AI-based chaos testing method and system for immersion liquid cooling server cabinet

Through the AI-based immersive liquid-cooled server cabinet chaos testing method, the multi-level working environment parameter queue and DFA algorithm are used to solve the problems of insufficient testing process fixation and environmental simulation in the existing technology, and more flexible and accurate liquid Cold server whole cabinet test.

CN118690125BActive Publication Date: 2024-11-26四川华鲲振宇智能科技有限责任公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411162898.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-23
Publication Date
2024-11-26
Estimated Expiration
2044-08-23

AI Technical Summary

Technical Problem

The test process of existing liquid-cooled server cabinet testing methods is fixed, making it difficult to adjust flexibly, and cannot effectively simulate a complex and changeable working environment, resulting in inaccurate test results.

Method used

Using AI-based immersive liquid-cooled server complete cabinet chaos testing method, the chaos simulation system and data acquisition module are deployed, multi-level working environment parameter queue is established, performance parameter data is collected in real time, exception processing and environment simulation is performed, and DFA algorithm is used for Data preprocessing and analysis.

Benefits of technology

It realizes more flexible and comprehensive environmental simulation and more powerful real-time adaptability adjustments, improves the accuracy and adaptability of tests, and enhances the system's processing efficiency and accuracy of input data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118690125B_ABST
    Figure CN118690125B_ABST
Patent Text Reader

Abstract

The present invention relates to an AI-based chaos testing method and system for an immersion liquid-cooled server cabinet, which deploys a chaos simulation system and a data acquisition module; simulates the first working environment parameter of a liquid-cooled server cabinet according to a first working environment parameter queue; collects data of various performance parameters, and determines whether various performance parameters are abnormal. Simulates the second working environment parameter, collects data of various performance parameters of each liquid-cooled server within a specified time length, and determines whether various performance parameters are abnormal; establishes a second working environment parameter queue in the chaos simulation system, simulates the working environment of the liquid-cooled server cabinet in turn according to the second working environment parameter queue, collects data of various performance parameters of each liquid-cooled server, and determines whether the cabinet is abnormal; simulates the working environment of the liquid-cooled server cabinet in turn according to the third working environment parameter queue, and determines whether each liquid-cooled server cabinet is abnormal. A more comprehensive and flexible environmental simulation is achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of server testing methods, and in particular to an AI-based immersion liquid-cooled server cabinet chaos testing method and system. Background Art

[0002] With the continuous development of data center and server technology, liquid cooling technology plays an increasingly important role in improving server performance and energy efficiency. Liquid cooling technology improves heat dissipation efficiency and reduces energy consumption by directly contacting the cooling medium with the heat source, and has gradually become one of the key technologies in the field of high-performance computing.

[0003] The test process of the liquid-cooled server cabinet test in the prior art is relatively fixed, including setting predetermined test conditions, such as fixed load, temperature, humidity, etc. The test process is generally executed as planned and cannot be flexibly adjusted. During the test process, the operating status of the entire server cabinet is simulated mainly by setting predetermined load conditions. These loads are usually defined in advance and will not be dynamically adjusted according to actual conditions. The simulation of the environment is relatively limited, and usually only simple load changes are considered, making it difficult to simulate a real and complex and changeable working environment. For the special environment of the liquid cooling system, the factors considered are limited. There is a lack of a mechanism for real-time adaptive adjustment during the test process. Once the test conditions are set, it is difficult to flexibly respond to fluctuations in system performance or environmental changes, resulting in test results that may not be accurate enough.

[0004] Therefore, how to improve the existing testing methods for the entire cabinet of liquid-cooled servers to achieve a more flexible testing process, more comprehensive environmental simulation, and more powerful real-time adaptive adjustment, so as to better meet the requirements of the chaos testing of the entire cabinet of liquid-cooled servers, is a technical problem that urgently needs to be solved. Summary of the invention

[0005] The purpose of the present invention is to improve the existing test method for testing the entire cabinet of a liquid-cooled server, to achieve a more flexible test process, a more comprehensive environmental simulation and a more powerful real-time adaptive adjustment, so as to better meet the requirements of the chaos test of the entire cabinet of a liquid-cooled server.

[0006] First, a chaos testing method for an immersion liquid cooling server cabinet based on AI is provided, comprising the following steps:

[0007] S1: deploying a chaos simulation system, establishing a communication connection between the chaos simulation system and a number of liquid cooling server cabinets, and deploying a data acquisition module inside each liquid cooling server;

[0008] S2: establishing a first working environment parameter queue in the chaos simulation system, assigning priorities to the working environment parameters in the first working environment parameter queue according to the degree of influence of the working environment parameters on the liquid-cooled server cabinet, reordering the working environment parameters in the queue according to the priority, and simulating the first working environment parameters of the liquid-cooled server cabinet according to the first working environment parameter queue;

[0009] S3: The data acquisition module collects the performance parameter data of each liquid-cooled server within a specified time length in real time, and compares the data with the corresponding preset performance parameter ranges to determine whether there are abnormalities in each performance parameter. If not, execute step S4. If so, perform abnormal processing on the liquid-cooled server cabinet with abnormal performance parameters;

[0010] S4: The second working environment parameter in the first working environment parameter queue is simulated in the liquid-cooled server cabinet by the chaos simulation system. The data acquisition module collects the performance parameter data of each liquid-cooled server within the specified time length in real time, and compares it with the corresponding preset performance parameter ranges to determine whether there is an abnormality in each performance parameter. If not, execute step S5. If so, perform abnormal processing on the liquid-cooled server cabinet with abnormal performance parameters.

