Two-phase cold plate liquid cooling method and system

Abstract

The invention provides a two-phase cold plate liquid cooling method and system, and the method comprises the steps: obtaining a cold plate design scheme and a historical cold plate design scheme, obtaining a cold plate information set and a historical cold plate information set based on the cold plate design scheme and the historical cold plate design scheme, carrying out the preprocessing, and carrying out the clustering analysis of the preprocessed cold plate information set and historical cold plate information set, the method comprises the steps of establishing a cold plate portrait and a historical cold plate portrait, performing portrait numbering on the cold plate portrait and the historical cold plate portrait, obtaining a cold plate improvement direction, performing direction numbering on the cold plate improvement direction, establishing a scheme database based on the cold plate portrait, the historical cold plate portrait and the cold plate improvement direction, and establishing a cold plate table framework in the scheme database. The scheme improvement model is constructed, the scheme improvement model and the scheme database are linked, the recommendation improvement direction is output based on the scheme improvement model, designers can be assisted in improving the two-phase cold plate design scheme through the recommendation improvement direction, and the research and development efficiency of the designers is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of liquid cooling and heat dissipation, and more specifically, particularly relates to a two-phase cold plate liquid cooling method and system. Background Art

[0002] With the rapid development of information technology, electronic equipment is developing rapidly towards high performance, miniaturization and integration. From server clusters in data centers to personal smartphones and laptops, to precision electronic instruments in the aerospace field, the computing speed and integration of chips inside the equipment are constantly improving, and the heat generated per unit area is also growing exponentially. Taking data centers as an example, with the widespread application of technologies such as artificial intelligence and big data, the power density of a single cabinet has soared from several kilowatts in the past to tens of kilowatts. The traditional air cooling method has limited heat dissipation efficiency and is difficult to meet such high heat dissipation requirements. Liquid cooling technology has gradually become the core solution to solve the problem of high heat density heat dissipation. Two-phase cold plate liquid cooling technology stands out among many liquid cooling solutions due to its efficient heat dissipation capabilities. This technology utilizes the characteristics of coolant absorbing and releasing a large amount of latent heat when phase change occurs in the cold plate. It can take away more heat with a smaller flow rate and has higher heat dissipation efficiency than single-phase liquid cooling.

[0003] However, the existing two-phase cold plate liquid cooling design scheme still has shortcomings. The optimization and improvement of the cold plate design scheme often lacks systematic evaluation reference, and most design schemes rely on empirical formulas and repeated experiments. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a two-phase cold plate liquid cooling method and system to solve the technical problems in the prior art that the design optimization and improvement of traditional two-phase cold plates often lack systematic evaluation references, and most design schemes rely on empirical formulas and repeated experiments.

[0005] The purpose and effect of a two-phase cold plate liquid cooling method and system of the present invention are achieved by the following specific technical means: A two-phase cold plate liquid cooling method comprises the following steps: Acquire a cold plate design scheme and a historical cold plate design scheme, and acquire a cold plate information set and a historical cold plate information set based on the cold plate design scheme and the historical cold plate design scheme; Preprocessing the cold plate information set and the historical cold plate information set, performing cluster analysis on the preprocessed cold plate information set and the historical cold plate information set, establishing cold plate portraits and historical cold plate portraits, and numbering the cold plate portraits and the historical cold plate portraits; Obtain the improvement directions of the cold plate, number the improvement directions of the cold plate, establish a solution database based on the cold plate image, the historical cold plate image and the improvement directions of the cold plate, and establish a cold plate table structure in the solution database; Construct a solution improvement model, establish a link between the solution improvement model and the solution database, and output recommended improvement directions based on the solution improvement model.

[0006] As a further solution of the present invention, the constructing a solution improvement model, establishing a link between the solution improvement model and the solution database, and outputting recommended improvement directions based on the solution improvement model includes: Establish an SVM model, establish a link between the SVM model and the solution database, and the SVM model calls the data of the historical cold plate information table and the improvement information table in the cold plate table structure; Take the data in the historical cold plate information table as SVM features, and take the data in the improvement information table as SVM labels; The SVM model generates a corresponding matching degree function for each direction number based on the SVM features and the SVM labels, and generates a matching degree score based on the matching degree function; Fuse the matching degree score with the SVM features to form a feature fusion matrix; Import the feature fusion matrix into the established random forest model, analyze the fusion matrix based on the decision tree voting strategy of the random forest model, and output a solution improvement model, which is used to obtain multiple direction matching probabilities.

