Liquid cooling impurity monitoring method and system

Abstract

The present invention provides a liquid cooling impurity monitoring method and system, which obtains the existing liquid cooling impurity attribute set through spectral analysis and microscopic imaging, deploys an impurity detection module in the liquid cooling channel and obtains the liquid cooling channel resonance spectrum diagram and liquid cooling channel environmental parameters, performs collaborative feature extraction on the liquid cooling channel resonance spectrum diagram and liquid cooling channel environmental data, generates a liquid cooling ecological vector, obtains the historical liquid cooling ecological vector corresponding to the liquid cooling channel historical data, constructs a liquid cooling impurity monitoring model and performs model pre-training to generate a liquid cooling impurity library, inputs the liquid cooling ecological vector into the liquid cooling impurity monitoring model, matches the liquid cooling ecological vector with the liquid cooling impurity library based on the cuckoo algorithm, obtains a liquid cooling impurity combination vector, obtains a pre-trained impurity purification decision model, inputs the liquid cooling impurity combination vector into the impurity purification decision model, generates an optimal liquid cooling impurity purification strategy, and the user can remove impurities from the liquid cooling system according to the optimal liquid cooling purification strategy.

Description

Technical Field

[0001] The present invention relates to the technical field of liquid cooling impurity monitoring, and in particular to a liquid cooling impurity monitoring method and system. Background Art

[0002] With the rapid development of high-performance computing, data centers, and new energy batteries, liquid cooling technology has been widely adopted due to its efficient heat dissipation capabilities. However, over long-term operation, coolant is susceptible to factors such as metal corrosion, particulate contamination, and microbial growth, leading to the accumulation of impurities, which can cause pipe blockages, reduced heat transfer efficiency, and even equipment failure. Therefore, real-time monitoring of coolant impurity status is crucial to ensuring system reliability and extending service life.

[0003] In summary, there is an urgent need for a liquid cooling impurity monitoring method and system that can accurately identify impurity types and contamination levels in real time and automatically generate optimization solutions to break through the limitations of existing technologies and meet the high reliability requirements of liquid cooling data centers. Summary of the Invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a liquid cooling impurity monitoring method, the method comprising:

[0005] S1: Obtain the existing liquid-cooled impurity property set through spectral analysis and microscopic imaging;

[0006] S2: Deploy an impurity detection module in the liquid cooling channel and obtain the liquid cooling channel resonance spectrum and liquid cooling channel environmental parameters based on the impurity detection module;

[0007] S3: Perform collaborative feature extraction on the liquid cooling channel resonance spectrum and the liquid cooling channel environment data to generate a liquid cooling ecological vector, wherein the liquid cooling ecological vector includes a main frequency band offset, a liquid cooling attenuation rate, a phase offset, and liquid cooling channel environment parameters;

[0008] S4: Obtain the historical liquid cooling ecological vector corresponding to the historical data of the liquid cooling channel, build a liquid cooling impurity monitoring model, input the existing liquid cooling impurity attribute set and the historical liquid cooling ecological vector into the liquid cooling impurity monitoring model for pre-training, and generate a liquid cooling impurity library;

[0009] S5: Input the liquid cooling ecological vector into a pre-trained liquid cooling impurity monitoring model, perform liquid cooling impurity matching between the liquid cooling ecological vector and the liquid cooling impurity library based on the cuckoo algorithm, and obtain a liquid cooling impurity combination vector, wherein the liquid cooling impurity combination vector includes a liquid cooling impurity type and an impurity monitoring factor;

[0010] S6: Obtain a pre-trained impurity purification decision model, input the liquid cooling impurity combination vector into the impurity purification decision model, and generate an optimal liquid cooling impurity purification strategy. The user can remove impurities from the liquid cooling system according to the optimal liquid cooling purification strategy.

[0011] As a further solution of the present invention, the method of obtaining a historical liquid cooling ecological vector corresponding to the historical data of the liquid cooling channel, constructing a liquid cooling impurity monitoring model, inputting the existing liquid cooling impurity attribute set and the historical liquid cooling ecological vector into the liquid cooling impurity monitoring model for pre-training, and generating a liquid cooling impurity library includes:

[0012] Based on the bidirectional attention mechanism, a dynamic association network of historical liquid cooling ecological vectors and existing liquid cooling impurity attribute sets is constructed, and bidirectional interactive calculations are performed on the historical liquid cooling ecological vectors and the existing liquid cooling impurity attribute sets to obtain interactive calculation results. Based on the interactive calculation results, the liquid cooling channel resonance spectrum diagram is dimensionally aligned with the existing liquid cooling impurity attribute set;

[0013] Performing environmental interference correction on the main frequency band offset, the liquid cooling attenuation rate, and the phase offset according to a liquid cooling compensation mechanism;

[0014] A mapping relationship between the liquid cooling impurity attributes included in the existing liquid cooling impurity attribute set and the main frequency band offset, liquid cooling attenuation rate and phase offset after liquid cooling compensation is obtained, and a liquid cooling impurity library is constructed based on the mapping relationship.

[0015] As a further solution of the present invention, performing environmental interference correction on the main frequency band offset, the liquid cooling attenuation rate, and the phase offset according to the liquid cooling compensation mechanism includes:

[0016] Constructing a linear regression function based on the liquid cooling channel environmental parameters included in the historical liquid cooling ecological vector, quantifying the linear perturbation of the liquid cooling channel environmental parameters on the main frequency band offset according to the linear regression function, obtaining a linear perturbation result, and correcting the main frequency band offset according to the linear perturbation result;

[0017] Analyzing the coupling relationship between the liquid cooling attenuation rate and the flow rate parameter in the liquid cooling channel environmental parameter by constructing a covariance matrix, obtaining a coupling result, and correcting the liquid cooling attenuation rate according to the coupling result;

[0018] A nonlinear drift function of the phase offset is constructed based on wavelet transform and Kalman filter operations, and the phase offset is corrected by the nonlinear drift function.

