within an optimal range, thereby improving the server's heat exchange efficiency and maintaining stable operation under high loads. The preset threshold can provide early warning of any changes in the cooling medium content that may affect the stability of the cooling system. For example, a low proportion of fluorocarbon cooling media may lead to uneven cooling, or an excessive proportion of hydrocarbon cooling media may affect electrical insulation. Timely adjustments can avoid these problems and maintain long-term system stability. Therefore, this addresses the problem of low cooling efficiency in immersion liquid-cooled servers in related technologies.

[0060] In an optional implementation, calculating the current content of the coolant based on the current dielectric characteristics and current spectral characteristics includes: calculating the type of at least one cooling medium included in the coolant and the first content corresponding to each of the at least one cooling medium based on the current spectral characteristics, wherein the current spectral characteristics are used to indicate the absorption or emission characteristics of the coolant to light; determining the second content corresponding to each of the at least one cooling medium based on the current dielectric characteristics and the reference dielectric characteristics corresponding to each of the at least one cooling medium, wherein the current dielectric characteristics are used to indicate the response of the coolant to an electric field; and calculating the current content based on the first content and the second content.

[0061] It should be noted that the reference dielectric characteristic refers to the dielectric characteristic of a known pure coolant or a coolant mixture with a specific component ratio under standard conditions, serving as the basis for comparison with the current dielectric characteristic. The first content can be the percentage of coolant included in the coolant as determined by spectral characteristics, and the second content can be the percentage of coolant included in the coolant as determined by dielectric characteristics.

[0062] In an optional implementation, based on spectral feature analysis, the types of cooling media contained in the coolant can be identified, and the initial content of each component can be preliminarily estimated. Subsequently, by comparing the dielectric features with those of a reference dielectric, the second content of each cooling media is further determined, thereby calculating the accurate current cooling media content by combining the first and second contents. By combining the chemical composition information provided by spectral features and the dielectric property fluctuations reflected by dielectric features, the system can quickly respond to changes in the coolant state, promptly detect and address component imbalances or other potential problems, thereby ensuring the efficient and stable operation of the immersion liquid-cooled server.

[0063] In an optional implementation, component identification and content calculation based on spectral characteristics can utilize a miniature fiber optic spectral probe to acquire real-time spectral data of the coolant. The spectral characteristics of the coolant are analyzed to identify the type of coolant present. Based on information such as the intensity and position of spectral characteristic peaks, chemometric methods (such as multiple linear regression, partial least squares, or neural network algorithms) are applied to calculate the first content. Content verification and adjustment based on dielectric characteristics involves acquiring the dielectric constant and dielectric loss data of the coolant. The current dielectric characteristics are compared with reference dielectric characteristics in a coolant characteristic database. Difference analysis or machine learning models are used to determine the second content of each coolant, verifying the accuracy of the spectral analysis results. By comparing the first and second content results, a weighted average or statistical fusion algorithm is used to calculate the final current content of the coolant. This algorithm considers the errors and uncertainties in spectral and dielectric measurements, aiming to provide the most accurate content information.

[0064] Example 2:

[0065] Assume the coolant in the liquid cooling system is a mixture of hydrocarbon (HC) and fluorocarbon (Fluorinated) fluid. The system is equipped with an electrochemical impedance spectroscopy array sensor and a miniature fiber optic spectral probe. The sensors continuously acquire dielectric and spectral characteristic data of the coolant and transmit this data to the intelligent classification and management system in real time.

[0066] The system identifies hydrocarbon and fluorine components in the coolant using a miniature fiber optic spectral probe, and calculates the initial content ratio by integrating the characteristic peaks of the Raman spectrum, for example, 60% hydrocarbon components and 40% fluorine components.

[0067] Simultaneously, the dielectric sensor measures the dielectric constant ε1 and dielectric loss tanδ1 of the coolant. These data are compared with known reference dielectric characteristics of hydrocarbon and fluorocarbon working fluids in a database to determine the second content ratio; for example, the model calculates a hydrocarbon component of 62% and a fluorocarbon component of 38%.

[0068] The system performs a comprehensive analysis of the first and second contents, and calculates the final current contents using a weighted average algorithm. The assumed result is that the hydrocarbon component accounts for 61% and the carbon-fluorine component accounts for 39%.

[0069] Assume the preset safe ranges for hydrocarbon and fluorocarbon components are 55%~65% and 35%~45%, respectively. If the system detects that the current content is within the safe range, no alarm is needed, but the data is recorded for subsequent cooling fluid management decisions; if the content exceeds the safe range, the system immediately triggers an alarm and initiates the corresponding maintenance procedure.

[0070] In an optional implementation, the current content is calculated based on the first and second contents, which can be achieved through a model. Multi-source data fusion algorithms based on machine learning can be used to infer coolant content from dielectric and spectral features. Several models can be considered: Neural network models: including traditional Artificial Neural Networks (ANNs), Convolutional Neural Networks (CNNs), or Long Short-Term Memory Networks (LSTMs), used to handle complex nonlinear relationships, suitable for data fusion and analysis of dielectric features (such as dielectric constant and dielectric loss factor) and spectral features (involving light intensity variations across multiple wavelengths). Support Vector Machines (SVMs): perform particularly well in handling high-dimensional feature spaces and can be used for classification and regression analysis, mapping dielectric and spectral features to multi-class or multi-variable predictions of coolant content. Random Forests: an ensemble learning method that improves prediction accuracy and robustness by constructing multiple decision tree models, suitable for multi-source data fusion and coolant content prediction. XGBoost: as an efficient gradient boosting framework, it maintains high speed and accuracy when processing large datasets, making it ideal for feature selection, weight allocation, and final coolant content prediction for multi-source data.

[0071] The above model is for illustrative purposes only; the specific implementation method should be selected according to the needs.

[0072] Example 3:

[0073] To illustrate the calculation process, the following example uses a neural network to explain how the current content is obtained:

[0074] In an optional implementation, a large amount of training data is collected, covering dielectric characteristics (dielectric constant, dielectric loss factor, etc.), spectral characteristics (Raman and infrared spectral data), and corresponding coolant content labels. The data is cleaned, normalized, and feature-engineered to ensure the quality of the model input data.

[0075] Design a multiple-input multiple-output (MIMO) neural network model comprising two branches: one for processing dielectric feature data and the other for processing spectral feature data. Each branch can be an independent neural network, such as a fully connected (dense) layer, mapping its respective dataset to a set of intermediate representations. These two intermediate representations are then merged (through operations such as concatenation or averaging) to form a composite feature vector, which is finally output as the coolant content through a series of fully connected layers.

[0076] The neural network is trained using a collected dataset, including dielectric features, spectral features, and coolant content labels. This typically involves tuning the network parameters to minimize the difference between the predicted content and the actual labels. The training process may employ techniques such as cross-validation, learning rate decay, and dropout to improve the model's generalization ability and avoid overfitting. The model's performance is evaluated on independent validation and test sets to ensure it can accurately predict coolant content even on unseen data.

[0077] The trained model is deployed into the intelligent classification and management system, receiving real-time inputs of dielectric and spectral characteristics from online monitoring devices, and outputting predicted coolant content for real-time monitoring and system regulation. In practical applications, the model should be retrained periodically with new datasets to adapt to the changing trends of coolant composition over time. Furthermore, the model's performance can be continuously optimized by adding more training data or adjusting hyperparameters to improve prediction accuracy.

[0078] Training Phase: Data Preparation: Collect data from the operation of the immersion liquid-cooled server cooling system, including dielectric characteristics, spectral characteristics, and the corresponding actual coolant content ratio at specific time points.

[0079] Model Setup: Two neural network branches are designed: one for processing dielectric feature data and the other for processing spectral feature data. Each branch contains at least two fully connected layers for feature extraction. In the last layer, the outputs of the two branches are concatenated using some kind of fusion operation (such as concatenation). Additional fully connected layers are added to finally output the predicted value of coolant content.

[0080] Training and tuning: The dataset is split into training, validation and test sets. The neural network model is trained using the training set data and the model parameters are tuned using the validation set data until the prediction results on the test set achieve satisfactory accuracy.