[0011] S5: Simulate the other working environment parameters of the first working environment parameter queue in turn, collect the performance parameter data of each liquid-cooled server after each working environment parameter simulation, and determine whether each liquid-cooled server cabinet has an abnormality, if not, execute step S6; if yes, perform abnormal processing on the liquid-cooled server cabinet with the abnormality;

[0012] S6: Establish a second working environment parameter queue in the chaos simulation system, where the second working environment parameter queue is obtained by combining two random working environment parameters in the first working environment parameter queue, and simulate the working environment of the liquid-cooled server cabinet in turn according to the second working environment parameter queue, and collect performance parameter data of each liquid-cooled server after each working environment parameter simulation, and determine whether each liquid-cooled server cabinet has an abnormality, if not, execute step S7; if yes, perform abnormal processing on the liquid-cooled server cabinet with the abnormality;

[0013] S7: Establish a third working environment parameter queue in the chaos simulation system, wherein the third working environment parameter queue is obtained by randomly combining three working environment parameters in the first working environment parameter queue, and simulate the working environment of the liquid-cooled server cabinet in sequence according to the third working environment parameter queue, and collect performance parameter data of each liquid-cooled server after each working environment parameter simulation, and determine whether each liquid-cooled server cabinet has an abnormality, if not, execute step S8; if yes, perform abnormal processing on the liquid-cooled server cabinet with the abnormality;

[0014] S8: Simulate other combinations of working environment parameters in the first working environment parameter queue in turn, and collect performance parameter data of each liquid-cooled server after each working environment parameter simulation, and determine whether there is an abnormality in each liquid-cooled server cabinet. If so, perform abnormal processing on the liquid-cooled server cabinet with the abnormality. If not, complete the chaos test.

[0015] Preferably, the working environment parameters of the liquid-cooled server simulated by the chaos simulation system include at least power supply fluctuations, temperature changes, network delays and voltage changes.

[0016] Preferably, after collecting various performance parameter data of the liquid-cooled server cabinet each time, data preprocessing is performed on the various performance parameter data. The data preprocessing process adopts the DFA algorithm. The specific process is as follows:

[0017] S31: Cleaning invalid data and duplicate data in the collected performance parameter data of the liquid cooling server cabinet;

[0018] S32: performing abnormal data processing on various performance parameter data of the cleaned liquid cooling server cabinet, including correction and filling of the abnormal data;

[0019] S33: normalizing the performance parameter data of the liquid-cooled server cabinet after the abnormal data processing.

[0020] Preferably, the first working environment parameter queue is set to [A, B, C, D], where A is temperature change, B is power supply fluctuation, C is network delay, and D is voltage change;

[0021] Four sub-working environment parameter queues are established in the first working environment parameter queue, namely:

[0022] [A1, A2, A3], [B1, B2, B3], [C1, C2, C3], [D1, D2, D3];

[0023] Wherein, A1 is the first temperature change, A2 is the second temperature change, and A3 is the third temperature change. A1, A2, and A3 respectively represent temperature changes of different gradients. The temperature change gradient of A1 is smaller than the temperature change gradient of A2, and the temperature change gradient of A2 is smaller than the temperature change gradient of A3.

[0024] B1 is the first power supply fluctuation, B2 is the second power supply fluctuation, and B3 is the third power supply fluctuation. B1, B2, and B3 represent power supply fluctuations of different degrees respectively. The power supply fluctuation degree of B1 is less than that of B2, and the power supply fluctuation degree of B2 is less than that of B3.

[0025] C1 is the first network delay, C2 is the second network delay, and C3 is the third network delay. C1, C2, and C3 represent different degrees of network delay respectively. The network delay degree of C1 is less than that of C2, and the network delay degree of C2 is less than that of C3.

[0026] D1 is a first voltage change, D2 is a second voltage change, and D3 is a third voltage change. D1, D2, and D3 respectively represent voltage changes of different degrees. The voltage change degree of D1 is smaller than that of D2, and the voltage change degree of D2 is smaller than that of D3.

[0027] Preferably, the second working environment parameter queue is [AB, AC, AD, BC, BD, CD],

[0028] Among them, AB means simulating both temperature change and power supply fluctuation, AC means simulating both temperature change and network delay, AD means simulating both temperature change and voltage change, BC means simulating both power supply fluctuation and network delay, and BD means simulating both power supply fluctuation and voltage change.