[0019] As a further solution of the present invention, the liquid cooling ecological vector is input into a pre-trained liquid cooling impurity monitoring model, and the liquid cooling ecological vector is matched with the liquid cooling impurity library based on the cuckoo algorithm to obtain the liquid cooling impurity combination vector, including:

[0020] Randomly generate multiple liquid-cooled cuckoos based on the liquid-cooled impurity library, each liquid-cooled cuckoo carries a set of cuckoo flight parameters, and the cuckoo flight parameters include a step size parameter, a direction vector, and a search radius;

[0021] The liquid-cooled cuckoo performs a random walk alternating between long and short steps based on the cuckoo flight parameters. During the walk, the liquid-cooled ecological vector and the liquid-cooled impurity similarity of each data point in the liquid-cooled impurity library are calculated in real time. An impurity matching score is generated based on the liquid-cooled impurity similarity, and the cuckoo flight parameters are dynamically adjusted according to the impurity matching score.

[0022] When the variance of the motion trajectory of the liquid-cooled cuckoo is continuously iterated below a preset convergence threshold, it is determined to be a stable solution state, the data node with the highest impurity matching score in the liquid-cooled impurity library is extracted, and a liquid-cooled impurity combination vector is constructed.

[0023] As a further embodiment of the present invention, the impurity monitoring factor includes:

[0024] The impurity detection factors include concentration factor, particle size distribution factor, confidence weight and abnormal flag. The concentration factor is used to indicate the concentration level of impurities in the coolant. The particle size distribution factor is used to indicate the average particle size of impurities in the coolant. The confidence weight is used to indicate the similarity of liquid cooling impurities calculated by the cuckoo algorithm. The abnormal flag is used to indicate the characteristic mark when the impurity is not matched in the liquid cooling impurity library.

[0025] As a further solution of the present invention, the method of obtaining a pre-trained impurity purification decision model, inputting the liquid cooling impurity combination vector into the impurity purification decision model, and generating an optimal liquid cooling impurity purification strategy, wherein the user can remove impurities from the liquid cooling system according to the optimal liquid cooling impurity purification strategy, includes:

[0026] Obtaining a liquid-cooled impurity purification parameter group, the liquid-cooled impurity purification parameter group including environmental constraint parameters of the liquid cooling system and power consumption of an impurity purification tool; matching a predefined purification rule library based on the liquid-cooled impurity combination vector to obtain a purification rule matching result; and combining the liquid-cooled impurity purification parameter group with the purification rule matching result to generate an impurity purification strategy candidate set;

[0027] Monte Carlo simulation is used to evaluate the purification efficiency, energy consumption and stability risk of the impurity purification strategy candidate set under timing constraints, and a non-dominated sorting algorithm is used to screen the optimal solution of the impurity purification strategy. Based on the optimal solution of the impurity purification strategy, the optimal liquid cooling impurity purification strategy is generated.

[0028] As a further solution of the present invention, the impurity detection module is deployed in the liquid cooling channel, and the liquid cooling channel resonance spectrum and liquid cooling channel environmental data are obtained based on the impurity detection module, including:

[0029] Acquiring the liquid cooling channel resonance frequency of the liquid cooling system in real time based on the acoustic sensor group included in the impurity detection module, dynamically superimposing the liquid cooling channel resonance frequency to generate a liquid cooling channel resonance spectrum diagram;

[0030] The environmental parameters of the liquid cooling channel are acquired in real time based on the environmental sensor group included in the impurity detection module. The environmental parameters of the liquid cooling channel include the temperature, flow rate and pressure parameters of the coolant in the liquid cooling channel.

[0031] As a further solution of the present invention, the method of obtaining the existing liquid-cooled impurity attribute set through spectral analysis and microscopic imaging includes:

[0032] Perform impurity analysis on the sample coolant based on spectral analysis to obtain the type and properties of organic impurities in the sample coolant;

[0033] Perform impurity analysis on the sample coolant based on microscopic imaging to obtain the type and properties of inorganic impurities in the sample coolant;

[0034] The sample coolants are represented by multiple groups of coolants of liquid cooling systems in different operating environments, and an existing liquid cooling impurity attribute set is constructed based on the organic impurity type and attribute and the inorganic impurity type and attribute.

[0035] As a further solution of the present invention, the collaborative feature extraction of the liquid cooling channel resonance spectrum and the liquid cooling channel environment data to generate the liquid cooling ecological vector includes:

[0036] Feature extraction is performed on the liquid cooling channel resonance spectrum to obtain the main frequency band offset, liquid cooling attenuation rate, and phase offset. The main frequency band offset, liquid cooling attenuation rate, and phase offset are combined with the liquid cooling channel environmental data to generate a liquid cooling ecological vector.

[0037] In another aspect, an embodiment of the present invention further provides a liquid cooling impurity monitoring system, comprising:

[0038] An impurity detection module, comprising an acoustic sensor group and an environmental sensor group, for obtaining a liquid cooling channel resonance spectrum and liquid cooling channel environmental data;

[0039] An acoustic sensor group, the acoustic sensor group is used to obtain the liquid cooling channel resonance frequency of the liquid cooling system, and dynamically superimpose the liquid cooling channel resonance frequency to generate a liquid cooling channel resonance spectrum diagram;

[0040] An environmental sensor group, the environmental sensor group is used to obtain temperature parameters, flow rate parameters and pressure parameters of the coolant in the liquid-cooling channel in real time, and combine the temperature parameters, the flow rate parameters and the pressure parameters into liquid-cooling channel environmental parameters;

[0041] A data presetting module, the data presetting module is used to obtain an existing liquid-cooled impurity attribute set through spectral analysis and microscopic imaging;

[0042] A collaborative extraction module is used to perform collaborative feature extraction on the liquid cooling channel resonance spectrum and the liquid cooling channel environment data to generate a liquid cooling ecological vector and obtain a historical liquid cooling ecological vector corresponding to the liquid cooling channel historical data;

[0043] A model building module, wherein the model building module is used to build a liquid cooling impurity monitoring model, pre-train the liquid cooling impurity monitoring model, and generate a liquid cooling impurity library;

[0044] An impurity matching module, configured to perform liquid cooling impurity matching between the liquid cooling ecological vector and the liquid cooling impurity library based on a cuckoo algorithm to obtain a liquid cooling impurity combination vector;

[0045] A decision generation module is used for the impurity purification decision model to generate an optimal liquid cooling purification strategy based on a liquid cooling impurity combination vector.