[0081] Real-time input during the prediction phase: The intelligent classification and management system acquires the latest dielectric and spectral characteristic data from online monitoring devices. Feature input model: Real-time data is input into two branches of the neural network for feature extraction. Data fusion and prediction: At the end of the network, the features extracted from the two branches are fused, and then passed through the final fully connected layer to obtain a real-time prediction of the coolant content. Output: The predicted coolant content data is sent to the system controller for dynamically adjusting the coolant replenishment or replacement strategy to ensure that the coolant composition is maintained at an optimal level.

[0082] In an optional implementation, a database of coolants can be established before managing the coolant. The database can store reference dielectric and spectral characteristics of each coolant to indicate the characteristics of each coolant.

[0083] It should be noted that the coolant characteristic database is one of the core technical supports for the management and maintenance of immersion liquid-cooled server systems. It stores key physicochemical properties of various coolants, aiming to provide a data foundation for system testing and classification. This database mainly includes, but is not limited to, the following types of data: Dielectric parameters: the dielectric constant and dielectric loss factor of the coolant. These parameters are crucial for understanding the electrical characteristics of the coolant, especially in immersion liquid-cooled systems, where the dielectric properties of the working fluid directly affect its heat transfer efficiency and electrical isolation capability with the server's electronic components. Spectral fingerprints: the chemical composition identifier of the coolant obtained through spectral analysis, such as Raman spectroscopy and infrared absorption spectroscopy. Each coolant possesses unique spectral characteristics due to its different molecular structure, which can be considered its "chemical fingerprint." Thermophysical property data: including thermodynamic parameters such as the thermal conductivity, specific heat capacity, and boiling point of the coolant. These data are extremely important for evaluating the heat transfer capacity and operating range of the coolant.

[0084] In an optional implementation, the cooling medium in the database can be stored in the following form:

[0085] Coolant type: fluorocarbon;

[0086] Characteristic feature: CF bond Raman peak, with a specific location of approximately 1135±5 cm. -1 The presence and intensity of this peak can serve as confirmation of fluorocarbon coolants, because the vibrational modes of the CF bond provide unique spectral signatures in Raman spectroscopy.

[0087] Detection threshold: Peak area ≥ 5000 counts. This is to ensure reliable detection of the presence and concentration of fluorocarbon coolants even under the influence of factors such as noise and background light. The peak area is directly related to the concentration of fluorocarbons in the coolant.

[0088] Coolant type: Silicone-based;

[0089] Characteristic: Si-O-Si infrared absorption, located at approximately 1080±10 cm. -1 In infrared spectroscopy, the stretching vibrations of Si-O-Si bonds exhibit a certain absorption intensity, a characteristic that serves as an important clue for identifying organosilicon coolants.

[0090] Detection threshold: Absorbance ≥ 0.8 AU. This threshold is set to ensure accurate identification of the composition and concentration of silicone coolant even in cases of slight contamination or degradation. The absorbance level reflects the concentration of specific chemical bonds in the coolant.

[0091] Coolant type: hydrocarbon-based;

[0092] Characteristic features: CH bending vibration, wavenumber approximately 1450±20cm -1 In fluorescence spectroscopy, different types of hydrocarbons exhibit unique absorption and emission lines, which helps to distinguish and quantify hydrocarbon coolants.

[0093] Detection threshold: Fluorescence intensity ≥ 3000 AU. Fluorescence intensity is an important indicator of hydrocarbon concentration. A threshold above 3000 AU means that the system can sensitively detect minute changes in hydrocarbon coolants, which is crucial for maintaining the purity and stability of the coolant.

[0094] The embodiments described above in this application, which combine spectral and dielectric measurement methods for content calculation, can effectively and accurately monitor the dynamic changes of the coolant in the coolant, which is of great significance for maintaining the stability of immersion liquid-cooled server systems and extending the service life of the coolant. The multi-source data fusion algorithm based on neural networks can effectively integrate dielectric and spectral characteristic data to provide accurate coolant content prediction, thereby achieving intelligent and refined management of coolant composition. This not only improves cooling efficiency but also extends the service life of the coolant and reduces system maintenance costs, representing a significant advancement in immersion liquid cooling technology.

[0095] In an optional implementation, calculating the types of at least one cooling medium included in the coolant and the first content corresponding to each of the at least one cooling medium based on the current spectral characteristics includes: determining the number of characteristic peaks in the current spectral characteristics as the number of types of at least one cooling medium, wherein the characteristic peaks are used to indicate local maxima in the current spectral characteristics; determining the type of at least one cooling medium based on the wavenumber corresponding to each of the at least one characteristic peak; and calculating the proportion corresponding to each of the at least one cooling medium based on the peak value of the at least one characteristic peak.

[0096] It should be noted that by analyzing spectral characteristics, the number of different types of coolant in the coolant can be directly estimated from the number of characteristic peaks. The wavenumber positions of characteristic peaks provide clues to the chemical identity of the coolant; by comparing them with a spectral library of known substances, specific components in the coolant can be identified. The intensity (peak height) of characteristic peaks is related to the concentration or content of the coolant; by analyzing the peak values ​​of characteristic peaks, the relative proportion of each coolant can be calculated.

[0097] In an optional implementation, a detection device equipped with a miniature fiber optic spectral probe can be used to acquire Raman or infrared spectral data of the coolant in real time, ensuring the quality and representativeness of the spectral data. Spectral analysis software is used to identify characteristic peaks in the spectrum, i.e., local maxima where absorption or reflection intensity is higher than the surrounding area. The number of characteristic peaks is calculated as a preliminary estimate of the number of coolant types.

[0098] The wavenumber position of each characteristic peak is analyzed and compared with existing coolant spectral databases to identify the type of each coolant. The content ratio of the coolant is calculated using chemometric methods by analyzing the peak intensity of the characteristic peaks. The intensity of the characteristic peaks is then converted into the concentration or mass fraction of the coolant using a pre-defined standard curve or empirical formula.

[0099] The embodiments described in this application not only identify the components in the coolant but also calculate their current content ratios, providing crucial data for subsequent coolant management. This real-time spectral analysis and content calculation scheme is essential for maintaining the stable operation and efficient cooling of data center immersion liquid cooling systems, and improves the accuracy of subsequent adjustments to the coolant content.

[0100] In an optional implementation, the proportion of each of the at least one cooling working fluid is calculated based on the peak value of at least one characteristic peak, including one of the following:

[0101] 1) Given the Raman spectrum of the coolant indicated by the current spectral features, identify at least one Raman characteristic peak in the current spectral features; calculate the peak area corresponding to each of the at least one Raman characteristic peak, and determine the proportion of the at least one peak area as the proportion of each of the at least one coolant.

[0102] 2) Given the infrared spectrum of the coolant indicated by the current spectral characteristics, determine at least one absorption characteristic peak in the current spectral characteristics; calculate the absorbance corresponding to each of the at least one absorption characteristic peak, and determine the proportion of each of the at least one absorbance as the proportion of each of the at least one coolant.

[0103] Raman characteristic peaks are characteristic peaks in Raman spectra that reflect the Raman shifts of specific chemical bonds in a substance. Their positions and intensities can also be used to identify substances or assess their state.

[0104] Infrared characteristic peaks are absorption peaks that different chemical bonds and groups (such as CH, OH, C=O, Si-O-Si, etc.) exhibit at specific wavenumbers in the infrared spectrum. These peaks are the infrared characteristic peaks of a substance and can provide detailed information about its molecular structure.

[0105] The peak value can be the point of maximum intensity of a characteristic peak in a spectrum, used to quantify the significance of the characteristic peak. Peak area, in Raman spectroscopy, is the integral area of ​​a characteristic peak, directly proportional to the content of the component. Absorbance, in infrared spectroscopy, is the intensity index of an absorption peak, directly proportional to the concentration of the component.

[0106] In an optional implementation, S1. Selecting characteristic peaks: First, identify the Raman characteristic peaks of specific components. Different chemical bonds or functional groups will have different shifts and intensities in the Raman spectrum.

[0107] S2. Integral Calculation: Using appropriate software or hardware tools, perform integral calculations on the selected characteristic peak region to obtain the peak area. The peak area typically reflects the relative concentration of the corresponding chemical bond or functional group.

[0108] S3. Calibration and Standardization: To convert peak area into actual concentration, Raman spectra need to be calibrated. This typically involves establishing a calibration curve using standard samples of known concentrations, linking peak area to concentration. Furthermore, to eliminate the influence of factors such as laser intensity and sample amount, peak area standardization is required.