[0029] A is temperature change, B is power supply fluctuation, C is network delay, and D is voltage change;

[0030] Preferably, the third working environment parameter queue is [ABC, ABD, ACD, BCD],

[0031] Among them, ABC means simulating three working environment parameters: temperature change, power supply fluctuation and network delay at the same time; ABD means simulating three working environment parameters: temperature change, power supply fluctuation and voltage change at the same time; ACD means simulating three working environment parameters: temperature change, network delay and voltage change at the same time; BCD means simulating three working environment parameters: power supply fluctuation, network delay and voltage change at the same time;

[0032] A fourth working environment parameter queue is also set, and the fourth working environment parameter queue is [ABCD], wherein ABCD represents simultaneous simulation of four working environment parameters: temperature change, power supply fluctuation, network delay, and voltage change.

[0033] Preferably, while simulating various working environment parameters through the chaos simulation system, the load of each server node in each liquid-cooled server cabinet is automatically adjusted, and various performance parameters of each server node in each liquid-cooled server cabinet under different loads are recorded.

[0034] Preferably, after collecting the performance parameter data of each liquid cooling server within a specified time length, the performance parameter data are analyzed to determine whether there is an abnormality in each performance parameter, which is achieved by analyzing with a DFA algorithm.

[0035] Preferably, abnormal processing of a liquid-cooled server cabinet having abnormal performance parameters includes adjusting technical parameters of the liquid-cooled server cabinet, including adjusting the flow velocity of the liquid-cooling medium, and adjusting the voltage and current.

[0036] In a second aspect, an AI-based immersion liquid-cooled server cabinet chaos testing system is provided, which is used to implement any one of the AI-based immersion liquid-cooled server cabinet chaos testing methods, including a chaos simulation system, a data acquisition module, a data preprocessing module, a main control module, an exception handling module and a visualization module, wherein the chaos simulation system, the data acquisition module, the data preprocessing module, the exception handling module and the visualization module are all connected to the main control module, the chaos simulation system is communicatively connected to a number of server liquid-cooling cabinets, and the server liquid-cooling cabinets are respectively connected to a number of data acquisition modules, the data acquisition module is connected to the data preprocessing module, the data preprocessing module is connected to the main control module, and the main control module is respectively connected to the exception handling module and the visualization module;

[0037] The chaos simulation system is used to realize a comprehensive simulation of the working environment of an immersion liquid cooling server cabinet based on chaos theory;

[0038] The data acquisition module is used to collect data on various performance parameters of each immersion liquid cooling server cabinet after the chaos simulation system is started;

[0039] The data preprocessing module is used to preprocess the performance parameter data collected by the data collection module, including data cleaning, correction and filling of abnormal data, and data normalization;

[0040] The main control module is used to control the start and stop of the chaos simulation system, the data acquisition module, the data preprocessing module, the exception handling module and the visual display module, and to perform data analysis and processing;

[0041] The exception handling module is used to handle the abnormality of the liquid-cooled server cabinet when it works abnormally in the simulated working environment of the chaos simulation system, including adjusting the technical parameters of the liquid-cooled server cabinet, including adjusting the flow velocity of the liquid cooling medium, adjusting the voltage and current, and detecting abnormal components and performing maintenance on the abnormal components;

[0042] The visualization display module is used to visualize the collected performance data, abnormal data of the liquid-cooled server cabinet, and the data related to the abnormal processing of the abnormal processing module.

[0043] The beneficial effects of the present invention include:

[0044] The AI-based chaos testing method and system for the whole cabinet of an immersion liquid-cooled server provided by the present invention deploys a chaos simulation system and a data acquisition module; simulates the first working environment parameter of the liquid-cooled server cabinet according to the first working environment parameter queue; collects data of each performance parameter, and determines whether each performance parameter is abnormal. Simulates the second working environment parameter, collects data of each performance parameter of each liquid-cooled server within a specified time length, and determines whether each performance parameter is abnormal; establishes a second working environment parameter queue in the chaos simulation system, simulates the working environment of the liquid-cooled server cabinet in turn according to the second working environment parameter queue, collects data of each performance parameter of each liquid-cooled server, and determines whether the cabinet is abnormal, simulates the working environment of the liquid-cooled server cabinet in turn according to the third working environment parameter queue, and determines whether each liquid-cooled server cabinet is abnormal.

[0045] On the one hand, by establishing a multi-level working environment parameter queue, various working environments of the liquid-cooled server cabinet are simulated in turn according to the multi-level working environment parameter queue and the priority of each level of queue, a more comprehensive, flexible and realistic environmental simulation of the chaos simulation system is achieved.

[0046] On the other hand, by adopting the DFA algorithm for data processing and analysis, the accuracy and adaptability of the liquid-cooled server cabinet test are improved.

[0047] On the other hand, the DFA algorithm is used to pre-process the various performance parameter data collected in real time, thereby improving the system's processing efficiency and accuracy of input data. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 The figure is a flow chart of the AI-based immersion liquid-cooled server cabinet chaos testing method of the present invention.

[0049] Figure 2 It is a structural schematic diagram of a chaos test system in the prior art.