[0046] Based on the above aspects, the embodiment of the present application first obtains the existing liquid cooling impurity attribute set through spectral analysis and microscopic imaging, obtains and integrates the impurity types and corresponding attributes that may be generated in the liquid cooling system, and expands the scope of subsequent impurity matching monitoring;

[0047] Subsequently, based on the impurity detection module, the liquid cooling channel resonance spectrum and liquid cooling channel environmental parameters are obtained, and collaborative feature extraction is performed on the liquid cooling channel resonance spectrum and liquid cooling channel environmental data to generate a liquid cooling ecological vector. The historical liquid cooling ecological vector corresponding to the historical data of the liquid cooling channel is obtained, and a liquid cooling impurity monitoring model is constructed. The existing liquid cooling impurity attribute set and the historical liquid cooling ecological vector are input into the liquid cooling impurity monitoring model for pre-training to generate a liquid cooling impurity library. The various liquid cooling channel resonance frequencies in the liquid cooling channel resonance spectrum are aligned with the impurity attributes in the existing liquid cooling impurity attribute set, so that no complicated operations are required for subsequent impurity matching;

[0048] Finally, the liquid cooling ecological vector is input into the pre-trained liquid cooling impurity monitoring model. Based on the cuckoo algorithm, the liquid cooling ecological vector is matched with the liquid cooling impurity library for liquid cooling impurities to obtain a liquid cooling impurity combination vector. The liquid cooling impurity combination vector includes the liquid cooling impurity type and the impurity monitoring factor. A pre-trained impurity purification decision model is obtained, and the liquid cooling impurity combination vector is input into the impurity purification decision model to generate the optimal liquid cooling impurity purification strategy. Users can remove impurities from the liquid cooling system according to the optimal liquid cooling purification strategy. This method can improve the efficiency of detecting coolant impurities in the liquid cooling system. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1It is a schematic diagram of the execution flow of a liquid cooling impurity monitoring method provided by an embodiment of the present invention.

[0050] Figure 2 Schematic diagram of a liquid-cooled impurity monitoring system provided in an embodiment of the present invention.

[0051] Figure 3 Schematic diagram of an impurity detection module in a liquid-cooled impurity monitoring system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The present invention will be described in detail below with reference to the accompanying drawings. Figure 1 This is a schematic diagram of the execution flow of a liquid cooling impurity monitoring method provided by an embodiment of the present invention. The liquid cooling impurity monitoring method is introduced in detail below.

[0053] Step S1: Acquire the existing liquid cooling impurity attribute set through spectral analysis and microscopic imaging.

[0054] Specifically, impurity analysis is performed on the sample coolant based on spectral analysis to obtain the type and properties of organic impurities in the sample coolant, and impurity analysis is performed on the sample coolant based on microscopic imaging to obtain the type and properties of inorganic impurities in the sample coolant. The sample coolant is represented as multiple groups of liquid cooling system coolants under different operating environments, and an existing liquid cooling impurity property set is constructed based on the organic impurity types and properties and the inorganic impurity types and properties.

[0055] For example, a liquid cooling system is in the cold start phase at 10℃, and the coolant flow rate in the liquid cooling channel is 0.8m / s. The organic impurities in the coolant in the liquid cooling channel under the running state are detected by Fourier transform infrared spectroscopy. A Si-CH3 characteristic peak is detected at the coolant, confirming that it is low-temperature precipitation of the silicone oil additive. The corresponding organic impurity type and properties are output in the form of a vector, that is, the organic impurity vector = {Type: polydimethylsiloxane, Properties: precipitation temperature threshold -15°C, surface tension 18 mN / m}. Confocal microscopy is used to observe irregular particles of 10-50 μm in the coolant. Combined with laser-induced breakdown spectroscopy, it is shown that the Fe content is 70%, the Cr content is 18%, and the Ni content is 12%. The inorganic impurity is determined to be wear debris from the stainless steel pump body. The corresponding inorganic impurity type and properties are output in the form of a vector, that is, the inorganic impurity vector = {Type: 316L stainless steel, Properties: weak magnetic response, density 7.9 g / cm³}. The organic impurity vector and the inorganic impurity vector are combined to generate a liquid cooling impurity property record, that is, A{Environmental label: "Low-temperature cold start", Organic matter: "Silicone oil_precipitation -15°C", Inorganic matter: "316L steel chips_non-magnetic", Purification priority: 2}.

[0056] Step S2: deploy an impurity detection module in the liquid cooling channel, and obtain a liquid cooling channel resonance spectrum and liquid cooling channel environmental parameters based on the impurity detection module.

[0057] Specifically, based on the acoustic sensor group included in the impurity detection module, the liquid cooling channel resonance frequency of the liquid cooling system is obtained in real time, and the liquid cooling channel resonance frequency is dynamically superimposed to generate a liquid cooling channel resonance spectrum diagram.

[0058] Specifically, the environmental parameters of the liquid cooling channel are acquired in real time based on the environmental sensor group included in the impurity detection module. The liquid cooling channel environmental parameters include the temperature, flow rate and pressure parameters of the coolant in the liquid cooling channel.

[0059] Step S3: Perform collaborative feature extraction on the liquid cooling channel resonance spectrum and the liquid cooling channel environment data to generate a liquid cooling ecological vector. The liquid cooling ecological vector includes a main frequency band offset, a liquid cooling attenuation rate, a phase offset, and liquid cooling channel environment parameters.

[0060] In this embodiment, step S3 includes:

[0061] Step S31 , extracting features from the liquid-cooled channel resonance spectrum to obtain the main frequency band offset, liquid cooling attenuation rate, and phase offset.