[0109] S4. Component Proportion Calculation: After obtaining the characteristic peak areas of all components of interest, their actual concentrations can be calculated by comparing them with their respective calibration curves. Then, using this concentration data, the proportion of each component in the mixture can be calculated.

[0110] In an optional implementation, the basic principle of calculating component proportions using infrared spectroscopy is similar to that of Raman spectroscopy: S1. Selecting characteristic absorption peaks: First, determine the characteristic infrared absorption peaks for each component. Unlike Raman spectroscopy, infrared spectroscopy is generated based on the absorption of infrared light of a specific frequency by a substance; therefore, it is the absorption peaks, not the scattering peaks, that need to be identified.

[0111] S2. Absorbance Measurement: Next, measure the absorbance of the selected absorption peak. Absorbance (A) is related to the transmittance (T) and optical path length (L) of the incident light, following the Lambert-Beer Law: A = -log(T) = εCL, where ε is the molar absorptivity and C is the concentration. Absorbance is directly proportional to concentration, but the ε value differs for different chemical bonds or functional groups.

[0112] S3. Establish a calibration curve: Using standard samples of known concentration, establish a calibration curve between the absorbance of the infrared absorption peak and the concentration. This step is similar to Raman spectroscopy, and is to convert the spectral data into actual concentrations.

[0113] S4. Calculate Concentration and Ratio: Finally, by comparing the absorbance of the sample with the calibration curve, the concentration of each component is calculated. These concentration data are then used to calculate the component ratios, thereby understanding the compositional changes of the mixed coolant.

[0114] Both Raman and infrared spectroscopy can be used to calculate the component ratios in a mixed coolant by integrating the area of ​​characteristic peaks or measuring absorbance, combined with calibration and standardization processes. While these two spectroscopic techniques differ in principle, both provide accurate analysis of coolant composition in practical applications, which is crucial for maintaining the stability of immersion liquid cooling systems and improving cooling efficiency.

[0115] The embodiments described above enable real-time monitoring of the proportions of various components in the coolant, and automatic adjustment or early warning based on the detection results to ensure that the coolant component proportions remain within the optimal range, thereby maintaining efficient and stable system operation. This process utilizes the high resolution of Raman or infrared spectroscopy, combined with intelligent algorithms, to achieve precise management of coolant component content.

[0116] In an optional implementation, determining the second content of each of the at least one cooling medium based on the current dielectric characteristic and the reference dielectric characteristic corresponding to each of the at least one cooling medium includes: determining the reference dielectric characteristic corresponding to each of the at least one cooling medium; determining the weight corresponding to each of the at least one reference dielectric characteristic based on the at least one reference dielectric characteristic and the current dielectric characteristic, wherein the current dielectric characteristic is obtained by weighted summation of the at least one reference dielectric characteristic; and determining the ratio of the weights corresponding to each of the at least one reference dielectric characteristic as the second content.

[0117] It should be noted that the reference dielectric characteristics can be pre-measured or calibrated, representing the ideal dielectric properties of each individual component in a pure or mixed cooling medium. These data are typically obtained under laboratory conditions and are used to establish a model relating different component contents to their dielectric characteristics. They can be directly obtained from the database established above.

[0118] It should be noted that a set of reference dielectric characteristics representing pure or single-component coolants is required. These characteristics are then compared with the current dielectric characteristics of the actual coolant (which may be a mixture of multiple coolants). The mixing ratio of each coolant is inferred by calculating the weighted sum ratio. This ratio is known as the "second content".

[0119] In immersion liquid cooling systems, the dielectric properties of the coolant mixture are determined by the dielectric properties of its individual components, and these properties vary with the proportions of different components. Therefore, by measuring the current dielectric characteristics of the mixed coolant and comparing them with the reference dielectric characteristics of each pure coolant, the component proportions of the current mixture can be reconstructed using mathematical methods (such as weighted summation). Determining the "second content" depends not only on the measurement of the current dielectric characteristics but also on a comprehensive reference dielectric characteristic library containing dielectric performance data for all possible coolants. This allows for the reverse calculation of the content of each coolant in the current mixture based on the principle of weighted summation.

[0120] In an optional implementation, for each possible coolant (hydrocarbon, silicone, and fluorocarbon), its dielectric constant and dielectric loss in pure liquid form are determined under laboratory conditions and recorded in a database. Using an electrochemical impedance spectroscopy array or other dielectric property sensors, the dielectric constant and dielectric loss data of the coolant are acquired in real time to form a current dielectric characteristic record. The current dielectric characteristic is considered as a weighted sum of the reference dielectric characteristics of different coolants according to their content proportions. Using least squares or machine learning algorithms (such as neural networks), a set of weight values ​​that best fit the current dielectric characteristic is found; these weight values ​​correspond to the percentage content of each coolant in the mixture. Finally, the calculated weight ratios are converted into a second content representation of the coolant, i.e., the specific proportion of each component in the mixture.

[0121] Through the above-described embodiments of this application, not only can the "second content" of each component in the coolant be estimated independently, but the model can also be continuously optimized according to actual conditions to ensure the accuracy and reliability of the detection results. This method is particularly suitable for coolant combinations with similar spectral characteristics that are difficult to distinguish using a single technique, and can provide a more comprehensive solution for detecting the proportion of mixed coolant components.

[0122] In an optional implementation, calculating the current content of the coolant based on the current dielectric characteristics and the current spectral characteristics includes: obtaining candidate dielectric characteristics of the coolant, wherein the candidate dielectric characteristics are used to indicate the dielectric characteristics of the coolant over a period of time; calculating the dielectric change rate of the coolant based on the current dielectric characteristics and the candidate dielectric characteristics; and calculating the current content of the coolant based on the current dielectric characteristics and the current spectral characteristics if the dielectric change rate is greater than a preset change threshold.

[0123] It should be noted that candidate dielectric features can be a dataset representing the historical dielectric properties of the coolant, used to analyze the trend of coolant dielectric properties changing over time. The dielectric change rate can be the degree of difference between the current dielectric feature and the candidate dielectric features, usually expressed as a percentage or absolute rate of change. The preset change threshold can be a boundary value set by the system to distinguish whether the change in coolant dielectric properties is a normal fluctuation or an abnormal change.

[0124] It's important to note that the system not only acquires the current dielectric properties of the coolant but also records and stores historical dielectric properties over a past period as historical data (i.e., candidate dielectric characteristics). By comparing the current dielectric characteristics with average or typical values ​​over the past period, the system can quantify the rate of change of dielectric properties to monitor coolant stability. When the rate of change of dielectric properties exceeds a pre-set threshold, it indicates a significant change in the chemical composition or state of the coolant. In this case, the current composition needs to be recalculated by combining the current dielectric and spectral characteristics. This step aims to ensure the accuracy of coolant component control through more comprehensive data analysis.

[0125] Specifically, a high-frequency dielectric sensor can be used to continuously monitor the dielectric properties of the coolant, and this data can be periodically stored as historical records. The average dielectric characteristic value over a period of time is extracted from the historical records as candidate dielectric characteristics. The difference or ratio between the current dielectric characteristic and the candidate dielectric characteristics is calculated periodically or in real time to obtain the dielectric change rate. The dielectric change rate can be calculated cumulatively over time intervals or tracked in real time using a sliding window method. A preset threshold for the dielectric change rate is established, based on the properties of the coolant and the stability requirements of the system. When the calculated dielectric change rate exceeds the preset threshold, the system automatically triggers further content detection procedures. When the dielectric change rate trigger condition is met, the system simultaneously uses a spectral analysis device to obtain the spectral characteristics of the coolant. Using a previously developed model or algorithm, combined with the current dielectric and spectral characteristics, the current content of each component in the coolant is calculated. Based on the calculated current content, the health status of the coolant is assessed, and warning signals are issued or the coolant replenishment system is automatically adjusted if necessary. Coolant management decisions are updated, such as when to replace the coolant or add new components to maintain system performance.

[0126] The above-described embodiments of this application not only enhance the accuracy of coolant component monitoring, but also ensure that the system can respond quickly and take necessary management measures when the dielectric properties change, thereby maintaining the stability of the coolant and the efficient operation of the server system.