[0050] Figure 3 It is a structural schematic diagram of the AI-based immersion liquid-cooled server cabinet chaos testing system of the present invention. DETAILED DESCRIPTION

[0051] The following is combined with Figure 1-3 The present invention is further described in detail:

[0052] See also Figure 1 , the AI-based immersion liquid cooling server cabinet chaos testing method includes the following steps:

[0053] S1: deploying a chaos simulation system, establishing a communication connection between the chaos simulation system and a number of liquid cooling server cabinets, and deploying a data acquisition module inside each liquid cooling server;

[0054] S2: establishing a first working environment parameter queue in the chaos simulation system, assigning priorities to the working environment parameters in the first working environment parameter queue according to the degree of influence of the working environment parameters on the liquid-cooled server cabinet, reordering the working environment parameters in the queue according to the priority, and simulating the first working environment parameters of the liquid-cooled server cabinet according to the first working environment parameter queue;

[0055] S3: The data acquisition module collects the performance parameter data of each liquid-cooled server within a specified time length in real time, and compares the data with the corresponding preset performance parameter ranges to determine whether there are abnormalities in each performance parameter. If not, execute step S4. If so, perform abnormal processing on the liquid-cooled server cabinet with abnormal performance parameters;

[0056] S4: The second working environment parameter in the first working environment parameter queue is simulated in the liquid-cooled server cabinet by the chaos simulation system. The data acquisition module collects the performance parameter data of each liquid-cooled server within the specified time length in real time, and compares it with the corresponding preset performance parameter ranges to determine whether there is an abnormality in each performance parameter. If not, execute step S5. If so, perform abnormal processing on the liquid-cooled server cabinet with abnormal performance parameters.

[0057] S5: Simulate the other working environment parameters of the first working environment parameter queue in turn, collect the performance parameter data of each liquid-cooled server after each working environment parameter simulation, and determine whether each liquid-cooled server cabinet has an abnormality, if not, execute step S6; if yes, perform abnormal processing on the liquid-cooled server cabinet with the abnormality;

[0058] S6: Establish a second working environment parameter queue in the chaos simulation system, where the second working environment parameter queue is obtained by combining two random working environment parameters in the first working environment parameter queue, and simulate the working environment of the liquid-cooled server cabinet in turn according to the second working environment parameter queue, and collect performance parameter data of each liquid-cooled server after each working environment parameter simulation, and determine whether each liquid-cooled server cabinet has an abnormality, if not, execute step S7; if yes, perform abnormal processing on the liquid-cooled server cabinet with the abnormality;

[0059] S7: Establish a third working environment parameter queue in the chaos simulation system, wherein the third working environment parameter queue is obtained by randomly combining three working environment parameters in the first working environment parameter queue, and simulate the working environment of the liquid-cooled server cabinet in sequence according to the third working environment parameter queue, and collect performance parameter data of each liquid-cooled server after each working environment parameter simulation, and determine whether each liquid-cooled server cabinet has an abnormality, if not, execute step S8; if yes, perform abnormal processing on the liquid-cooled server cabinet with the abnormality;

[0060] S8: Simulate other combinations of working environment parameters in the first working environment parameter queue in turn, and collect performance parameter data of each liquid-cooled server after each working environment parameter simulation, and determine whether there is an abnormality in each liquid-cooled server cabinet. If so, perform abnormal processing on the liquid-cooled server cabinet with the abnormality. If not, complete the chaos test.

[0061] Since the relevant environmental parameters of the actual working environment of the immersed liquid-cooled server cabinet are uncertain, for example, the temperature of its specific working environment changes in real time, and its power supply fluctuation may or may not exist. When there is power supply fluctuation, the degree of fluctuation is also uncertain, and the degree of fluctuation may be large or small. The network delay of the actual working environment may also exist or may not exist. When there is network delay, the degree of network delay is also uncertain, and the degree of network delay may be large or small. The other environmental parameters of the working environment are also uncertain, and one or two or three or any other combination of temperature changes, power supply fluctuations, network delays or other environmental parameters may occur. Therefore, when simulating the working environment of the immersed liquid-cooled server cabinet through a chaotic simulation system, a comprehensive simulation is required, and the working environment parameters of the liquid-cooled server simulated by the chaotic simulation system include at least power supply fluctuations, temperature changes, network delays and voltage changes.

[0062] Therefore, the AI-based chaos testing method for the entire cabinet of an immersion liquid-cooled server of the present invention first deploys a chaos simulation system, establishes a communication connection between the chaos simulation system and several liquid-cooled server cabinets, and deploys a data acquisition module inside each liquid-cooled server. After the deployment is completed, before simulating the working environment through the chaos simulation system, it is first necessary to establish a working environment parameter queue, which includes all possible environmental parameters. Since the impact of each environmental parameter on its working state is different during the operation of the liquid-cooled server cabinet, it is necessary to prioritize the impact of each environmental parameter, and then prioritize the simulation of the working environment parameters with a greater impact, and then re-arrange the working environment parameters in the working environment parameter queue according to the impact of each environmental parameter, and then simulate a single working environment parameter in the re-arranged order of the working environment parameters.