[0007] As a further solution of the present invention, the method further includes: After the solution improvement model is linked to the solution database, the solution improvement model calls the data of the cold plate information table in the cold plate table structure. The solution improvement model generates multiple direction matching probabilities based on the data of the cold plate information table in the cold plate table structure. The multiple direction matching probabilities correspond to the direction numbers respectively. Based on the direction matching probabilities, the first X direction numbers are used as recommended improvement directions, and the recommended improvement directions are imported into the solution database. The recommended improvement directions correspond to the image numbers of the cold plate image.

[0008] As a further solution of the present invention, the preprocessing of the cold plate information set and the historical cold plate information set, and the clustering analysis of the preprocessed cold plate information set and the historical cold plate information set to establish a cold plate image and a historical cold plate image includes: Obtain a cold plate evaluation vector based on the cold plate information set, obtain a historical cold plate evaluation vector based on the historical cold plate information set, obtain the scheme test standard time, perform time window partitioning on the cold plate evaluation vector and the historical cold plate evaluation vector based on the scheme test standard time, and establish a cold plate evaluation sequence and a historical cold plate evaluation sequence; Query for data missing in the cold plate evaluation sequence and the historical cold plate evaluation sequence, perform linear interpolation filling on the cold plate evaluation sequence and the historical cold plate evaluation sequence with data missing, and at the same time standardize the cold plate evaluation sequence and the historical cold plate evaluation sequence; Obtain the cold plate evaluation DTW distance between the cold plate design scheme and the historical cold plate design scheme, generate a cold plate evaluation DTW distance matrix based on the DTW distance, perform hierarchical clustering on the cold plate evaluation DTW distance matrix to generate an evaluation dendrogram, obtain the initial cluster center based on the evaluation dendrogram, and update the initial cluster center based on the cold plate evaluation DTW distance; Define an evaluation label, extract cold plate evaluation sequence features based on the updated initial cluster center, match the cold plate evaluation sequence features with the evaluation label, and establish a cold plate portrait and a historical cold plate portrait based on the evaluation label.

[0009] As a further solution of the present invention, the solution further includes: The cold plate information set includes a scheme name, heat dissipation test data, uniformity test data, and response test data. Generate a cold plate evaluation vector based on the heat dissipation test data, the uniformity test data, and the response test data. Each cold plate portrait corresponds to the scheme name; The historical cold plate information set includes a historical scheme name, historical heat dissipation test data, historical uniformity test data, and historical response test data. Generate a historical cold plate evaluation vector based on the historical heat dissipation test data, the historical uniformity test data, and the historical response test data. Each historical cold plate portrait corresponds to the historical scheme name.

[0010] As a further solution of the present invention, the numbering of the cold plate portrait and the historical cold plate portrait includes: The portrait numbers of the cold plate portrait and the historical cold plate portrait both include a number segment, a time period segment, and a name segment; The portrait number of the cold plate portrait is represented as NUM-DATE-NAME, and the portrait number of the historical cold plate portrait is represented as NUM-DATE-HNAME; Among them, NUM represents the number segment, DATE represents the time period segment, NAME represents the name segment of the portrait number of the cold plate portrait, and HNAME represents the name segment of the portrait number of the historical cold plate portrait.

[0011] As a further solution of the present invention, a solution database is established based on the cold plate image, the historical cold plate image and the cold plate improvement direction, and a cold plate table structure is established in the solution database, including: The solution database selects a relational database, and the cold plate table structure includes a cold plate information table, a historical cold plate information table, an improvement information table and a recommended improvement information table; Links are established for the cold plate information table, the historical cold plate information table, the improvement information table and the recommended improvement information table, and a database index is established at the same time. The primary keys of the cold plate information table and the historical cold plate information table are the image numbers of the cold plate image and the historical cold plate image, the primary key of the improvement information table is the direction number, and the primary key of the recommended improvement information table is the image number of the cold plate image.

[0012] As a further solution of the present invention, the method further includes: When establishing the solution database, the storage area in the solution database is divided into Y intervals based on a fixed time span. The Y intervals include a cold interval, a hot interval and a temporary storage interval. The cold interval is represented as a closed interval, the hot interval is represented as an unclosed interval, and the temporary storage interval is represented as an interval for data storage buffering; When the solution database stores data, the data is stored in the temporary storage interval, and an open or closed interval mark is made for the data based on the data storage time. The data marked with an open interval is stored in the hot interval, and the data marked with a closed mark is stored in the cold interval.

[0013] As a further solution of the present invention, the method further includes: A backup database is established, and the backup database is used to synchronously back up and store the data in the solution database; An update countdown is obtained. The update countdown represents the time when the backup database is updated. The update countdown is Z hours. When the update countdown ends, the backup database performs a storage update.