[0127] In an optional implementation, after obtaining the current dielectric characteristics and current spectral characteristics of the coolant, the process includes: calculating at least one degradation index of the coolant based on the current dielectric characteristics and current spectral characteristics, wherein the degradation index is a parameter used to indicate a decrease in coolant performance or a deterioration in quality; and generating a first prompt message when at least one degradation index meets the preset degradation conditions corresponding to each degradation index.

[0128] It should be noted that degradation indicators refer to parameters that quantify the degree of performance degradation or quality deterioration of the coolant. Degradation indicators may include, but are not limited to, shifts in the CF bond Raman peak, absorbance of acidic products, and decreased thermal stability, which correspond to changes in the coolant's chemical, electrochemical, and thermodynamic properties, respectively. Preset degradation conditions are threshold ranges for a series of degradation indicators set by the system to ensure the coolant's health. Once a detected degradation indicator exceeds these preset ranges, it is considered that the coolant has degraded.

[0129] The first alert is a warning or notification generated by the system when it detects a decline in coolant performance that meets preset degradation conditions. This alert informs maintenance personnel that the coolant needs to be checked, replenished, or replaced. The system first continuously acquires the dielectric and spectral characteristics of the coolant using a multi-sensor fusion detection module. Then, based on this real-time detection data, an intelligent algorithm calculates the coolant degradation index to determine if the coolant has deteriorated. Finally, when the system confirms that the coolant is in poor condition, it issues a maintenance notification.

[0130] In optional implementations, continuous real-time monitoring of the dielectric and spectral characteristics of the coolant is crucial in immersion liquid-cooled server cooling systems, as these characteristics directly reflect the coolant's electrical insulation properties and chemical stability. An intelligent classification management system compares these characteristics with standard characteristics of healthy coolants stored in a database to calculate degradation indicators reflecting changes in coolant performance. When any degradation indicator reaches or exceeds preset degradation conditions, it indicates that the coolant may have become contaminated or undergone changes in phase, electrochemical, or thermal properties. The system immediately generates an initial alert, prompting maintenance personnel to take appropriate action to prevent coolant degradation from adversely affecting server operation and system performance.

[0131] In an optional implementation, once at least one degradation indicator meets the degradation conditions, the system will generate a first alert message, notifying maintenance personnel via a local display screen, email, SMS, or integrated remote monitoring platform. The alert message should include details of the specific degradation indicator, suggested operating procedures, and possible coolant recovery strategies. Upon receiving the first alert message, maintenance personnel will perform necessary coolant checks, replenishments, or replacements according to the alert content, and record the results in the system log or maintenance log for subsequent analysis and optimization.

[0132] Through the above-described embodiments of this application, the immersion liquid-cooled server cooling system can monitor the health status of the coolant in real time, promptly detect and prevent coolant deterioration, ensure long-term stable operation of the system, reduce the risk of unexpected downtime, and improve the operation and maintenance efficiency and security of the data center.

[0133] In an optional implementation, at least one degradation index of the coolant is calculated based on the current dielectric characteristics and current spectral characteristics, including at least one of the following:

[0134] 1) Calculate the dielectric constant of the coolant based on the current dielectric characteristics, and determine the ratio of the imaginary part to the real part of the dielectric constant as the first degradation index, wherein the dielectric constant is used to indicate the capacitance of the coolant;

[0135] 2) Calculate the first shift of the first characteristic peak based on the current spectral characteristics, and determine the first shift as the second degradation index, wherein the first characteristic peak is used to indicate the decomposable cooling medium;

[0136] 3) Given the infrared spectrum of the coolant as indicated by the current spectral characteristics, determine the first absorbance of the second characteristic peak and use the first absorbance as the third degradation index;

[0137] If at least one degradation indicator meets the preset degradation conditions corresponding to that indicator, a first warning message is generated, including at least one of the following:

[0138] 1) If the rate of change of the first deterioration index is greater than the rate of change of the preset factor, generate the first prompt message;

[0139] 2) If the second degradation index is greater than the preset offset, generate the first prompt message;

[0140] 3) If the third degradation index is greater than the preset absorbance, generate the first prompt message.

[0141] It should be noted that the first degradation indicator can be the dielectric loss factor. The first characteristic peak refers to the spectral characteristic peak of a specific chemical component or structure exhibited by the coolant in spectral analysis. In fluorocarbon coolants, such as the Raman peak of the CF bond, it can indicate the decomposition status of fluorides in the coolant and is an important indicator for monitoring coolant degradation. The second characteristic peak may refer to the spectral characteristic peak of acidic products or other degradation byproducts generated in the coolant, such as specific absorption peaks in infrared spectroscopy, used to detect whether the coolant has undergone undesirable chemical changes, such as the formation of acidic products.

[0142] It should be noted that the first degradation index determines whether the electrical properties of the coolant have deteriorated by analyzing the dielectric constant of the coolant, particularly the change in the ratio of its imaginary to real parts. This index can reveal an increase in impurities or moisture in the coolant, as these will affect the capacitive portion of the dielectric constant.

[0143] The second degradation index monitors whether decomposable coolant (such as fluorides in fluorocarbons) has decomposed by shifting the position of the first characteristic peak in spectral analysis, indicating a decrease in the chemical stability of the coolant. Detecting changes in the absorbance of the second characteristic peak in the coolant's infrared spectrum to determine whether acidic products or other degradation byproducts have formed in the coolant is another method for monitoring the coolant's chemical properties.

[0144] The dielectric loss factor (the primary degradation indicator), usually expressed as tanδ (tan delta), where Δtanδ represents the relative change in the dielectric loss factor. In immersion liquid-cooled server systems, the coolant must not only possess excellent thermal conductivity but also be non-conductive or low-conductive to ensure the safety of server components. The dielectric properties of the coolant are crucial to its suitability; any significant change in the dielectric loss factor may reflect abnormal coolant conditions, such as the introduction of impurities, aging, or performance degradation.

[0145] In an optional implementation, the dielectric loss factor changes beyond a certain threshold (0.01 in this case) within a short period (e.g., 10 seconds). This sudden change is typically not caused by conventional factors such as temperature fluctuations or pressure changes, but rather indicates a potential chemical change or impurity intrusion in the coolant. For example, the coolant may degrade due to prolonged use; fluorocarbon coolants may decompose to produce fluorides; or hydrocarbon or silicone coolants may be oxidized, leading to a deterioration in their dielectric properties. This sudden change is detected by continuously recording the dielectric loss factor value and calculating its change at preset time intervals (e.g., every 10 seconds). When Δtanδ exceeds the threshold of 0.01, the intelligent classification management system will trigger an early warning signal, alerting maintenance personnel to the coolant condition and potentially requiring coolant replacement or purification procedures to prevent potential thermal management failures and protect server components from damage.

[0146] It should be noted that the second degradation indicator can indicate fluoride decomposition, which is achieved by monitoring 1135 cm⁻¹. -1 The wavenumber shift of a nearby Raman spectral characteristic peak is used for assessment. This characteristic peak corresponds to the vibration of the CF bond in fluorocarbons and is a hallmark spectral feature of fluorocarbon coolants. When the fluoride in the coolant begins to decompose, the structure of the CF bond may change, causing a shift in the position of the characteristic peak at that specific wavenumber in the Raman spectrum. This shift is direct evidence that the coolant has begun to deteriorate, indicating that the CF bond may be broken due to thermal, optical, or chemical reactions, affecting the dielectric properties and thermal stability of the coolant. A wavenumber shift threshold is set to trigger an alarm. This means that if, during continuous monitoring, the wavenumber shift exceeds 1135 cm⁻¹, an alarm will be triggered. -1 The characteristic peak shift at wavenumber exceeds ±3cm -1 The system will automatically issue a warning, indicating that the coolant may have decomposed and requires further analysis or replacement.

[0147] It should be noted that the third degradation index can indicate the formation of acidic products, as shown by infrared spectroscopy analysis at 1700 cm⁻¹. -1 The presence of a new absorption peak at this point is used to determine the absorption frequency. This wavenumber region corresponds to the stretching vibration of the carbonyl group (C=O) and is the characteristic absorption region for many acidic products and carbonyl compounds. During the operation of an immersion liquid cooling system, the coolant may produce acidic products, such as organic acids or acetic acid, due to high temperature, oxidation, or other chemical reactions. These acidic products typically contain carbonyl structures, and their absorption characteristics in the infrared spectrum can be observed at 1700 cm⁻¹. -1 It was detected nearby. If at 1700cm... -1If the absorbance of a newly detected peak in the vicinity exceeds 0.05 Absorbance Units (AU), the system will trigger an alarm. Absorbance reflects the intensity of absorption of infrared light of a specific wavelength by the coolant. Increased absorbance indicates an increased concentration of acidic products, which may pose a risk of corrosion to server hardware and reduce the dielectric properties and heat dissipation efficiency of the coolant.