[0063] For example, simulate a working environment where only temperature changes occur, and then collect the performance parameters of the liquid-cooled server cabinet in the working environment where only temperature changes occur. When an abnormality occurs, handle the abnormality to improve the ability of the liquid-cooled cabinet to operate normally under temperature changes. Then simulate a working environment where only power fluctuations occur, and then collect the performance parameters of the liquid-cooled server cabinet in the working environment where only power fluctuations occur. When an abnormality occurs, handle the abnormality to improve the ability of the liquid-cooled cabinet to operate normally under power fluctuations. Then simulate a working environment where only network delays or voltage changes occur, and then collect the performance parameters of the liquid-cooled server cabinet in the working environment where only network delays or voltage changes occur. When an abnormality occurs, handle the abnormality to improve the ability of the liquid-cooled cabinet to operate normally under network delays or voltage changes.

[0064] However, during the operation of the liquid-cooled server cabinet, one, two, three, four or more environmental parameters such as power fluctuations or temperature changes or network delays or voltage changes may appear. Therefore, it is necessary to simulate two, three and multiple types one by one to simulate all possible working environment states and ensure the comprehensiveness of the simulated working environment. Therefore, after simulating each working environment parameter in the first working environment parameter queue, a second working environment parameter queue is established. The second working environment parameter queue is obtained by combining two random working environment parameters in the first working environment parameter queue. Similarly, the two working environment parameters are reordered according to the degree of influence of the liquid-cooled server cabinet, and the working environment of the liquid-cooled server cabinet is simulated in sequence according to the reordered second working environment parameter queue. After each working environment parameter simulation, the performance parameter data of each liquid-cooled server is collected, and it is determined whether each liquid-cooled server cabinet has an abnormality. When an abnormality exists, the abnormality handling module performs corresponding processing, including adjustment of technical parameters or maintenance processing of related components.

[0065] Then establish a third working environment parameter queue. The third working environment parameter queue is obtained by randomly combining three working environment parameters in the first working environment parameter queue. The priority is set according to the degree of influence. The parameters are reordered according to the priority. The working environment of the liquid-cooled server cabinet is simulated in turn according to the reordered third working environment parameter queue. After each working environment parameter simulation, the performance parameter data of each liquid-cooled server is collected, and it is determined whether there is an abnormality in each liquid-cooled server cabinet. When an abnormality exists, corresponding processing is performed through the exception handling module, including adjustment of technical parameters or maintenance processing of related components.

[0066] When there are 4 working environment parameters in the first working environment parameter queue, a fourth working environment parameter queue is also established. When the first working environment parameter queue is [A, B, C, D], the fourth working environment parameter queue at this time is [ABCD]. When simulating the working environment parameters in the fourth working environment parameter queue, the four working environment parameters ABCD are simulated at the same time, A is temperature change, B is power supply fluctuation, C is network delay, and D is voltage change, that is, temperature change, power supply fluctuation, network delay and voltage change are simulated at the same time, and the performance status of the server liquid cooling cabinet when temperature change, power supply fluctuation, network delay and voltage change exist at the same time is collected. Similarly, when there is an abnormality, the abnormality handling module performs corresponding processing, including adjustment of technical parameters or maintenance processing of related components.

[0067] Similarly, when there are 5 working environment parameters in a working environment parameter queue, a fourth working environment parameter queue is also established. That is to say, no matter how many working environment parameters the liquid cooling server has, all possible working environment simulations are performed through the chaos simulation system. After collecting the performance parameter data of each liquid cooling server within a specified time length, the performance parameter data is analyzed to determine whether there are any abnormalities in each performance parameter, which is achieved through the DFA algorithm.

[0068] In the above scheme, after collecting the performance parameter data of the liquid-cooled server cabinet each time, the performance parameter data is preprocessed. The data preprocessing process adopts the DFA algorithm. The specific process is as follows:

[0069] S31: Cleaning invalid data and duplicate data in the collected performance parameter data of the liquid cooling server cabinet;

[0070] S32: performing abnormal data processing on various performance parameter data of the cleaned liquid cooling server cabinet, including correction and filling of the abnormal data;

[0071] S33: normalizing the performance parameter data of the liquid-cooled server cabinet after the abnormal data processing.

[0072] In one embodiment, let the first working environment parameter queue be [A, B, C, D], where A is temperature change, B is power supply fluctuation, C is network delay, and D is voltage change;

[0073] Four sub-working environment parameter queues are established in the first working environment parameter queue, namely:

[0074] [A1, A2, A3], [B1, B2, B3], [C1, C2, C3], [D1, D2, D3];

[0075] Wherein, A1 is the first temperature change, A2 is the second temperature change, and A3 is the third temperature change. A1, A2, and A3 respectively represent temperature changes of different gradients. The temperature change gradient of A1 is smaller than the temperature change gradient of A2, and the temperature change gradient of A2 is smaller than the temperature change gradient of A3.

[0076] B1 is the first power supply fluctuation, B2 is the second power supply fluctuation, and B3 is the third power supply fluctuation. B1, B2, and B3 represent power supply fluctuations of different degrees respectively. The power supply fluctuation degree of B1 is less than that of B2, and the power supply fluctuation degree of B2 is less than that of B3.

[0077] C1 is the first network delay, C2 is the second network delay, and C3 is the third network delay. C1, C2, and C3 represent different degrees of network delay respectively. The network delay degree of C1 is less than that of C2, and the network delay degree of C2 is less than that of C3.