[0014] A two-phase cold plate liquid cooling system includes: An acquisition module, which is used to acquire the cold plate design solution and the historical cold plate design solution, and at the same time acquire the cold plate information set and the historical cold plate information set based on the cold plate design solution and the historical cold plate design solution, and acquire the cold plate improvement direction; A data processing module, which is used to preprocess the cold plate information set and the historical cold plate information set; A building module, which is used to perform clustering analysis on the cold plate information set and the historical cold plate information set, and establish a cold plate image and a historical cold plate image; A database module, the database module is used to establish a solution database and establish a cold plate table architecture in the solution database; A numbering module, which is used to number the cold plate portraits and historical cold plate portraits, and to number the improvement directions of the cold plate; The improvement module is used to construct a solution improvement model and output a recommended improvement direction based on the solution improvement model.

[0015] Compared with the prior art, the present invention has the following beneficial effects: Firstly, the cold plate design scheme and the historical cold plate design scheme are obtained, and the cold plate information set and the historical cold plate information set are obtained based on the above two design schemes, and the data in the cold plate information set and the historical cold plate information set are preprocessed, and then the preprocessed cold plate information set and the historical cold plate information set are clustered and analyzed to establish the cold plate portrait and the historical cold plate portrait, and the cold plate portrait and the historical cold plate portrait are numbered, and then the cold plate improvement direction is obtained, the cold plate improvement direction is numbered, and a scheme database is established, and a cold plate table structure is established in the scheme database. The established cold plate table structure can make it easier for designers to query historical related data, and finally a scheme improvement model is constructed, and the scheme improvement model is linked to the scheme database, and the recommended improvement direction is output based on the scheme improvement model. A method for assisting designers in improving the two-phase cold plate design scheme is provided, and designers can use the recommended improvement direction output by the scheme improvement model as a reference to assist in improving the liquid cooling design scheme of the two-phase cold plate, thereby improving the efficiency of designers in the research and development of the two-phase cold plate. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of the steps of a two-phase cold plate liquid cooling method of the present invention.

[0017] Figure 2 It is a flow chart of the steps of constructing a scheme improvement model in a two-phase cold plate liquid cooling method of the present invention.

[0018] Figure 3 It is a schematic diagram of a two-phase cold plate liquid cooling system of the present invention. DETAILED DESCRIPTION

[0019] The following is a further detailed description of the embodiments of the present invention in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the technical solutions of the present invention, but cannot be used to limit the scope of protection of the present invention. Embodiment 1

[0020] As attached Figure 1 and Figure 2 As shown: The present invention provides a two-phase cold plate liquid cooling method, which is applicable to the liquid cooling scheme design of a two-phase cold plate, and includes the following steps: Step S1, obtain the cold plate design scheme and the historical cold plate design scheme, and obtain the cold plate information set and the historical cold plate information set based on the cold plate design scheme and the historical cold plate design scheme.

[0021] Specifically, the cold plate information set includes the scheme name, heat dissipation test data, uniformity test data, and response test data. The historical cold plate information set includes the historical scheme name, historical heat dissipation test data, historical uniformity test data, and historical response test data. Designers can obtain the cold plate information set and the historical cold plate information set of the cold plate design scheme and the historical cold plate design scheme in simulation software or actual tests. The cold plate information set and the historical cold plate information set can reflect the liquid cooling performance of the two-phase cold plate design scheme, and provide a data basis for the establishment of the cold plate portrait and the historical cold plate portrait and the establishment of the subsequent scheme improvement model.

[0022] It can be understood that through the heat dissipation test data, the steady-state peak heat dissipation requirements when the chip is at full load power can be ensured. Through the uniformity test data, the temperature difference between different regions of the chip can be prevented from being too large, so as to ensure the reliability of the steady-state equilibrium temperature. Through the response test data, the speed of transient heat dissipation response can be ensured. The three groups of data, namely the heat dissipation test data, the uniformity test data, and the response test data, can be directly associated with devices such as IT devices or servers, so as to evaluate whether the core temperature of IT devices or servers and other devices will exceed the standard, whether the performance will be limited, and whether they can work stably in a dynamically changing environment when the two-phase cold plate design scheme is adopted. The heat dissipation test data, the uniformity test data, and the response test data can effectively and comprehensively evaluate the design scheme of the two-phase cold plate.

[0023] Step S2, preprocess the cold plate information set and the historical cold plate information set, perform clustering analysis on the preprocessed cold plate information set and the historical cold plate information set, establish the cold plate portrait and the historical cold plate portrait, and number the cold plate portrait and the historical cold plate portrait.