[0062] Specifically, feature extraction is performed on the liquid-cooled channel resonance spectrum, and multi-window spectrum analysis is performed on the liquid-cooled channel resonance spectrum to identify the main frequency band corresponding to the energy density peak. The main frequency band offset is obtained by calibration and comparison with a preset baseline. For example, in a high temperature environment of 70°C, the viscosity of the coolant decreases, resulting in an increase in the propagation speed of the sound wave, and the main frequency band shifts from the baseline of 12.5kHz to 13.2kHz. At this time, the main frequency band offset Δf=+0.7kHz;

[0063] Within the selected main frequency band, the harmonic energy attenuation slope is calculated, and the spectrum envelope is fitted using linear regression. The absolute value of the spectrum envelope slope is used as the liquid cooling attenuation rate. For example, metal particle contamination causes steep attenuation in the 15-20 kHz high frequency band. At this time, the fitting slope of the spectrum envelope is α = −18.5, which is significantly higher than −6.2 of pure liquid.

[0064] The instantaneous phase of the acoustic signal is extracted through Hilbert transform, and its time-domain average offset relative to the preset value is calculated. The time-domain average offset is used as the phase offset. For example, when the flow rate increases from 1.2 m / s to 2.0 m / s, the fluid turbulence causes phase jitter, and the phase offset Δϕ increases from 0.1 rad to 0.35 rad.

[0065] In step S32 , the main frequency band offset, the liquid cooling attenuation rate, the phase offset, and the liquid cooling channel environment parameters are integrated to generate a liquid cooling ecological vector.

[0066] Specifically, the main frequency band offset, liquid cooling attenuation rate, phase offset and liquid cooling channel environment parameters are normalized and spliced ​​to generate a liquid cooling ecological vector, which can be expressed as:

[0067] ;

[0068] in, Indicates the main frequency band offset, It is the historical standard deviation of the main frequency band offset. Expressed as the liquid cooling attenuation rate, Expressed as the historical standard deviation of the liquid cooling decay rate, Expressed as a phase offset, Expressed as the historical standard deviation of the phase offset, Expressed as the liquid cooling channel temperature parameter, It is expressed as the temperature parameter benchmark of liquid cooling channel. Expressed as the liquid cooling channel flow rate parameter, Expressed as the liquid cooling channel flow rate parameter benchmark, Expressed as the liquid cooling channel pressure parameter, Expressed as the liquid cooling channel pressure parameter benchmark, Represented as a liquid cooling eco vector.

[0069] For example, a data center is under full load, and the corresponding parameters of the liquid cooling system of the data center are =65℃, =2.2m / s, =110kPa, =+0.4kHz, =-14.2, =0.28rad, and =0.3, =5.6, =0.15, =25℃, =1.5m / s, =101.3kPa, then the corresponding =[1.33,-2.54,1.87,1.0,1.4,1.09].

[0070] Step S4, obtain the historical liquid cooling ecological vector corresponding to the historical data of the liquid cooling channel, build a liquid cooling impurity monitoring model, input the existing liquid cooling impurity attribute set and the historical liquid cooling ecological vector into the liquid cooling impurity monitoring model for pre-training, and generate a liquid cooling impurity library.

[0071] In this embodiment, step S4 includes:

[0072] Step S41: construct a dynamic association network between the historical liquid cooling ecological vector and the existing liquid cooling impurity attribute set based on the bidirectional attention mechanism, and perform bidirectional interactive calculation on the historical liquid cooling ecological vector and the existing liquid cooling impurity attribute set to obtain the interactive calculation result, and align the liquid cooling channel resonance spectrum diagram with the existing liquid cooling impurity attribute set based on the interactive calculation result.

[0073] Specifically, the historical liquid cooling ecological vector and the existing liquid cooling impurity attribute set are mapped to a unified dimension through a fully connected layer to form an input vector, and the input vector is used as the input basis for attention calculation. The forward attention flow uses the historical liquid cooling ecological vector as the query vector and the impurity attributes contained in the existing liquid cooling impurity attribute set as the key value vector to calculate the association weight of the spectral features to the impurity attributes. This process generates attention distribution through the Softmax function, identifies key feature combinations, such as the strong correlation between high-frequency attenuation rate and metal particle density, and outputs the forward attention flow results.

[0074] The reverse attention flow reverses the query and key-value roles, takes the impurity attribute as the starting point, locates redundant features in the historical liquid cooling ecological vector that are interfered with by environmental noise, and dynamically shields abnormal data through binary masks, such as phase jitter caused by high flow rate, and outputs the reverse attention results.

[0075] Furthermore, the output results of the forward attention stream and the reverse attention stream are spliced ​​and residual-connected, and then fused into a joint feature vector using layer normalization. The historical liquid cooling ecological vector and the joint feature vector are mapped to the same dimension, and a cosine similarity comparison is performed. The spectral features and attribute features corresponding to the same type of impurities are closely clustered, while heterogeneous features are significantly separated. The spectral features are represented by the main frequency band offset, liquid cooling attenuation rate, and phase offset.

[0076] Step S42 , performing environmental interference correction on the main frequency band offset, the liquid cooling attenuation rate, and the phase offset according to a liquid cooling compensation mechanism.

[0077] It can be understood that a linear regression function is constructed based on the liquid cooling channel environmental parameters contained in the historical liquid cooling ecological vector, and the linear perturbation of the liquid cooling channel environmental parameters on the main frequency band offset is quantified according to the linear regression function, and the linear perturbation result is obtained. The main frequency band offset is corrected according to the linear perturbation result. The linear regression function between the main frequency band offset and the temperature is expressed as:

[0078] ;

[0079] in, Expressed as the uncorrected phase offset, is the actual offset caused by impurities, Expressed as the liquid cooling channel temperature parameter, for example, when =45℃, =+3.0kHz, where the ambient temperature rise contribution is 0.15×(45-25)=+3.0kHz, and the actual impurity offset 0kHz, then the corrected main frequency band offset is -3.0=0kHz.

[0080] It can be understood that by constructing a covariance matrix to analyze the coupling relationship between the liquid cooling attenuation rate and the flow rate parameter in the liquid cooling channel environmental parameter, a coupling result is obtained, and the liquid cooling attenuation rate is corrected by combining the coupling result with the liquid cooling attenuation rate correction formula. The liquid cooling attenuation rate correction formula can be expressed as:

[0081] ;

[0082] in, Expressed as the corrected liquid cooling attenuation rate, Expressed as the uncorrected liquid cooling decay rate, Expressed as the liquid cooling channel flow rate parameter.