[0148] As can be seen from the above embodiments of this application, the system can effectively identify the deterioration status of the coolant by monitoring its dielectric and spectral characteristics, generate the first warning information in a timely manner, provide key decision support for maintenance personnel, and ensure the efficient and stable operation of the immersion liquid cooling system.

[0149] In an optional implementation, after obtaining the current dielectric characteristics and current spectral characteristics of the coolant, the process includes: obtaining the current thermal characteristics of the coolant, and calculating at least one degradation index of the coolant based on the current thermal characteristics; and generating a first prompt message when at least one degradation index meets the preset degradation conditions corresponding to each degradation index.

[0150] It should be noted that thermal characteristics are parameters reflecting the thermal properties of the coolant, such as thermal conductivity, specific heat capacity, and thermal stability, used to assess the coolant's heat exchange capacity and thermal stability. After acquiring the dielectric and spectral characteristics of the coolant, the system further acquires the current thermal characteristics of the coolant. This is accomplished through a series of thermal performance monitoring devices, including but not limited to miniature thermal conductivity sensors, temperature sensors, and heat flow meters, which can provide thermal property data of the coolant under current operating conditions. Subsequently, the system calculates at least one degradation index of the coolant based on these thermal characteristics, such as changes in thermal stability or a decrease in thermal conductivity. These indicators are used to determine whether the coolant has begun to age or become contaminated. If any calculated degradation index meets a preset degradation condition (usually exceeding a certain threshold), the system will generate an initial alert, prompting the management or automation system to take immediate action, such as replacing the coolant or initiating a cleaning procedure, to prevent further decline in cooling efficiency or damage to the system.

[0151] Through the above-described embodiments of this application, by real-time monitoring and analysis of the thermal characteristics of the coolant, combined with preset deterioration conditions, the immersion liquid cooling system can proactively identify the deterioration state of the coolant and take timely measures to ensure the continuous reliability and efficiency of the cooling process.

[0152] In an optional implementation, calculating at least one degradation index of the coolant based on current thermal characteristics includes: if the coolant is a first coolant, calculating the thermal conductivity of the first coolant based on current thermal characteristics and determining the thermal conductivity as a fourth degradation index, wherein the first coolant is used to indicate a coolant that remains liquid during the transfer of heat from the server; if the coolant is a second coolant, calculating the boiling point of the second coolant based on current thermal characteristics and determining the boiling point as a fifth degradation index, wherein the second coolant is used to indicate a coolant that undergoes liquid-to-gas transition during the transfer of heat from the server.

[0153] It should be noted that the first coolant refers to the coolant that remains liquid during the heat transfer process, typically used in single-phase liquid cooling systems. The second coolant refers to the coolant that undergoes a liquid-to-gas change during the heat transfer process, typically used in phase-change liquid cooling systems. The fourth degradation indicator can be the change in thermal conductivity of the first coolant (which remains liquid). The fifth degradation indicator can be the change in the boiling point of the second coolant (phase-change coolant).

[0154] The system calculates one or more indicators reflecting coolant performance degradation based on real-time monitored coolant thermal characteristics, such as decreased thermal conductivity or changes in boiling point. If the coolant is the primary coolant used for single-phase liquid cooling, the system calculates its thermal conductivity based on real-time thermal characteristics. Changes in thermal conductivity are identified as a degradation indicator, used to determine whether the primary coolant's heat exchange performance has declined due to contamination, deterioration, or other reasons. If the coolant is the secondary coolant used for phase-change liquid cooling, the system calculates its boiling point based on real-time thermal characteristics. Changes in boiling point are identified as another degradation indicator, used to determine whether the secondary coolant's phase-change cooling performance has declined due to changes in composition or deterioration.

[0155] It should be noted that the system assesses the coolant's performance status by monitoring its thermal characteristics. For the first coolant in a single-phase liquid cooling system, the system calculates its thermal conductivity and uses the change in thermal conductivity as the fourth degradation indicator to determine whether the first coolant still possesses good heat exchange performance. For the second coolant in a phase change liquid cooling system, the system calculates its boiling point and uses the change in boiling point as the fifth degradation indicator to determine whether the phase change cooling performance of the second coolant has deteriorated. The assessment of these two degradation indicators helps ensure that the immersion liquid cooling system can maintain efficient and stable cooling performance during server operation.

[0156] Through the above-described embodiments of this application, by real-time monitoring and analysis of the thermal characteristics of the coolant, combined with intelligent algorithm processing, this embodiment can effectively judge and respond to the deterioration of coolant performance, ensuring that the immersion liquid cooling system continues to operate efficiently in the data center, while maintaining the stability of server operation and the quality of coolant.

[0157] Example 5:

[0158] Figure 3 This is a schematic diagram of an optional immersion server according to an embodiment of this application; as shown... Figure 3 The system described can be a single-phase immersion liquid cooling system, comprising a cooling system, an outdoor cold source, a server, and a coolant. It achieves efficient heat transfer by directly immersing the server or its components entirely in a liquid cooling medium, making it suitable for modern computing devices requiring high heat dissipation efficiency. In a single-phase immersion liquid cooling system, the cooling medium (typically a non-conductive dielectric liquid, such as coolant) remains in a liquid state and does not undergo a phase change (from liquid to gas or vice versa).

[0159] During operation, the heat generated by the server is directly transferred to the surrounding liquid cooling medium via heat conduction. The system maintains the server hardware within its optimal operating temperature range by dynamically adjusting the coolant flow rate and temperature, as well as the cooling efficiency of the heat exchanger. This may include regulating the inlet and outlet temperatures of the coolant, and the cooling intensity at the heat exchanger.

[0160] Figure 4 This is a schematic diagram of an optional coolant management method according to an embodiment of this application; as shown Figure 4 As shown, a management method for the coolant management system of a single-phase immersion server is presented.

[0161] A dielectric sensor continuously acquires dielectric loss factor data of the coolant at two frequency points: 1 MHz and 10 MHz. This real-time, multi-frequency monitoring can capture minute changes in the dielectric properties of the coolant, providing timely warnings of coolant degradation or contamination. A Raman spectrometer simultaneously acquires the spectral information of the coolant, with an integration time of 200 ms. By analyzing the Raman spectrum, the system can identify characteristic peaks of specific compounds in the coolant, such as CH bond vibrations in hydrocarbons or Si-O-Si bond vibrations in organosilicon compounds, thereby determining the type and purity of the coolant. The acquired dielectric loss factor and Raman spectral data are transmitted in real time to an intelligent classification management system. This system compares the data with a preset coolant characteristic database and uses machine learning algorithms for data fusion analysis to identify the type of coolant and determine whether it meets preset performance standards. A steady-state thermal conductivity measurement device continuously monitors the thermal conductivity of the coolant, comparing it with initial or standard thermal conductivity to assess whether the coolant's heat transfer efficiency has decreased. A decrease in thermal conductivity may indicate coolant contamination, aging, or compositional changes. The intelligent classification management system (terminal equipment) determines the state of the coolant based on the analysis results of dielectric loss factor, Raman spectral characteristics, and thermal conductivity. If any monitored parameter deviates from a predetermined threshold, the system will issue an early warning, prompting operators to replace or purify the coolant to maintain the heat dissipation efficiency and safety of the server immersion liquid cooling system. After the warning is issued, the system may activate dynamic control mechanisms, such as adjusting the coolant flow rate, temperature, or pressure, to adapt to changes in the coolant state and maintain optimal heat dissipation performance.

[0162] Example 6:

[0163] Figure 5 This is a schematic diagram of another optional immersion server according to an embodiment of this application; as shown Figure 5 As shown, this could be a phase change immersion liquid cooling system, including a cooling system, an outdoor cold source, a server, and a coolant. Unlike single-phase liquid cooling systems, phase change immersion liquid cooling utilizes the process of a liquid cooling medium absorbing heat and then transforming into a gaseous state to dissipate heat. This process involves a phase transition of the cooling medium.