[0024] Specifically, the portrait numbers of the cold plate portrait and the historical cold plate portrait both include a number segment, a time segment, and a name segment. The portrait number of the cold plate portrait is expressed as NUM-DATE-NAME, and the portrait number of the historical cold plate portrait is expressed as NUM-DATE-HNAME. Among them, NUM represents the number segment, DATE represents the time segment, NAME represents the name segment of the portrait number of the cold plate portrait, and HNAME represents the name segment of the portrait number of the historical cold plate portrait.

[0025] For example, the portrait number of the historical cold plate design solution constructed on February 10, 2024 can be expressed as 14-20240210-Hza123, where the number segment is 14, the time period is 20240210, and the name segment is Hza123. Or the portrait number of the newly constructed cold plate design solution on March 10, 2025 can be expressed as 4-20250310-nw444, where the number segment is 4, the time period is 20250310, and the name segment is nw444.

[0026] In this embodiment, step S2 includes: Step S21: Obtain a cold plate evaluation vector based on the cold plate information set, obtain a historical cold plate evaluation vector based on the historical cold plate information set, obtain the scheme test standard time, divide the cold plate evaluation vector and the historical cold plate evaluation vector by time window based on the scheme test standard time, and establish a cold plate evaluation sequence and a historical cold plate evaluation sequence.

[0027] Specifically, the cold plate information set includes a scheme name, heat dissipation test data, uniformity test data, and response test data. A cold plate evaluation vector is generated based on the heat dissipation test data, the uniformity test data, and the response test data. Each cold plate portrait corresponds to the scheme name. The historical cold plate information set includes a historical scheme name, historical heat dissipation test data, historical uniformity test data, and historical response test data. A historical cold plate evaluation vector is generated based on the historical heat dissipation test data, the historical uniformity test data, and the historical response test data. Each historical cold plate portrait corresponds to the historical scheme name.

[0028] It can be understood that through the scheme test standard time, such as 15 minutes, 20 minutes, or 25 minutes, the heat dissipation test data, uniformity test data, and response test data in the cold plate evaluation vector, and the historical heat dissipation test data, historical uniformity test data, and historical response test data in the historical cold plate evaluation vector are converted into time series. For example, in a 45-minute test time period, if the scheme test standard time is set to 15 minutes, then the cold plate evaluation vector and the historical cold plate evaluation vector are converted into time series based on 15 minutes as the time window.

[0029] Step S22: Query for data missing in the cold plate evaluation sequence and the historical cold plate evaluation sequence, perform linear interpolation filling on the cold plate evaluation sequence and the historical cold plate evaluation sequence with data missing, and simultaneously perform standardization on the cold plate evaluation sequence and the historical cold plate evaluation sequence.

[0030] Specifically, linear interpolation filling can be used to fill in the missing data in the cold plate evaluation sequence and the historical cold plate evaluation sequence, and at the same time, standardize the data in the cold plate evaluation sequence and the historical cold plate evaluation sequence to unify the dimensions of the three groups of different data in the cold plate evaluation sequence and the historical cold plate evaluation sequence.

[0031] It can be understood that filling in the missing data in the cold plate evaluation sequence and the historical cold plate evaluation sequence can prevent data loss when analyzing the data in the cold plate evaluation sequence and the historical cold plate evaluation sequence using the K-MEANS clustering algorithm later. By standardizing the cold plate evaluation sequence and the historical cold plate evaluation sequence, the influence of dimensions can be eliminated, so that the units of each data in the cold plate evaluation sequence and the historical cold plate evaluation sequence are unified units, which is convenient for the subsequent analysis and processing of the cold plate evaluation sequence and the historical cold plate evaluation sequence by the K-MEANS clustering algorithm.

[0032] Step S23: Obtain the cold plate evaluation DTW distance between the cold plate design scheme and the historical cold plate design scheme, generate a cold plate evaluation DTW distance matrix based on the DTW distance, perform hierarchical clustering on the cold plate evaluation DTW distance matrix to generate an evaluation dendrogram, obtain the initial cluster centers based on the evaluation dendrogram, and update the initial cluster centers based on the cold plate evaluation DTW distance.

[0033] It can be understood that DTW (Dynamic Time Warping) is used to capture the similarity patterns between the cold plate evaluation sequence and the historical cold plate evaluation sequence, replacing the Euclidean distance of traditional clustering algorithms. When performing weighted calculations, a weighted calculation method with a learning weight of 60% and an internship weight of 40% is adopted, and at the same time, the cold plate evaluation DTW distance matrix is mapped to a low-dimensional space using multidimensional scaling.

[0034] Furthermore, after generating the evaluation dendrogram, cut the evaluation dendrogram into N parts to obtain N clusters, and select the sample with the smallest average cold plate evaluation DTW distance from other samples in other clusters in each cluster as the initial cluster center.