[0083] For example, when the coolant flow rate in the liquid cooling channel increases from 1.5m / s to 3.0m / s, the liquid cooling attenuation rate From -12 to -18, =-18×(1+0.4×3)=-39.6.

[0084] A nonlinear drift function of the phase offset is constructed based on wavelet transform and Kalman filtering operations, and the phase offset is corrected by the nonlinear drift function. The wavelet transform is represented by decomposing the phase signal through the Daubechies wavelet basis to separate high-frequency noise greater than 100 Hz and low-frequency real drift less than 10 Hz.

[0085] Step S43 , obtaining a mapping relationship between the liquid cooling impurity attributes included in the existing liquid cooling impurity attribute set and the main frequency band offset, liquid cooling attenuation rate, and phase offset after liquid cooling compensation, and constructing a liquid cooling impurity library based on the mapping relationship.

[0086] For example, the mapping relationship can be expressed as follows: [+1.2kHz, −32dB / octave, 0.15rad] and the impurity attribute [metal, 8.9g / cm³, 20μm] correspond to the occurrence of copper particle oxide layer rupture, and [−0.8kHz, −9dB / octave, 0.8rad] and the impurity attribute [gas phase, 0.001g / cm³, 200μm] correspond to the occurrence of bubble clusters.

[0087] In step S5, the liquid cooling ecological vector is input into a pre-trained liquid cooling impurity monitoring model, and the liquid cooling ecological vector is matched with the liquid cooling impurity library based on the cuckoo algorithm to obtain a liquid cooling impurity combination vector, where the liquid cooling impurity combination vector includes the liquid cooling impurity type and the impurity monitoring factor.

[0088] In this embodiment, step S5 includes:

[0089] Step S51 : randomly generating a plurality of liquid-cooled cuckoos based on a liquid-cooled impurity library, each liquid-cooled cuckoo carrying a set of cuckoo flight parameters, wherein the cuckoo flight parameters include a step size parameter, a direction vector, and a search radius.

[0090] Specifically, the step size parameter is expressed as the ratio of the distance moved in the feature space and obeys the Levy flight distribution, which is expressed as ,in Expressed as the step size parameter, It is in the range of [1,10], the direction vector is expressed as the feature dimension adjustment direction, and the search radius is expressed as the local search range radius.

[0091] For example, the flight parameters of liquid-cooled cuckoo A are {0.28, [0.6, -0.3, 0.5], 0.35}, and the flight parameters of liquid-cooled cuckoo B are {0.41, [-0.8, 0.2, 0.7], 0.22}.

[0092] In step S52, the liquid-cooled cuckoo performs a random walk with alternating long and short steps based on the cuckoo flight parameters. During the walk, the liquid-cooled ecological vector and the liquid-cooled impurity similarity of each data point in the liquid-cooled impurity library are calculated in real time, an impurity matching score is generated based on the liquid-cooled impurity similarity, and the cuckoo flight parameters are dynamically adjusted according to the impurity matching score.

[0093] Specifically, the long-short step alternating random walk includes long-step exploration and short-step development. The long-step exploration is represented by a long-distance jump with a probability of 70%, and the end development is represented by a local search within a radius r with a probability of 30%. The liquid-cooled impurity similarity calculation is represented by calculation based on the liquid-cooled impurity similarity calculation formula, which can be expressed as:

[0094] ;

[0095] in It represents the weight of the main frequency band offset, liquid cooling attenuation rate and phase offset, and the weight is one of [0.5, 0.3, 0.2], that is, is the main frequency band offset weight, is the liquid cooling attenuation rate weight, is the phase offset weight, and , Indicates liquid-cooled cuckoo dimensional data, i.e. Indicates the current main frequency offset of the liquid-cooled cuckoo. Expressed as the current liquid cooling attenuation rate of the liquid cooling cuckoo, Expressed as the current phase offset of the liquid-cooled cuckoo, Indicated as the first Dimensional benchmark value.

[0096] For example, the initial position of the liquid-cooled cuckoo A is [1.2kHz, -28, 0.15rad]. The initial position performs a long-distance jump according to the flight parameters of the liquid-cooled cuckoo A. The position after the jump is [1.2+0.28×0.6kHz, -28+(-0.3×0.28), 0.15+0.28×0.5rad]=[-0.78kHZ, -27.916, 0.29rad], and the corresponding benchmark in the matched liquid-cooled impurity library is [-0.612kHZ, -25.816, 0.20rad]. The liquid-cooled impurity similarity between the liquid-cooled cuckoo A and the data point matched in this flight is 0.47.

[0097] The initial position of the liquid-cooled cuckoo B is [-0.9kHz, -10, 0.75rad], and the local perturbation is added A short-distance search is performed at the initial position, and the new position is [-0.78 kHz, -8.5, 0.75 rad], and the corresponding benchmark in the matched liquid-cooled impurity library is [-0.68 kHz, -8.0, 0.65 rad]. The similarity between the liquid-cooled cuckoo B and the liquid-cooled impurity of the data point matched in this flight is 0.78.

[0098] Furthermore, an impurity matching score is obtained according to the impurity matching scoring mechanism and combined with the liquid cooling impurity similarity. The impurity matching scoring mechanism is expressed as assigning a score of (90, 100] to data with a liquid cooling impurity similarity within (0.8, 1.0], a score of (60, 80] to data with a liquid cooling impurity similarity within (0.5, 0.8], and a score of [0, 60] to data with a liquid cooling impurity similarity within [0.0, 0.5].

[0099] For example, if the liquid-cooled impurity similarity between liquid-cooled cuckoo A and the data points matched to this flight is 0.47, then the impurity matching score assigned to this flight is 49; if the liquid-cooled impurity similarity between liquid-cooled cuckoo B and the data points matched to this flight is 0.78, then the impurity matching score assigned to this flight is 79.