[0164] Figure 6 This is a schematic diagram of another optional coolant management method according to an embodiment of this application; as shown Figure 6 As shown, a management method for the coolant system of a phase change immersion server is presented.

[0165] Pulsed capacitive probes monitor changes in the dielectric constant of the coolant by emitting short-cycle electrical pulse signals, thus reflecting changes in the coolant content. Mid-infrared attenuated total reflectance (ATR) probes typically operate in the 2.5 to 25 μm range. ATR probes utilize infrared spectroscopy to analyze the vibration and rotation of coolant molecules, identifying the chemical composition and changes in the coolant, particularly in tracking the degradation of fluorocarbon coolants. Boiling point dynamic tracking devices (boiling point detectors) monitor the boiling point of the coolant in real time. Pulsed capacitive probes are placed inside an immersion liquid cooling tank, but not in direct contact with steam; instead, they are in close contact with the liquid coolant through a steam barrier membrane. The steam barrier membrane prevents steam from directly entering the capacitive probe and affecting the measurement results, while allowing the capacitive probe to accurately measure the dielectric properties of the liquid coolant. Mid-infrared attenuated total reflectance (ATR) probes are also placed inside the cooling tank, but their location needs careful selection to ensure contact with the liquid coolant while being away from areas of high steam density, in order to more accurately detect the chemical composition of the coolant. High-precision temperature sensors are deployed at multiple key locations within the cooling tank, including the bottom of the tank, near the condenser, and specific points along the coolant circulation path. These sensors continuously monitor temperature changes in the coolant, especially in areas prone to phase transitions, such as hot spots near the CPU or GPU. By analyzing temperature data, the system automatically identifies the boiling point at which the coolant changes from liquid to gas and continuously tracks this temperature over time. The intelligent classification and management system analyzes the thermal stability and phase transition performance of the coolant based on dynamic changes in the boiling point. An increase or decrease in the boiling point may indicate changes in coolant purity, compositional fluctuations, or performance degradation. When a boiling point change exceeds a preset threshold, the system triggers an alert, indicating potential coolant deterioration and providing corresponding maintenance recommendations, such as replacing the coolant or checking the condenser's proper functioning.

[0166] In an optional implementation, obtaining the current dielectric characteristics and current spectral characteristics of the coolant includes: estimating based on candidate dielectric characteristics and candidate spectral characteristics to obtain predicted dielectric characteristics and predicted spectral characteristics, wherein the candidate dielectric characteristics are used to indicate the dielectric characteristics of the coolant over a period of time, and the candidate spectral characteristics are used to indicate the spectral characteristics of the coolant over a period of time; calculating a first difference between the predicted dielectric characteristics and a first dielectric characteristic, and a second difference between the predicted spectral characteristics and the first spectral characteristics, wherein the first dielectric characteristics are used to indicate the measured dielectric characteristics of the coolant, and the first spectral characteristics are used to indicate the measured spectral characteristics of the coolant; determining the current dielectric characteristics based on the predicted dielectric characteristics and the first difference, and determining the current spectral characteristics based on the predicted spectral characteristics and the second difference.

[0167] It should be noted that the predicted dielectric characteristics are the result of predicting the dielectric properties at future times using an adaptive Kalman filter algorithm, based on a series of current and past data points. Candidate spectral characteristics are a sequence of spectral features over a period of time based on recent spectral measurements, reflecting the dynamic changes in the coolant's chemical composition. Predicted spectral characteristics are future spectral feature values ​​predicted by the Kalman filter algorithm based on historical spectral data and the current measurement state. The first dielectric characteristic and the first spectral characteristic refer to the actual dielectric and spectral characteristics of the coolant directly measured by the sensor at the current time point. The first difference and the second difference correspond to the deviations between the currently measured dielectric and spectral feature values ​​and the Kalman filter predictions, respectively.

[0168] It is important to note that in immersion liquid cooling systems, especially phase change liquid cooling systems, the coolant undergoes a cyclic phase change from liquid to gas and back to liquid to absorb and dissipate heat generated by electronic devices. During this process, air bubbles may become trapped in the coolant. The presence of these bubbles significantly interferes with the coolant flow rate, thermal conductivity, and dielectric properties of the liquid cooling system. Specifically, the thermal conductivity of air bubbles is much lower than that of the coolant; their presence forms an insulating layer, hindering effective heat transfer and thus reducing cooling efficiency. Air bubbles alter the dielectric constant and dielectric loss of the coolant, which affects the accuracy of any detection that relies on dielectric properties, such as electric field sensing or capacitance measurements. The presence of air bubbles can cause anomalies in optical measurements (such as Raman spectroscopy and light absorption measurements) because air bubbles scatter or block light, causing fluctuations and uncertainties in the measurement signal.

[0169] In optional implementations, in immersion liquid cooling systems, the dielectric and spectral characteristics of the coolant are directly related to its cooling efficiency and insulation performance, making them key parameters for real-time monitoring of system health. However, due to complex factors during system operation, such as the random generation and disappearance of bubbles, actual measurements often exhibit fluctuations and errors. To obtain more accurate current dielectric and spectral characteristics, this invention employs an adaptive Kalman filter algorithm for data processing. First, based on candidate dielectric and spectral characteristics of the coolant over a period of time, predicted dielectric and spectral characteristics are obtained. This step utilizes the predictive ability of the Kalman filter to assess system state. Next, by calculating the first and second differences between the predicted and current measured values, the real-time measurement data is compared with the predicted data to evaluate the measurement error. Finally, combining the difference between the prediction and measurement, the Kalman filter adjusts its parameters to minimize the error, determining more accurate current dielectric and spectral characteristics, thus achieving real-time data correction and noise reduction.

[0170] Specifically, dielectric and spectral sensors are installed to continuously acquire dielectric and spectral data of the coolant. Historical data construction involves collecting dielectric and spectral characteristic data over a period of time to form a candidate feature sequence. This data can be continuous measurement results from the system under normal operating conditions. The initial state, state transition matrix, and observation matrix of the filter, as well as the covariance matrices of state noise and observation noise, are set. Using the filter output and state transition matrix from the previous time step, the dielectric and spectral characteristics for the next time step are predicted. Simultaneously, the first dielectric and spectral characteristics are acquired in real time. The predicted features are compared with the measured features, and the first and second differences between the dielectric and spectral characteristics are calculated to evaluate the measurement error. Using the Kalman gain, the filter state estimate is corrected based on the differences, optimizing the parameters of the prediction model. Through continuous prediction, measurement, and adjustment, the Kalman filter can automatically identify and suppress measurement fluctuations caused by bubbles, etc.

[0171] Example 7:

[0172] Assuming that a certain amount of air bubbles are generated in the coolant during the operation of the immersion liquid cooling system, these bubbles will interfere with the readings of the dielectric and spectral sensors. The system uses an adaptive Kalman filter algorithm for real-time monitoring and correction.

[0173] Over the past ten minutes, the system has continuously monitored the dielectric properties (dielectric constant ε=2.5, dielectric loss factor tanδ=0.002) and spectral properties (Raman spectral characteristic value S=10000 counts, infrared spectral characteristic value I=0.7AU).

[0174] The initial state is set as the average of the past ten measurements. The state transition matrix A is set according to the physical properties of the coolant, the observation matrix H is set as the identity matrix, and the state and observation noise covariance matrices are estimated based on the coolant environmental conditions and sensor performance. Based on the current state and the measurement results of the previous moment, the dielectric characteristics ε' and spectral characteristics S' and I' of the next moment are predicted as follows: ε=2.5, tanδ=0.002, S=10000counts, I=0.7AU.

[0175] At the next moment of the measurement phase, the dielectric and spectral characteristics measured by the sensor are ε=2.55, tanδ=0.003, S=10100counts, and I=0.71AU, respectively. It can be seen that the bubble caused a slight fluctuation in the measurement results.

[0176] The differences between the predicted and measured values ​​are calculated as Δε=0.05, Δtanδ=0.001, ΔS=100counts, and ΔI=0.01AU, respectively.

[0177] Update phase:

[0178] The Kalman gain K is calculated, and its magnitude depends on the current measurement error and the estimated system prediction error. Based on Δε and Δtanδ, the readings of the electrochemical impedance spectroscopy sensor are adjusted to update the predicted dielectric characteristics. Based on ΔS and ΔI, the measurement results of the miniature fiber optic spectral probe are corrected to update the predicted spectral characteristics.