[0035] It can be understood that after converting the heat dissipation test data, uniformity test data, response test data, historical heat dissipation test data, historical uniformity test data, and historical response test data into time series, using DWT (Discrete Wavelet Transform) to calculate the similarity between the cold plate evaluation sequence of different cold plate design schemes and the historical cold plate design scheme and the historical cold plate evaluation sequence can eliminate the interference caused by time axis stretching. Based on the cold plate evaluation DTW distance matrix, hierarchical clustering is used to generate the initial cluster centers, replacing the random initiality of traditional clustering algorithms, which can improve the convergence speed and stability. When the cold plate design scheme becomes the historical cold plate design scheme, its historical cold plate evaluation sequence can be added and incremental clustering can be triggered to avoid recalculating all the data from scratch.

[0036] Step S24, define an evaluation label, extract the cold plate evaluation sequence features based on the updated initial cluster center, match the cold plate evaluation sequence features with the evaluation label, and establish a cold plate portrait and a historical cold plate portrait based on the evaluation label.

[0037] It can be understood that defining an evaluation label means assigning an understandable label to each cluster to describe the core features of the group. This evaluation label can be directly defined by designers. For example, the evaluation label can be defined as "good heat dissipation", "poor heat dissipation", etc. Extract the cold plate evaluation sequence based on the updated cluster center, match the cold plate evaluation sequence and the historical cold plate evaluation sequence with the evaluation label, and establish a cold plate portrait and a historical cold plate portrait that match the scheme name and the historical scheme name.

[0038] Step S3, obtain the cold plate improvement direction, number the cold plate improvement direction, establish a scheme database based on the cold plate portrait, the historical cold plate portrait and the cold plate improvement direction, and establish a cold plate table structure in the scheme database.

[0039] Specifically, the scheme database selects a relational database. The cold plate table structure includes a cold plate information table, a historical cold plate information table, an improvement information table and a recommended improvement information table. The cold plate information table is used to store data related to the cold plate design scheme and the cold plate portrait. The historical cold plate information table is used to store data related to the historical cold plate design scheme and the historical cold plate portrait. The improvement information table is used to store the cold plate improvement direction and the direction number. The recommended improvement information table stores the recommended improvement direction and the cold plate design scheme corresponding to the recommended improvement direction.

[0040] It can be understood that a relational database is used to store structured data and stores data in a table form. Each row represents a record, and each column represents an attribute. This two-dimensional table structure is intuitive and easy to understand, facilitating designers to understand and operate. Whether it is simple data storage or complex business logic processing, it can clearly display the relationship between data, making the organization and management of data more orderly, and designers can intuitively view relevant data.

[0041] Furthermore, establish links between the cold plate information table, the historical cold plate information table, the improvement information table and the recommended improvement information table, and at the same time establish a database index. The primary keys of the cold plate information table and the historical cold plate information table are the portrait numbers of the cold plate portrait and the historical cold plate portrait. The primary key of the improvement information table is the direction number. The primary key of the recommended improvement information table is the portrait number of the cold plate portrait.

[0042] It should be noted that when establishing the solution database, the storage area in the solution database is divided into Y intervals based on a fixed time span. The Y intervals include a cold interval, a hot interval, and a temporary storage interval. The cold interval is represented as a closed interval, the hot interval is represented as an unclosed interval, and the temporary storage interval is represented as an interval for data storage buffering.

[0043] Furthermore, when the solution database stores data, the data is stored in the temporary storage interval. Based on the data storage time, an open or closed interval mark is made for the data. The data marked with an open interval is stored in the hot interval, and the data marked with a closed mark is stored in the cold interval.

[0044] It can be understood that by adopting a method similar to curve integration to establish a relational database storage system, data can be temporarily stored in the temporary storage interval. During the period when the data is temporarily stored, the time span of the hot interval and the cold interval can be directly queried in advance, and at the same time, the time of the data to be stored can be queried. Through the query of the time span of the hot interval and the cold interval, an open or closed interval mark is made for the data, and the data is stored in the corresponding storage interval through the interval mark and time, which speeds up the storage speed and avoids the chaotic storage of data each time when storing, thereby improving the speed of the solution database when reading or storing data.

[0045] In addition, when establishing the solution database, a backup database is established. The backup database is used to synchronously back up and store the data in the solution database, and an update countdown is obtained. The update countdown represents that when the backup database is updated, the update countdown is Z hours. When the update countdown ends, the backup database performs a storage update. By establishing the backup database, data loss caused by the collapse of the solution database can be prevented.

[0046] Step S4, construct a solution improvement model, establish a link between the solution improvement model and the solution database, and output a recommended improvement direction based on the solution improvement model.