[0078] D1 is a first voltage change, D2 is a second voltage change, and D3 is a third voltage change. D1, D2, and D3 respectively represent voltage changes of different degrees. The voltage change degree of D1 is smaller than that of D2, and the voltage change degree of D2 is smaller than that of D3.

[0079] The second working environment parameter queue is [AB, AC, AD, BC, BD, CD], wherein AB represents the simultaneous simulation of two working environment parameters, temperature change and power supply fluctuation, AC represents the simultaneous simulation of two working environment parameters, temperature change and network delay, AD represents the simultaneous simulation of two working environment parameters, temperature change and voltage change, BC represents the simultaneous simulation of two working environment parameters, power supply fluctuation and network delay, and BD represents the simultaneous simulation of two working environment parameters, power supply fluctuation and voltage change.

[0080] The third working environment parameter queue is [ABC, ABD, ACD, BCD], where ABC represents the simultaneous simulation of three working environment parameters: temperature change, power supply fluctuation, and network delay; ABD represents the simultaneous simulation of three working environment parameters: temperature change, power supply fluctuation, and voltage change; ACD represents the simultaneous simulation of three working environment parameters: temperature change, network delay, and voltage change; and BCD represents the simultaneous simulation of three working environment parameters: power supply fluctuation, network delay, and voltage change. A fourth working environment parameter queue is also set, and the fourth working environment parameter queue is [ABCD], where ABCD represents the simultaneous simulation of four working environment parameters: temperature change, power supply fluctuation, network delay, and voltage change.

[0081] Different degrees are set for different working environment parameters in the second working environment parameter queue and the third working environment parameter queue, such as temperature changes of different gradients, power supply fluctuations of different degrees, network delays of different degrees, and voltage changes of different degrees, so as to further simulate various possible working environments in a more refined manner.

[0082] While simulating various working environment parameters through the chaos simulation system, the load of each server node in each liquid-cooled server cabinet is automatically adjusted based on AI to simulate the performance of the system under different workloads and record the various performance parameters of each server node in each liquid-cooled server cabinet under different loads.

[0083] Abnormal processing for a liquid-cooled server cabinet with abnormal performance parameters includes adjusting the technical parameters of the liquid-cooled server cabinet, including adjusting the flow speed of the liquid cooling medium, and adjusting the voltage and current.

[0084] See also Figure 3 In the prior art, a chaotic simulation system is only deployed in the server liquid cooling cabinet to simulate the working environment of the server liquid cooling cabinet, and then the performance parameters of the server liquid cooling cabinet are collected through a data acquisition module in the server liquid cooling cabinet, and the performance of the server liquid cooling cabinet under different working environments is obtained based on the collected performance data.

[0085] See also Figure 2 The AI-based immersion liquid-cooling server cabinet chaos testing system provided by the present invention is used to implement any one of the AI-based immersion liquid-cooling server cabinet chaos testing methods, including a chaos simulation system, a data acquisition module, a data preprocessing module, a main control module, an exception handling module and a visualization module. The chaos simulation system, the data acquisition module, the data preprocessing module, the exception handling module and the visualization module are all connected to the main control module, the chaos simulation system is communicatively connected to a number of server liquid-cooling cabinets, and the server liquid-cooling cabinets are respectively connected to a number of data acquisition modules, the data acquisition module is connected to the data preprocessing module, the data preprocessing module is connected to the main control module, and the main control module is respectively connected to the exception handling module and the visualization module.

[0086] The chaos simulation system is used to realize a comprehensive simulation of the working environment of the immersion liquid cooling server cabinet based on chaos theory. The data acquisition module is used to collect data on various performance parameters of each immersion liquid cooling server cabinet after the chaos simulation system is started. The data preprocessing module is used to preprocess the various performance parameter data collected by the data acquisition module, including data cleaning, correction and filling of abnormal data, and data normalization. The main control module is used to control the start and shutdown of the chaos simulation system, the data acquisition module, the data preprocessing module, the abnormality handling module and the visual display module, and perform data analysis and processing. The abnormality handling module is used to perform abnormality processing when the liquid cooling server cabinet works abnormally in the simulated working environment of the chaos simulation system, including adjusting the technical parameters of the liquid cooling server cabinet, including adjusting the flow velocity of the liquid cooling medium, adjusting the voltage and current, and detecting abnormal components and performing maintenance processing on the abnormal components. The visual display module is used to visually display the various performance data, abnormal data and related data of the abnormality processing of the abnormality handling module of the collected liquid cooling server cabinet.

[0087] In summary, by deploying a chaos simulation system and a data acquisition module; simulating the first working environment parameter of a liquid-cooled server cabinet according to the first working environment parameter queue; collecting data of each performance parameter, and judging whether each performance parameter is abnormal. Simulating the second working environment parameter, collecting data of each performance parameter of each liquid-cooled server within a specified time length, and judging whether each performance parameter is abnormal; establishing a second working environment parameter queue in the chaos simulation system, simulating the working environment of the liquid-cooled server cabinet in turn according to the second working environment parameter queue, collecting data of each performance parameter of each liquid-cooled server, and judging whether the cabinet is abnormal, simulating the working environment of the liquid-cooled server cabinet in turn according to the third working environment parameter queue, and judging whether each liquid-cooled server cabinet is abnormal.