[0179] Through the above-described embodiments of this application, by adaptively adjusting the parameters of the Kalman filter, the system can identify and eliminate measurement fluctuations caused by bubbles in real time, resulting in more accurate determination of the current dielectric and spectral characteristics, approaching the ideal measurement results without bubble interference. Even when bubbles appear in the coolant, it still provides stable and reliable monitoring of dielectric and spectral characteristics, providing a solid data foundation for dynamic control and maintenance decisions of immersion liquid cooling systems.

[0180] In an optional implementation, by automatically identifying and adjusting the coolant composition, coolant recycling and self-cleaning can be achieved simultaneously, reducing maintenance costs and environmental impact. The system can automatically adjust the coolant circulation rate and direction based on real-time monitoring of the coolant content and the server's heat load, optimizing heat exchange efficiency.

[0181] The system uses a multi-sensor fusion detection module (such as dielectric sensors, Raman spectroscopy probes, and infrared spectroscopy units) to collect real-time data on the content of hydrocarbon, organosilicon, and fluorocarbon coolants in the coolant. This data is transmitted wirelessly or via wired connection to the edge computing node of the intelligent classification management system (terminal device). Temperature sensors and heat flow meters installed near server components continuously monitor the server's heat generation, including the temperature and heat load of key components such as the CPU, GPU, and memory modules. Simultaneously, it calculates predicted heat load values ​​in real-time based on the server's operating load (such as processor utilization and network traffic). Using a pre-trained machine learning model, the system analyzes the coolant content data to identify whether the coolant component ratio deviates from preset thresholds and whether there are signs of contaminants or coolant degradation. The system combines real-time temperature data and server load information to assess the current heat load and predict future heat load trends based on historical data and server operating modes. Based on a comprehensive analysis of coolant content and heat load, the system generates an optimized control strategy. For example, when a decrease in the content of fluorocarbon working fluid in the coolant is detected and the server's heat load increases, it may be necessary to increase the coolant circulation speed to enhance cooling capacity; if the heat load decreases, the circulation speed can be appropriately reduced to save energy.

[0182] The intelligent classification and management system sends control signals to the circulation pump, dynamically adjusting the pump speed according to the optimization strategy, thereby changing the coolant circulation rate. If improved cooling efficiency is needed, the system instructs the pump to increase its speed and increase the coolant flow rate; conversely, it reduces the speed to minimize unnecessary energy consumption. For complex liquid cooling systems, multi-way valves or intelligent distributors may be needed to control the coolant flow direction, ensuring that areas with high heat density are cooled preferentially. The system dynamically adjusts the coolant distribution strategy based on the server's heat load distribution, guiding more coolant to high-heat-load areas for more efficient heat management. The system continuously monitors the effects of adjustments, including coolant temperature changes and server hardware temperature drops. If the adjusted cooling effect does not meet expectations, the system automatically reanalyzes the data and performs further optimization until optimal cooling is achieved.

[0183] When the coolant content falls below a safe threshold or the server's heat load exceeds the cooling system's design limits, the intelligent classification management system will immediately trigger an early warning mechanism, notifying maintenance personnel to take emergency measures, such as replenishing coolant or temporarily reducing the server load. Based on the degree of coolant degradation and server heat load prediction, the system can issue maintenance recommendations in advance, such as regularly replacing the coolant, adding cleaning procedures, or optimizing the configuration of the cooling subsystem to prevent potential heat dissipation failures.

[0184] Through the above implementation methods, the immersion liquid cooling system can automatically adjust the circulation rate and direction of the coolant based on real-time monitoring of the coolant content and server heat load. This not only improves cooling efficiency but also enhances the system's adaptability and reliability. It contributes to efficient energy utilization and the long-term stable operation of server hardware.

[0185] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method.

[0186] Embodiments of this application also provide a coolant management device. Figure 7 This is a structural block diagram of an optional coolant management device according to an embodiment of this application, such as... Figure 7 As shown, the device includes:

[0187] The feature acquisition module 702 is used to acquire the current dielectric characteristics and current spectral characteristics of the coolant, wherein the coolant is used to immerse the server and absorb and transfer the server's heat.

[0188] The content calculation module 704 is used to calculate the current content of the coolant based on the current dielectric characteristics and the current spectral characteristics, wherein the current content is used to indicate the proportion of each of the at least one cooling working fluid included in the coolant.

[0189] The content adjustment module 706 is used to adjust the current content of the coolant to the target content when the current content meets the preset content threshold condition.

[0190] Optionally, the content calculation module 704 is further configured to: calculate the type of at least one cooling medium included in the coolant and the first content corresponding to each of the at least one cooling medium based on the current spectral characteristics, wherein the current spectral characteristics are used to indicate the absorption or emission characteristics of the coolant to light; determine the second content corresponding to each of the at least one cooling medium based on the current dielectric characteristics and the reference dielectric characteristics corresponding to each of the at least one cooling medium, wherein the current dielectric characteristics are used to indicate the response of the coolant to an electric field; and calculate the current content based on the first content and the second content.

[0191] Optionally, the content calculation module 704 is further configured to: determine the number of characteristic peaks in the current spectral features as the number of at least one type of cooling medium, wherein the characteristic peaks are used to indicate local maximum points in the current spectral features; determine the type of at least one cooling medium based on the wavenumber corresponding to each of the at least one characteristic peaks; and calculate the proportion corresponding to each of the at least one cooling medium based on the peak value of each of the at least one characteristic peaks.

[0192] Optionally, the content calculation module 704 is further configured to: determine at least one Raman characteristic peak in the current spectral features when the current spectral features indicate the Raman spectrum of the coolant; calculate the peak area corresponding to each of the at least one Raman characteristic peak, and determine the proportion of the at least one peak area as the proportion of each of the at least one cooling medium; determine at least one absorption characteristic peak in the current spectral features when the current spectral features indicate the infrared spectrum of the coolant; calculate the absorbance corresponding to each of the at least one absorption characteristic peak, and determine the proportion of the at least one absorbance as the proportion of each of the at least one cooling medium.

[0193] Optionally, the content calculation module 704 is further configured to: determine the reference dielectric characteristics corresponding to each of the at least one cooling working fluid; determine the weights corresponding to each of the at least one reference dielectric characteristics based on the at least one reference dielectric characteristic and the current dielectric characteristic, wherein the current dielectric characteristic is obtained by weighted summation of the at least one reference dielectric characteristic; and determine the ratio of the weights corresponding to each of the at least one reference dielectric characteristic as the second content.

[0194] Optionally, the content calculation module 704 is further configured to: acquire candidate dielectric features of the coolant, wherein the candidate dielectric features are used to indicate the dielectric features of the coolant over a period of time; calculate the dielectric change rate of the coolant based on the current dielectric features and the candidate dielectric features; and calculate the current content of the coolant based on the current dielectric features and the current spectral features when the dielectric change rate is greater than a preset change threshold.

[0195] Optionally, the feature acquisition module 702 is further configured to: calculate at least one degradation index of the coolant based on the current dielectric characteristics and the current spectral characteristics, wherein the degradation index is a parameter used to indicate the decline in coolant performance or the deterioration in quality; and generate a first prompt message when at least one degradation index meets the preset degradation conditions corresponding to each degradation index.

[0196] Optionally, the feature acquisition module 702 is further configured to: calculate the dielectric constant of the coolant based on the current dielectric characteristics, and determine the ratio of the imaginary part to the real part of the dielectric constant as a first degradation index, wherein the dielectric constant is used to indicate the capacitance of the coolant; calculate the first shift of the first characteristic peak based on the current spectral characteristics, and determine the first shift as a second degradation index, wherein the first characteristic peak is used to indicate a decomposable cooling medium; determine the first absorbance of the second characteristic peak when the current spectral characteristics indicate the infrared spectrum of the coolant, and determine the first absorbance as a third degradation index; generate a first prompt message when the rate of change of the first degradation index is greater than the rate of change of a preset factor; generate a first prompt message when the second degradation index is greater than the preset shift; and generate a first prompt message when the third degradation index is greater than the preset absorbance.