[0047] In this embodiment, step S4 includes: Step S41, establish an SVM model, establish a link between the SVM model and the solution database, and the SVM model calls the data in the historical cold plate information table and the improvement information table in the cold plate table structure.

[0048] Specifically, the SVM model adopts a one-versus-all strategy, and a binary classification SVM model is independently trained for each direction number. The historical cold plate evaluation sequence corresponding to the direction number of the binary classification SVM model is used as a positive sample, and the remaining historical cold plate evaluation sequences are used as negative samples.

[0049] Step S42: Use the data in the historical cold plate information table as SVM features and the data in the improvement information table as SVM labels.

[0050] Step S43: The SVM model generates a corresponding matching degree function for each direction number based on the SVM features and the SVM labels, and generates a matching degree score based on the matching degree function.

[0051] Specifically, use the historical cold plate evaluation sequence of the historical cold plate design scheme as SVM features, use the improvement number corresponding to the historical cold plate design scheme as SVM labels, import the SVM features and SVM labels into the SVM model, generate a corresponding matching degree function for the SVM labels and SVM features imported once based on the SVM model, and loop in this way until the SVM model generates a corresponding matching degree function for each improvement number and generates the corresponding matching degree score. Use the matching degree score as a new feature to fuse with the SVM features to form a feature fusion matrix.

[0052] Step S44: Fuse the matching degree score with the SVM features to form a feature fusion matrix.

[0053] Step S45: Import the feature fusion matrix into the established random forest model, analyze the fusion matrix based on the decision tree voting strategy of the random forest model, and output a scheme improvement model, which is used to obtain the matching probabilities in multiple directions.

[0054] Furthermore, import the feature fusion matrix into the random forest model, analyze the complex interaction relationship between the new features and SVM features in the feature fusion feature matrix based on the decision tree voting strategy and iterative training of the random forest model. When making the final prediction output, it can rely on the kernel function mapping of the SVM model, and can capture more implicit matching patterns through the random forest model, so as to output a scheme improvement model.

[0055] It can be understood that the scheme improvement model can quickly screen out the cold plate improvement directions with high matching degrees through the SVM model, while the random forest model can comprehensively mine more complex rules based on the original capabilities and the results of the SVM model, expand the recommendation scope, retain the accuracy of the SVM model, and also use the flexibility and comprehensiveness of the random forest model.

[0056] Specifically, after the solution improvement model establishes a link with the solution database, the solution improvement model calls the data in the cold plate information table in the cold plate table structure. The solution improvement model generates multiple direction matching probabilities based on the data in the cold plate information table in the cold plate table structure. The multiple direction matching probabilities correspond to the direction numbers respectively. Based on the direction matching probabilities, the first X direction numbers are used as the recommended improvement directions, and the recommended improvement directions are imported into the solution database. The recommended improvement directions correspond to the portrait numbers of the cold plate portraits. Designers can use the recommended improvement directions as a reference to improve the design solution of the two-phase cold plate and improve the R & D speed.

[0057] As shown in the Figure 3 appendix: A two-phase cold plate liquid cooling system includes: An acquisition module 101, which is used to acquire the cold plate design solution and the historical cold plate design solution. At the same time, based on the cold plate design solution and the historical cold plate design solution, it acquires the cold plate information set and the historical cold plate information set, and acquires the cold plate improvement direction.

[0058] A data processing module 102, which is used to preprocess the cold plate information set and the historical cold plate information set.

[0059] A building module 103, which is used to perform clustering analysis on the cold plate information set and the historical cold plate information set, and build a cold plate portrait and a historical cold plate portrait.

[0060] A database module 104, which is used to establish a solution database and establish a cold plate table structure in the solution database.

[0061] A numbering module 105, which is used to number the cold plate portrait and the historical cold plate portrait, and number the cold plate improvement direction.

[0062] An improvement module 106, which is used to build a solution improvement model and output recommended improvement directions based on the solution improvement model.

[0063] The specific usage and function of the first embodiment: First, obtain the cold plate design scheme and the historical cold plate design scheme. Based on the above two design schemes, obtain the cold plate information set and the historical cold plate information set. Preprocess the data in the cold plate information set and the historical cold plate information set. Subsequently, perform clustering analysis on the preprocessed cold plate information set and the historical cold plate information set, thereby establishing a cold plate portrait and a historical cold plate portrait. Number the cold plate portrait and the historical cold plate portrait. Then, obtain the cold plate improvement direction, number the direction of the cold plate improvement direction, and establish a solution database. Establish a cold plate table structure in the solution database. Through the established cold plate table structure, it is more convenient for designers to query relevant historical data. Finally, construct a solution improvement model, establish a link between the solution improvement model and the solution database, and output recommended improvement directions based on the solution improvement model, providing a method that can assist designers in improving the two-phase cold plate design scheme. Designers can use the recommended improvement directions output by the solution improvement model as a reference to assist in improving the liquid cooling design scheme of the two-phase cold plate, thereby improving the efficiency of designers in the research and development of the two-phase cold plate.