[0100] It can be understood that the dynamic adjustment of the cuckoo flight parameters according to the impurity matching score is expressed as follows: if the impurity matching score is less than 50 for two consecutive times, the step size parameter Perform an amplification operation, which is expressed as The amplification coefficient negatively correlated with the score is multiplied as the basis. The amplified step size drives the cuckoo to perform a longer distance Levy flight, quickly escape from the local optimal area, and expand the global exploration range. After each round of iteration, the flight direction is adjusted based on the gradient information of the impurity matching score, that is, the partial derivative of the score of the current position with respect to the main frequency offset, liquid cooling attenuation rate, and phase offset is calculated to generate a gradient vector. The direction vector is adjusted based on the gradient vector. When the cuckoo's matching score is higher than 80 points, it is determined that it has entered a high potential area, and the search radius is contracted. The radius contraction forces the cuckoo to perform Gaussian random walks in a narrow space and refine the exploration of local optimal solutions. If the score decreases in subsequent iterations, the radius automatically returns to the radius baseline value, which is expressed as the inverse of the number of adjacent records, and the breadth search capability is reactivated.

[0101] It is understandable that there is priority coordination in the adjustment of cuckoo flight parameters. Step size enlargement is a global escape strategy, which takes precedence over local optimization of radius contraction. When the cuckoo meets both the step size enlargement conditions and the radius contraction conditions, the system prioritizes step size enlargement to ensure that it escapes from the invalid area before conducting a detailed search.

[0102] Step S53: constructing a liquid-cooling impurity combination vector.

[0103] For example, in the input liquid cooling eco vector =-0.7kHz, =-9, =0.82rad, where the impurity type matched by the data in the liquid cooling impurity library is bubble generation, and The corresponding reference is -0.8kHz, The corresponding benchmark is -8, The corresponding benchmark is 0.8rad, and the impurity probability is generated by the SoftMax model. The obtained impurity probability is expressed as [0.02, 0.02, 0.96] after data normalization. The first 0.02 is the metal impurity score, the second 0.02 is the organic impurity score, and the 0.96 is the bubble score. The concentration factor is 1.55, the particle size distribution factor is 0.98, the confidence weight is 0.92, and the abnormal flag is 0. The final constructed liquid cooling impurity combination vector is expressed as V[0.02, 0.02, 0.96, 1.55, 0.98, 0.92, 0].

[0104] Step S6: Obtain a pre-trained impurity purification decision model, input the liquid cooling impurity combination vector into the impurity purification decision model, and generate an optimal liquid cooling impurity purification strategy. The user can remove impurities from the liquid cooling system according to the optimal liquid cooling purification strategy.

[0105] In this embodiment, step S6 includes:

[0106] Step S61: obtain a liquid-cooled impurity purification parameter group, which includes the environmental constraint parameters of the liquid cooling system and the power consumption of the impurity purification tool. Based on the liquid-cooled impurity combination vector, a predefined purification rule library is matched to obtain the purification rule matching result, and the liquid-cooled impurity purification parameter group and the purification rule matching result are combined to generate an impurity purification strategy candidate set.

[0107] Step S62 , evaluating the purification efficiency, energy consumption, and stability risk of the impurity purification strategy candidate set under timing constraints through Monte Carlo simulation, and using a non-dominated sorting algorithm to screen the optimal solution of the impurity purification strategy, and generating an optimal liquid cooling impurity purification strategy based on the optimal solution of the impurity purification strategy.

[0108] Figure 2 A schematic diagram of a liquid-cooled impurity monitoring system provided by some embodiments of the present application and capable of realizing the concept of the present application is shown. Figure 3 A schematic diagram of an impurity detection module in a liquid-cooled impurity monitoring system provided by some embodiments of the present application and capable of realizing the concept of the present application is shown.

[0109] Specifically, a liquid cooling impurity monitoring system includes:

[0110] The impurity detection module includes an acoustic sensor group and an environmental sensor group, which are used to obtain the liquid cooling channel resonance spectrum diagram and liquid cooling channel environmental data.

[0111] The acoustic sensor group is used to obtain the liquid cooling channel resonance frequency of the liquid cooling system and dynamically superimpose the liquid cooling channel resonance frequency to generate a liquid cooling channel resonance spectrum diagram.

[0112] The environmental sensor group is used to obtain the temperature parameters, flow rate parameters and pressure parameters of the coolant in the liquid-cooled cooling channel in real time, and combine the temperature parameters, the flow rate parameters and the pressure parameters into the liquid-cooled cooling channel environmental parameters.

[0113] A data preset module is used to obtain an existing liquid-cooled impurity attribute set through spectral analysis and microscopic imaging.

[0114] A collaborative extraction module is used to perform collaborative feature extraction on the liquid cooling channel resonance spectrum diagram and the liquid cooling channel environmental data, generate a liquid cooling ecological vector, and obtain a historical liquid cooling ecological vector corresponding to the liquid cooling channel historical data.

[0115] The model construction module is used to construct a liquid cooling impurity monitoring model, pre-train the liquid cooling impurity monitoring model, and generate a liquid cooling impurity library.

[0116] The impurity matching module is used to perform liquid cooling impurity matching between the liquid cooling ecological vector and the liquid cooling impurity library based on the cuckoo algorithm to obtain a liquid cooling impurity combination vector.

[0117] A decision generation module is used for the impurity purification decision model to generate an optimal liquid cooling purification strategy based on a liquid cooling impurity combination vector.