[0197] Optionally, the feature acquisition module 702 is further configured to: acquire the current thermal energy characteristics of the coolant, and calculate at least one degradation index of the coolant based on the current thermal energy characteristics; and generate a first prompt message when at least one degradation index meets the preset degradation conditions corresponding to each degradation index.

[0198] Optionally, the feature acquisition module 702 is further configured to: calculate the thermal conductivity of the first coolant based on the current thermal characteristics when the coolant is a first coolant, and determine the thermal conductivity as a fourth degradation index, wherein the first coolant is used to indicate a coolant that remains liquid during the process of transferring heat from the server; and calculate the boiling point of the second coolant based on the current thermal characteristics when the coolant is a second coolant, and determine the boiling point as a fifth degradation index, wherein the second coolant is used to indicate a coolant that achieves liquid-to-gas change during the process of transferring heat from the server.

[0199] Optionally, the feature acquisition module 702 is further configured to: estimate based on candidate dielectric features and candidate spectral features to obtain predicted dielectric features and predicted spectral features, wherein the candidate dielectric features are used to indicate the dielectric features of the coolant over a period of time, and the candidate spectral features are used to indicate the spectral features of the coolant over a period of time; calculate a first difference between the predicted dielectric features and a first dielectric feature, and a second difference between the predicted spectral features and the first spectral features, wherein the first dielectric feature is used to indicate the measured dielectric features of the coolant, and the first spectral feature is used to indicate the measured spectral features of the coolant; determine the current dielectric feature based on the predicted dielectric features and the first difference, and determine the current spectral feature based on the predicted spectral features and the second difference.

[0200] Optionally, the content adjustment module 706 is further configured to: adjust the coolant pressure parameter to the target pressure when the coolant pressure parameter meets the preset adjustment conditions; and adjust the coolant level to the target level when the coolant level meets the preset adjustment conditions.

[0201] For a description of the features in the embodiment corresponding to the coolant management device, please refer to the relevant description in the embodiment corresponding to the coolant management method, which will not be repeated here.

[0202] Embodiments of this application also provide an electronic device, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the above embodiments of the coolant management method.

[0203] Embodiments of this application also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the above embodiments of the coolant management method when it is run.

[0204] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0205] Embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the above-described embodiments of the coolant management method.

[0206] Embodiments of this application also provide another computer program product, including a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in any of the above-described embodiments of the coolant management method.

[0207] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0208] The foregoing has provided a detailed description of a coolant management method and apparatus, storage medium, and electronic device provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only intended to aid in understanding the method and core ideas of this application. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of the claims of this application.

Claims

1. A method for managing coolant, characterized in that, include: Obtain the current dielectric and spectral characteristics of the coolant, wherein the coolant is used to immerse the server and absorb and transfer heat from the server; The current content of the coolant is calculated based on the current dielectric characteristics and the current spectral characteristics, wherein the current content is used to indicate the proportion of each of the at least one cooling working fluid included in the coolant; If the current content meets the preset content threshold condition, the current content of the coolant is adjusted to the target content, wherein the target content is the specific component content required according to the optimal operating state of the system; The step of calculating the current content of the coolant based on the current dielectric characteristics and the current spectral characteristics includes: Obtain candidate dielectric features of the coolant, wherein the candidate dielectric features are used to indicate the dielectric features of the coolant over a period of time; The dielectric change rate of the coolant is calculated based on the current dielectric characteristics and the candidate dielectric characteristics; If the dielectric change rate is greater than a preset change threshold, the current content of the coolant is calculated based on the current dielectric characteristics and the current spectral characteristics. The step of obtaining the current dielectric characteristics and current spectral characteristics of the coolant includes: Calculate at least one degradation index of the coolant based on the current dielectric characteristics and the current spectral characteristics, wherein the degradation index is a parameter used to indicate a decrease in the performance or deterioration in the quality of the coolant; If at least one degradation index meets the preset degradation conditions corresponding to each degradation index, a first prompt message is generated.

2. The method according to claim 1, characterized in that, The step of calculating the current content of the coolant based on the current dielectric characteristics and the current spectral characteristics includes: The type of at least one cooling medium included in the coolant and the first content of each of the at least one cooling medium are calculated based on the current spectral characteristics, wherein the current spectral characteristics are used to indicate the light absorption or emission characteristics of the coolant. The second content of each of the at least one cooling working fluid is determined based on the current dielectric characteristic and the reference dielectric characteristic corresponding to each of the at least one cooling working fluid, wherein the current dielectric characteristic is used to indicate the response of the coolant to an electric field; The current content is calculated based on the first content and the second content.

3. The method according to claim 2, characterized in that, The step of calculating the type of at least one cooling medium included in the coolant and the first content corresponding to each of the at least one cooling medium based on the current spectral characteristics includes: The number of characteristic peaks in the current spectral features is determined as the number of types of the at least one cooling medium, wherein the characteristic peaks are used to indicate local maxima in the current spectral features; The type of the at least one cooling medium is determined based on the wavenumber corresponding to each of the at least one characteristic peak; The proportion of each of the at least one cooling working fluid is calculated based on the peak value of at least one of the characteristic peaks.

4. The method according to claim 3, characterized in that, The step of calculating the proportion of each of the at least one cooling working fluid based on the peak value of at least one of the characteristic peaks includes one of the following: When the current spectral features indicate the Raman spectrum of the coolant, at least one Raman characteristic peak in the current spectral features is determined; the peak area corresponding to each of the at least one Raman characteristic peak is calculated, and the proportion of the at least one peak area is determined as the proportion corresponding to each of the at least one cooling medium; In cases where the current spectral features indicate the infrared spectrum of the coolant, at least one absorption characteristic peak in the current spectral features is determined; Calculate the absorbance corresponding to each of the at least one absorption characteristic peak, and determine the proportion of each of the at least one absorbance as the proportion of each of the at least one cooling working fluid.

5. The method according to claim 2, characterized in that, Determining the second content of each of the at least one cooling working fluids based on the current dielectric characteristics and the reference dielectric characteristics corresponding to each of the at least one cooling working fluids includes: Determine the reference dielectric characteristics corresponding to each of the at least one cooling working fluid; The weights corresponding to each of the at least one of the reference dielectric features are determined based on at least one of the reference dielectric features and the current dielectric feature, wherein the current dielectric feature is obtained by weighted summation of the at least one of the reference dielectric features; The weight value corresponding to each of the at least one of the reference dielectric features is determined as the second content.

6. The method according to claim 1, characterized in that, The calculation of at least one degradation index of the coolant based on the current dielectric characteristics and the current spectral characteristics includes at least one of the following: The dielectric constant of the coolant is calculated based on the current dielectric characteristics, and the ratio of the imaginary part to the real part of the dielectric constant is determined as the first degradation index, wherein the dielectric constant is used to indicate the capacitance of the coolant; The first shift of the first characteristic peak is calculated based on the current spectral characteristics, and the first shift is determined as the second degradation index, wherein the first characteristic peak is used to indicate the decomposable cooling medium; When the current spectral characteristics indicate the infrared spectrum of the coolant, the first absorbance of the second characteristic peak is determined, and the first absorbance is determined as the third degradation index; When at least one degradation indicator satisfies the preset degradation condition corresponding to each degradation indicator, a first prompt message is generated, including at least one of the following: If the rate of change of the first degradation index is greater than the rate of change of the preset factor, a first prompt message is generated. If the second degradation index is greater than the preset offset, a first prompt message is generated; If the third degradation index is greater than the preset absorbance, a first prompt message is generated.

7. The method according to claim 1, characterized in that, After obtaining the current dielectric and spectral characteristics of the coolant, the process includes: Obtain the current thermal characteristics of the coolant; When the coolant is a first coolant, the thermal conductivity of the first coolant is calculated based on the current thermal characteristics, and the thermal conductivity is determined as a fourth degradation index, wherein the first coolant is used to indicate the coolant that remains liquid during the heat transfer of the server. When the coolant is a second coolant, the boiling point of the second coolant is calculated based on the current thermal characteristics, and the boiling point is determined as the fifth degradation index. The second coolant is used to indicate the coolant that achieves liquid-to-gas change during the heat transfer of the server. The first prompt message is generated when at least one degradation index satisfies the preset degradation condition corresponding to each degradation index.

8. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the steps of the coolant management method as described in any one of claims 1 to 7.

Images