[0064] An electronic device, comprising: At least one processor; and a memory communicatively connected to at least one processor; wherein the memory stores instructions executable by at least one processor, and the instructions are executed by at least one processor to enable at least one processor to execute the method proposed in Embodiment 1 of the present invention.

[0065] The following specifically introduces each component of the electronic device: Among them, the processor is the control center of the electronic device, which can be a single processor or a collective term for multiple processing elements. For example, the processor is one or more central processing units (CPUs), or can be an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement Embodiment 1 of the present invention, such as: one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs).

[0066] Among them, 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.

[0067] The memory is used to store the software program for implementing the solution of the present invention and is controlled by the processor for execution. The specific implementation manner can refer to the above method embodiments and will not be elaborated here.

[0068] The memory may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store 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.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed 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 through the interface circuit of the electronic device. The embodiments of the present invention do not make specific limitations on this.

[0069] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware, or any other combination. 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 includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more collections of available media. The available medium may be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium may be a solid-state drive.

[0070] It should be understood that the term "and / or" in this text is merely a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. Additionally, the character " / " in this text generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, and the specific meaning can be understood by referring to the context before and after.

[0071] It should be understood that in the embodiments of the present invention, the magnitude of the sequence numbers of the above processes does not imply the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0072] The above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A two-phase cold plate liquid cooling method, characterized in that, It includes the following steps: Obtain the cold plate design scheme and the historical cold plate design scheme, and obtain the cold plate information set and the historical cold plate information set based on the cold plate design scheme and the historical cold plate design scheme; Preprocess the cold plate information set and the historical cold plate information set, perform clustering analysis on the preprocessed cold plate information set and the historical cold plate information set, establish a cold plate portrait and a historical cold plate portrait, and number the cold plate portrait and the historical cold plate portrait; Obtain the cold plate improvement direction, number the cold plate improvement direction, establish a scheme database based on the cold plate portrait, the historical cold plate portrait and the cold plate improvement direction, and establish a cold plate table structure in the scheme database; Construct a scheme improvement model, establish a link between the scheme improvement model and the scheme database, and output a recommended improvement direction based on the scheme improvement model.

2. The two-phase cold plate liquid cooling method according to claim 1, wherein The constructing the scheme improvement model, establishing a link between the scheme improvement model and the scheme database, and outputting a recommended improvement direction based on the scheme improvement model includes: Establish an SVM model, establish a link between the SVM model and the scheme database, and the SVM model calls the data in the historical cold plate information table and the improvement information table in the cold plate table structure; Use the data in the historical cold plate information table as SVM features and the data in the improvement information table as SVM labels; The SVM model generates a corresponding matching degree function for each direction number based on the SVM features and the SVM labels, and generates a matching degree score based on the matching degree function; Fuse the matching degree score with the SVM features into a feature fusion matrix; Import the feature fusion matrix into the established random forest model, analyze the fusion matrix based on the decision tree voting strategy of the random forest model, and output a scheme improvement model, which is used to obtain multiple direction matching probabilities.

3. A two-phase cold plate liquid cooling method according to claim 2, characterized in that, The method further includes: After the scheme improvement model is linked to the scheme database, the scheme improvement model calls the data in the cold plate information table in the cold plate table structure. The scheme improvement model generates multiple direction matching probabilities based on the data in the cold plate information table in the cold plate table structure. The multiple direction matching probabilities correspond to the direction numbers respectively. Based on the direction matching probabilities, the first X direction numbers are used as the recommended improvement directions, and the recommended improvement directions are imported into the scheme database. The recommended improvement directions correspond to the portrait numbers of the cold plate portraits.