[0088] By establishing a multi-level working environment parameter queue, various working environments of the liquid-cooled server cabinet are simulated in turn according to the multi-level working environment parameter queue and the priority of each level of queue, a more comprehensive, flexible and realistic environmental simulation of the chaos simulation system is achieved. Then, the accuracy and adaptability of the liquid-cooled server cabinet test are improved by using the DFA algorithm for data processing and analysis. Finally, the DFA algorithm is used to pre-process the various performance parameter data collected in real time, which improves the system's processing efficiency and accuracy of input data.

Claims

1. An AI-based chaotic testing method for an immersion liquid-cooled server cabinet, characterized in that: The following steps are involved: S1: deploying a chaos simulation system, establishing a communication connection between the chaos simulation system and a number of liquid cooling server cabinets, and deploying a data acquisition module inside each liquid cooling server; S2: establishing a first working environment parameter queue in the chaos simulation system, assigning priorities to the working environment parameters in the first working environment parameter queue according to the degree of influence of the working environment parameters on the liquid-cooled server cabinet, reordering the working environment parameters in the queue according to the priority, and simulating the first working environment parameters of the liquid-cooled server cabinet according to the first working environment parameter queue; S3: The data acquisition module collects the performance parameter data of each liquid-cooled server within a specified time length in real time, and compares the data with the corresponding preset performance parameter ranges to determine whether there are abnormalities in each performance parameter. If not, execute step S4. If so, perform abnormal processing on the liquid-cooled server cabinet with abnormal performance parameters; S4: The second working environment parameter in the first working environment parameter queue is simulated in the liquid-cooled server cabinet by the chaos simulation system. The data acquisition module collects the performance parameter data of each liquid-cooled server within the specified time length in real time, and compares it with the corresponding preset performance parameter ranges to determine whether there is an abnormality in each performance parameter. If not, execute step S5. If so, perform abnormal processing on the liquid-cooled server cabinet with abnormal performance parameters. S5: Simulate the other working environment parameters of the first working environment parameter queue in turn, collect the performance parameter data of each liquid-cooled server after each working environment parameter simulation, and determine whether each liquid-cooled server cabinet has an abnormality, if not, execute step S6; if yes, perform abnormal processing on the liquid-cooled server cabinet with the abnormality; S6: Establish a second working environment parameter queue in the chaos simulation system, where the second working environment parameter queue is obtained by combining two random working environment parameters in the first working environment parameter queue, and simulate the working environment of the liquid-cooled server cabinet in turn according to the second working environment parameter queue, and collect performance parameter data of each liquid-cooled server after each working environment parameter simulation, and determine whether each liquid-cooled server cabinet has an abnormality, if not, execute step S7; if yes, perform abnormal processing on the liquid-cooled server cabinet with the abnormality; S7: Establish a third working environment parameter queue in the chaos simulation system, wherein the third working environment parameter queue is obtained by randomly combining three working environment parameters in the first working environment parameter queue, and simulate the working environment of the liquid-cooled server cabinet in sequence according to the third working environment parameter queue, and collect performance parameter data of each liquid-cooled server after each working environment parameter simulation, and determine whether each liquid-cooled server cabinet has an abnormality, if not, execute step S8; if yes, perform abnormal processing on the liquid-cooled server cabinet with the abnormality; S8: Simulate other combinations of working environment parameters in the first working environment parameter queue in turn, and collect performance parameter data of each liquid-cooled server after each working environment parameter simulation, and determine whether there is an abnormality in each liquid-cooled server cabinet. If so, perform abnormal processing on the liquid-cooled server cabinet with the abnormality. If not, complete the chaos test.

2. The AI-based immersion liquid cooling server cabinet chaos testing method according to claim 1 is characterized in that: The working environment parameters of the liquid-cooled server simulated by the chaos simulation system include at least power supply fluctuations, temperature changes, network delays and voltage changes.

3. The AI-based immersion liquid cooling server cabinet chaos testing method according to claim 1 is characterized in that: After collecting the performance parameter data of the liquid-cooled server cabinet each time, the performance parameter data is preprocessed. The data preprocessing process uses the DFA algorithm. The specific process is as follows: S31: Cleaning invalid data and duplicate data in the collected performance parameter data of the liquid cooling server cabinet; S32: performing abnormal data processing on various performance parameter data of the cleaned liquid cooling server cabinet, including correction and filling of the abnormal data; S33: normalizing the performance parameter data of the liquid-cooled server cabinet after the abnormal data processing.