[0118] The specific usage and function of this embodiment are described below:

[0119] First, the existing liquid cooling impurity attribute set is obtained through spectral analysis and microscopic imaging, and the impurity types and corresponding attributes that may be generated in the liquid cooling system are obtained and integrated to expand the scope of subsequent impurity matching monitoring. Then, based on the impurity detection module, the liquid cooling channel resonance spectrum and liquid cooling channel environmental parameters are obtained, and the liquid cooling channel resonance spectrum and liquid cooling channel environmental data are collaboratively extracted to generate a liquid cooling ecological vector, and the historical liquid cooling ecological vector corresponding to the liquid cooling channel historical data is obtained to construct a liquid cooling impurity monitoring model. The existing liquid cooling impurity attribute set and the historical liquid cooling ecological vector are input into the liquid cooling impurity monitoring model for pre-training to generate a liquid cooling impurity library. The various liquid cooling channel resonance frequencies in the liquid cooling channel resonance spectrum are compared. Align the impurity attributes with the existing liquid cooling impurity attribute set so that no complicated operations are required for subsequent impurity matching. Finally, input the liquid cooling ecological vector into the pre-trained liquid cooling impurity monitoring model, and match the liquid cooling ecological vector with the liquid cooling impurity library based on the cuckoo algorithm to obtain the liquid cooling impurity combination vector. The liquid cooling impurity combination vector includes the liquid cooling impurity type and the impurity monitoring factor. Obtain a pre-trained impurity purification decision model, input the liquid cooling impurity combination vector into the impurity purification decision model, and generate the optimal liquid cooling impurity purification strategy. Users can remove impurities from the liquid cooling system according to the optimal liquid cooling purification strategy. This method can improve the efficiency of detecting coolant impurities in the liquid cooling system.

[0120] In addition, an embodiment of the present invention further provides an electronic device, including:

[0121] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method proposed in the first embodiment of the present invention.

[0122] The following is a detailed introduction to the various components of electronic equipment:

[0123] The term "processor" is the control center of an electronic device and can be a single processor or a collective term for multiple processing elements. For example, the processor can be one or more central processing units (CPUs), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the first embodiment of the present invention, such as one or more digital signal processors (DSPs) or one or more field programmable gate arrays (FPGAs).

[0124] The processor can execute various functions of the electronic device by running or executing software programs stored in the memory and calling data stored in the memory.

[0125] The memory is used to store the software program for executing the solution of the present invention, and the execution is controlled by the processor. The specific implementation method can refer to the above method embodiment and will not be repeated here.

[0126] The memory may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only (CD-ROM), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto. The memory may be integrated with the processor or exist independently and be coupled to the processor via an interface circuit of the electronic device, and this is not specifically limited in the embodiments of the present invention.

[0127] The above embodiments can be implemented in whole or in part via software, hardware (e.g., circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in accordance with the embodiments of the present invention are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wireless communication (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer, or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0128] It should be understood that the term "and / or" as used herein simply describes an association between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent the existence of A alone, the existence of both A and B, or the existence of B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the related objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0129] It should be understood that in the embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0130] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A liquid cooling impurity monitoring method, characterized in that: The method comprises: S1: Obtain the existing liquid-cooled impurity property set through spectral analysis and microscopic imaging; S2: Deploy an impurity detection module in the liquid cooling channel and obtain the liquid cooling channel resonance spectrum and liquid cooling channel environmental parameters based on the impurity detection module; S3: Perform collaborative feature extraction on the liquid cooling channel resonance spectrum and the liquid cooling channel environment data to generate a liquid cooling ecological vector, wherein the liquid cooling ecological vector includes a main frequency band offset, a liquid cooling attenuation rate, a phase offset, and liquid cooling channel environment parameters; The collaborative feature extraction includes extracting features from the liquid cooling channel resonance spectrum to obtain a main frequency band offset, a liquid cooling attenuation rate, and a phase offset, and combining the main frequency band offset, the liquid cooling attenuation rate, and the phase offset with liquid cooling channel environmental data to generate a liquid cooling ecological vector; S4: Obtain the historical liquid cooling ecological vector corresponding to the historical data of the liquid cooling channel, build a liquid cooling impurity monitoring model, input the existing liquid cooling impurity attribute set and the historical liquid cooling ecological vector into the liquid cooling impurity monitoring model for pre-training, and generate a liquid cooling impurity library; S5: Input the liquid cooling ecological vector into a pre-trained liquid cooling impurity monitoring model, perform liquid cooling impurity matching between the liquid cooling ecological vector and the liquid cooling impurity library based on the cuckoo algorithm, and obtain a liquid cooling impurity combination vector, wherein the liquid cooling impurity combination vector includes a liquid cooling impurity type and an impurity monitoring factor; S6: Obtain a pre-trained impurity purification decision model, input the liquid cooling impurity combination vector into the impurity purification decision model, generate an optimal liquid cooling impurity purification strategy, and the user removes impurities from the liquid cooling system according to the optimal liquid cooling purification strategy.

2. A liquid cooling impurity monitoring method according to claim 1, characterized in that: The process of obtaining a historical liquid cooling ecological vector corresponding to the historical data of the liquid cooling channel, constructing a liquid cooling impurity monitoring model, inputting the existing liquid cooling impurity attribute set and the historical liquid cooling ecological vector into the liquid cooling impurity monitoring model for pre-training, and generating a liquid cooling impurity library includes: Based on the bidirectional attention mechanism, a dynamic association network of historical liquid cooling ecological vectors and existing liquid cooling impurity attribute sets is constructed, and bidirectional interactive calculations are performed on the historical liquid cooling ecological vectors and the existing liquid cooling impurity attribute sets to obtain interactive calculation results. Based on the interactive calculation results, the liquid cooling channel resonance spectrum diagram is dimensionally aligned with the existing liquid cooling impurity attribute set; Performing environmental interference correction on the main frequency band offset, the liquid cooling attenuation rate, and the phase offset according to a liquid cooling compensation mechanism; A mapping relationship between the liquid cooling impurity attributes included in the existing liquid cooling impurity attribute set and the main frequency band offset, liquid cooling attenuation rate and phase offset after liquid cooling compensation is obtained, and a liquid cooling impurity library is constructed based on the mapping relationship.

3. The liquid cooling impurity monitoring method according to claim 2, characterized in that: The performing environmental interference correction on the main frequency band offset, the liquid cooling attenuation rate, and the phase offset according to the liquid cooling compensation mechanism includes: Constructing a linear regression function based on the liquid cooling channel environmental parameters included in the historical liquid cooling ecological vector, quantifying the linear perturbation of the liquid cooling channel environmental parameters on the main frequency band offset according to the linear regression function, obtaining a linear perturbation result, and correcting the main frequency band offset according to the linear perturbation result; Analyzing the coupling relationship between the liquid cooling attenuation rate and the flow rate parameter in the liquid cooling channel environmental parameter by constructing a covariance matrix, obtaining a coupling result, and correcting the liquid cooling attenuation rate according to the coupling result; A nonlinear drift function of the phase offset is constructed based on wavelet transform and Kalman filter operations, and the phase offset is corrected by the nonlinear drift function.