4. A two-phase cold plate liquid cooling method according to claim 1, characterized in that, The preprocessing the cold plate information set and the historical cold plate information set, performing clustering analysis on the preprocessed cold plate information set and the historical cold plate information set, and establishing a cold plate portrait and a historical cold plate portrait includes: Obtain a cold plate evaluation vector based on the cold plate information set, obtain a historical cold plate evaluation vector based on the historical cold plate information set, obtain a scheme test standard time, divide the cold plate evaluation vector and the historical cold plate evaluation vector into time windows based on the scheme test standard time, and establish a cold plate evaluation sequence and a historical cold plate evaluation sequence; Query for data missing in the cold plate evaluation sequence and the historical cold plate evaluation sequence, perform linear interpolation filling on the cold plate evaluation sequence and the historical cold plate evaluation sequence with data missing, and at the same time perform standardization on the cold plate evaluation sequence and the historical cold plate evaluation sequence; Obtain the cold plate evaluation DTW distance between the cold plate design scheme and the historical cold plate design scheme, generate a cold plate evaluation DTW distance matrix based on the DTW distance, perform hierarchical clustering on the cold plate evaluation DTW distance matrix to generate an evaluation dendrogram, obtain the initial cluster centers based on the evaluation dendrogram, and update the initial cluster centers based on the cold plate evaluation DTW distance; Define evaluation labels, extract cold plate evaluation sequence features based on the updated initial cluster centers, match the cold plate evaluation sequence features with the evaluation labels, and establish cold plate portraits and historical cold plate portraits based on the evaluation labels.

5. A two-phase cold plate liquid cooling method according to claim 4, characterized in that The solution further includes: The cold plate information set includes a scheme name, heat dissipation test data, uniformity test data, and response test data. Generate a cold plate evaluation vector based on the heat dissipation test data, the uniformity test data, and the response test data. Each cold plate portrait corresponds to the scheme name; The historical cold plate information set includes a historical scheme name, historical heat dissipation test data, historical uniformity test data, and historical response test data. Generate a historical cold plate evaluation vector based on the historical heat dissipation test data, the historical uniformity test data, and the historical response test data. Each historical cold plate portrait corresponds to the historical scheme name.

6. A two-phase cold plate liquid cooling method according to claim 1, characterized in that The numbering of the cold plate portrait and the historical cold plate portrait includes: The numbering of the cold plate portrait and the numbering of the historical cold plate portrait both include a number segment, a time period, and a name segment; The numbering of the cold plate portrait is represented as NUM-DATE-NAME, and the numbering of the historical cold plate portrait is represented as NUM-DATE-HNAME; Among them, NUM represents the number segment, DATE represents the time period, NAME represents the name segment of the numbering of the cold plate portrait, and HNAME represents the name segment of the numbering of the historical cold plate portrait.

7. A two-phase cold plate liquid cooling method according to claim 1, characterized in that, Establish a solution database based on the cold plate portrait, the historical cold plate portrait, and the cold plate improvement direction, and establish a cold plate table structure in the solution database, including: The solution database selects a relational database. The cold plate table structure includes a cold plate information table, a historical cold plate information table, an improvement information table, and a recommended improvement information table; Establish links between the cold plate information table, the historical cold plate information table, the improvement information table, and the recommended improvement information table, and at the same time establish a database index. The primary keys of the cold plate information table and the historical cold plate information table are the numberings of the cold plate portrait and the historical cold plate portrait. The primary key of the improvement information table is the direction number, and the primary key of the recommended improvement information table is the numbering of the cold plate portrait.

8. A two-phase cold plate liquid cooling method according to claim 7, characterized in that The method further includes: When establishing the solution database, the storage area in the solution database is divided into Y intervals based on a fixed time span. The Y intervals include a cold interval, a hot interval, and a temporary storage interval. The cold interval is represented as a closed interval, the hot interval is represented as an unclosed interval, and the temporary storage interval is represented as an interval for data storage buffering. When the solution database stores data, the data is stored in the temporary storage interval. Based on the data storage time, an open or closed interval mark is given to the data. The data marked with an open interval is stored in the hot interval, and the data marked with a closed mark is stored in the cold interval.

9. A two-phase cold plate liquid cooling method according to claim 1, characterized in that The method further includes: Establishing a backup database, which is used to synchronously back up and store the data in the solution database; Obtaining an update countdown, which is represented as Z hours when the backup database is updated. When the update countdown ends, the backup database performs a storage update.

10. A two-phase cold plate liquid cooling system, characterized in that, Including: An acquisition module, which is used to acquire the cold plate design solution and the historical cold plate design solution. At the same time, based on the cold plate design solution and the historical cold plate design solution, a cold plate information set and a historical cold plate information set are acquired, and the cold plate improvement direction is obtained; A data processing module, which is used to preprocess the cold plate information set and the historical cold plate information set; A building module, which is used to perform clustering analysis on the cold plate information set and the historical cold plate information set, and build a cold plate portrait and a historical cold plate portrait; A database module, which is used to establish a solution database and build a cold plate table structure in the solution database; A numbering module, which is used to number the cold plate portrait and the historical cold plate portrait, and number the cold plate improvement direction; An improvement module, which is used to build a solution improvement model and output a recommended improvement direction based on the solution improvement model.

Images