4. The AI-based immersion liquid cooling server cabinet chaos testing method according to claim 1 is characterized in that: Let the first working environment parameter queue be [A, B, C, D], where A is temperature change, B is power supply fluctuation, C is network delay, and D is voltage change; Four sub-working environment parameter queues are established in the first working environment parameter queue, namely: [A1, A2, A3], [B1, B2, B3], [C1, C2, C3], [D1, D2, D3]; Wherein, A1 is the first temperature change, A2 is the second temperature change, and A3 is the third temperature change. A1, A2, and A3 respectively represent temperature changes of different gradients. The temperature change gradient of A1 is smaller than the temperature change gradient of A2, and the temperature change gradient of A2 is smaller than the temperature change gradient of A3. B1 is the first power supply fluctuation, B2 is the second power supply fluctuation, and B3 is the third power supply fluctuation. B1, B2, and B3 represent power supply fluctuations of different degrees respectively. The power supply fluctuation degree of B1 is less than that of B2, and the power supply fluctuation degree of B2 is less than that of B3. C1 is the first network delay, C2 is the second network delay, and C3 is the third network delay. C1, C2, and C3 represent different degrees of network delay respectively. The network delay degree of C1 is less than that of C2, and the network delay degree of C2 is less than that of C3. D1 is a first voltage change, D2 is a second voltage change, and D3 is a third voltage change. D1, D2, and D3 respectively represent voltage changes of different degrees. The voltage change degree of D1 is smaller than that of D2, and the voltage change degree of D2 is smaller than that of D3.

5. The AI-based immersion liquid cooling server cabinet chaos testing method according to claim 4 is characterized in that: The second working environment parameter queue is [AB, AC, AD, BC, BD, CD], Among them, AB means simulating both temperature change and power supply fluctuation, AC means simulating both temperature change and network delay, AD means simulating both temperature change and voltage change, BC means simulating both power supply fluctuation and network delay, and BD means simulating both power supply fluctuation and voltage change.

6. The AI-based immersion liquid cooling server cabinet chaos testing method according to claim 1 is characterized in that: The third working environment parameter queue is [ABC, ABD, ACD, BCD], Among them, ABC means simulating three working environment parameters: temperature change, power supply fluctuation and network delay at the same time; ABD means simulating three working environment parameters: temperature change, power supply fluctuation and voltage change at the same time; ACD means simulating three working environment parameters: temperature change, network delay and voltage change at the same time; BCD means simulating three working environment parameters: power supply fluctuation, network delay and voltage change at the same time; A fourth working environment parameter queue is also set, and the fourth working environment parameter queue is [ABCD], wherein ABCD represents simultaneous simulation of four working environment parameters: temperature change, power supply fluctuation, network delay, and voltage change.

7. The AI-based immersion liquid cooling server cabinet chaos testing method according to claim 1 is characterized in that: While simulating various working environment parameters through the chaos simulation system, the load of each server node in each liquid-cooled server cabinet is automatically adjusted, and various performance parameters of each server node in each liquid-cooled server cabinet under different loads are recorded.

8. The AI-based immersion liquid cooling server cabinet chaos testing method according to claim 1 is characterized in that: After collecting the performance parameter data of each liquid-cooled server within a specified time length, the performance parameter data are analyzed to determine whether there are abnormalities in each performance parameter, which is achieved through the DFA algorithm.

9. The AI-based immersion liquid cooling server cabinet chaos testing method according to claim 1 is characterized in that: Abnormal processing for a liquid-cooled server cabinet with abnormal performance parameters includes adjusting the technical parameters of the liquid-cooled server cabinet, including adjusting the flow velocity of the liquid cooling medium, and adjusting the voltage and current.

10. An AI-based immersion liquid cooling server cabinet chaos testing system, used to implement the AI-based immersion liquid cooling server cabinet chaos testing method according to any one of claims 1 to 9, characterized in that: It includes a chaos simulation system, a data acquisition module, a data preprocessing module, a main control module, an exception handling module and a visual display module. The chaos simulation system, the data acquisition module, the data preprocessing module, the exception handling module and the visual display module are all connected to the main control module. The chaos simulation system is communicatively connected to a number of server liquid cooling cabinets. The server liquid cooling cabinets are respectively connected to a number of data acquisition modules. The data acquisition module is connected to the data preprocessing module. The data preprocessing module is connected to the main control module. The main control module is respectively connected to the exception handling module and the visual display module. The chaos simulation system is used to realize a comprehensive simulation of the working environment of an immersion liquid-cooled server cabinet based on chaos theory; The data acquisition module is used to collect data on various performance parameters of each immersion liquid cooling server cabinet after the chaos simulation system is started; The data preprocessing module is used to preprocess the performance parameter data collected by the data collection module, including data cleaning, correction and filling of abnormal data, and data normalization; The main control module is used to control the start and stop of the chaos simulation system, the data acquisition module, the data preprocessing module, the exception handling module and the visual display module, and to perform data analysis and processing; The exception handling module is used to handle the abnormality of the liquid-cooled server cabinet when it works abnormally in the simulated working environment of the chaos simulation system, including adjusting the technical parameters of the liquid-cooled server cabinet, including adjusting the flow velocity of the liquid cooling medium, adjusting the voltage and current, and detecting abnormal components and performing maintenance on the abnormal components; The visualization display module is used to visualize the collected performance data, abnormal data of the liquid-cooled server cabinet, and the data related to the abnormal processing of the abnormal processing module.