4. The liquid cooling impurity monitoring method according to claim 1, characterized in that: The liquid cooling ecological vector is input into a pre-trained liquid cooling impurity monitoring model, and the liquid cooling ecological vector is matched with the liquid cooling impurity library based on the cuckoo algorithm to obtain the liquid cooling impurity combination vector, including: Randomly generate multiple liquid-cooled cuckoos based on the liquid-cooled impurity library, each liquid-cooled cuckoo carries a set of cuckoo flight parameters, and the cuckoo flight parameters include a step size parameter, a direction vector, and a search radius; The liquid-cooled cuckoo performs a random walk alternating between long and short steps based on the cuckoo flight parameters. During the walk, the liquid-cooled ecological vector and the liquid-cooled impurity similarity of each data point in the liquid-cooled impurity library are calculated in real time. An impurity matching score is generated based on the liquid-cooled impurity similarity, and the cuckoo flight parameters are dynamically adjusted according to the impurity matching score. When the variance of the motion trajectory of the liquid-cooled cuckoo is continuously iterated below a preset convergence threshold, it is determined to be a stable solution state, the data node with the highest impurity matching score in the liquid-cooled impurity library is extracted, and a liquid-cooled impurity combination vector is constructed.

5. The liquid cooling impurity monitoring method according to claim 1, characterized in that: The impurity monitoring factors include: The impurity detection factors include concentration factor, particle size distribution factor, confidence weight and abnormal flag. The concentration factor is used to indicate the concentration level of impurities in the coolant. The particle size distribution factor is used to indicate the average particle size of impurities in the coolant. The confidence weight is used to indicate the similarity of liquid cooling impurities calculated by the cuckoo algorithm. The abnormal flag is used to indicate the characteristic mark when the impurity is not matched in the liquid cooling impurity library.

6. The liquid cooling impurity monitoring method according to claim 1, characterized in that: The method of obtaining a pre-trained impurity purification decision model, inputting a liquid cooling impurity combination vector into the impurity purification decision model, generating an optimal liquid cooling impurity purification strategy, and performing impurity removal on the liquid cooling system according to the optimal liquid cooling impurity purification strategy includes: Obtaining a liquid-cooled impurity purification parameter group, the liquid-cooled impurity purification parameter group including environmental constraint parameters of the liquid cooling system and power consumption of an impurity purification tool; matching a predefined purification rule library based on the liquid-cooled impurity combination vector to obtain a purification rule matching result; and combining the liquid-cooled impurity purification parameter group with the purification rule matching result to generate an impurity purification strategy candidate set; Monte Carlo simulation is used to evaluate the purification efficiency, energy consumption and stability risk of the impurity purification strategy candidate set under timing constraints, and a non-dominated sorting algorithm is used to screen the optimal solution of the impurity purification strategy. Based on the optimal solution of the impurity purification strategy, the optimal liquid cooling impurity purification strategy is generated.

7. The liquid cooling impurity monitoring method according to claim 1, characterized in that: The impurity detection module is deployed in the liquid cooling channel, and the liquid cooling channel resonance spectrum and liquid cooling channel environmental data are obtained based on the impurity detection module, including: Acquiring the liquid cooling channel resonance frequency of the liquid cooling system in real time based on the acoustic sensor group included in the impurity detection module, dynamically superimposing the liquid cooling channel resonance frequency to generate a liquid cooling channel resonance spectrum diagram; The environmental parameters of the liquid cooling channel are acquired in real time based on the environmental sensor group included in the impurity detection module. The environmental parameters of the liquid cooling channel include the temperature, flow rate and pressure parameters of the coolant in the liquid cooling channel.

8. The liquid cooling impurity monitoring method according to claim 1, characterized in that: The method of obtaining the existing liquid cooling impurity attribute set through spectral analysis and microscopic imaging includes: Perform impurity analysis on the sample coolant based on spectral analysis to obtain the type and properties of organic impurities in the sample coolant; Perform impurity analysis on the sample coolant based on microscopic imaging to obtain the type and properties of inorganic impurities in the sample coolant; The sample coolants are represented by multiple groups of coolants of liquid cooling systems in different operating environments, and an existing liquid cooling impurity attribute set is constructed based on the organic impurity type and attribute and the inorganic impurity type and attribute.

9. A liquid cooling impurity monitoring system, characterized in that: include: An impurity detection module, comprising an acoustic sensor group and an environmental sensor group, for obtaining a liquid cooling channel resonance spectrum and liquid cooling channel environmental data; An acoustic sensor group, the acoustic sensor group is used to obtain the liquid cooling channel resonance frequency of the liquid cooling system, and dynamically superimpose the liquid cooling channel resonance frequency to generate a liquid cooling channel resonance spectrum diagram; An environmental sensor group, the environmental sensor group is used to obtain temperature parameters, flow rate parameters and pressure parameters of the coolant in the liquid-cooling channel in real time, and combine the temperature parameters, the flow rate parameters and the pressure parameters into liquid-cooling channel environmental parameters; A data presetting module, the data presetting module is used to obtain an existing liquid-cooled impurity attribute set through spectral analysis and microscopic imaging; A collaborative extraction module is used to perform collaborative feature extraction on the liquid cooling channel resonance spectrum and the liquid cooling channel environment data to generate a liquid cooling ecological vector and obtain a historical liquid cooling ecological vector corresponding to the liquid cooling channel historical data; A model building module, wherein the model building module is used to build a liquid cooling impurity monitoring model, pre-train the liquid cooling impurity monitoring model, and generate a liquid cooling impurity library; An impurity matching module, configured to perform liquid cooling impurity matching between the liquid cooling ecological vector and the liquid cooling impurity library based on a cuckoo algorithm to obtain a liquid cooling impurity combination vector; A decision generation module is used for the impurity purification decision model to generate an optimal liquid cooling purification strategy based on a liquid cooling impurity combination vector.

Images