A Data Center AI Failure Probability Prevention Analysis Method

Abstract

This invention discloses a data center AI-based fault probability prevention analysis method, belonging to the field of fault analysis technology. The method includes: retrieving replacement records of target components to obtain the working duration of the replaced components; dividing the replaced components into several regions to obtain the aging level of each region and calculating the first aging degree of the replaced components; calculating the second aging degree of the replaced components based on the location of each calibration point to obtain the total aging degree of the replaced components; acquiring the component to be tested and several replaced components, obtaining the component weight of each replaced component; obtaining the fault warning duration of the component to be tested, intelligently judging the warning time of the component to be tested, and providing timely warning prompts. This invention, by extracting replacement records and comprehensively analyzing grayscale images, performance tests, and environmental factors, intelligently judges the warning time and provides warning prompts, improving the reliability of component replacement timing and helping to ensure the continuous and stable operation of the data center and reduce the risk of downtime failures.

Description

Technical Field

[0001] This invention relates to the field of fault analysis technology, specifically a data center AI fault probability prevention analysis method. Background Technology

[0002] As the information hub of the digital age, data centers bear the fundamental core position of network operation and business support. As the core carrier of data processing and application operation, main servers undertake key tasks such as computing and data storage. Whether it is the operation of enterprise core databases, the computing power output of cloud computing services, or big data analysis and AI model training, all rely on the main servers to provide stable and powerful hardware support. They are the core driving force for the data center to perform effectively. With the rapid development of information technology, the computing load of main servers is increasing, so the requirements for the heat dissipation speed of data centers are also increasing.

[0003] Currently, the replacement of critical target components on heat dissipation equipment is often done after a component fails or before it ages. There is a lack of reliable methods for calculating the degree of aging. Replacing components after a failure can lead to heat dissipation interruptions, causing business delays or even data center downtime losses. On the other hand, blindly replacing components before they age can lead to a waste of hardware resources. Since heat dissipation is a core function for maintaining efficient computing and data storage in data centers, intelligently assessing the degree of aging of target components and promptly issuing warnings and replacing defective components is crucial for ensuring the continuous and stable operation of data centers and reducing the risk of downtime. Summary of the Invention

[0004] The purpose of this invention is to provide a data center AI failure probability prevention analysis method to solve the problems raised in the prior art.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] A data center AI failure probability prevention analysis method includes the following steps:

[0007] Retrieve the historical replacement records of the target component installed on the heat dissipation equipment in the data center. The heat dissipation equipment is used to dissipate heat from component X. Extract the replaced components from the replacement records, as well as the installation time and replacement time of the replaced components. Based on the real-time temperature data of component X, obtain the working time of the replaced components.

[0008] Several reference components with uniform material degradation but different aging levels were obtained. Grayscale images of the surfaces of the reference components and the replacement components were taken respectively. The replacement components were divided into several regions, and the aging level of each region was obtained.

[0009] Based on the location of each area, the calibration points on the replacement parts are obtained, and the first degree of aging of the replacement parts is calculated.

[0010] Performance tests were conducted on the replacement parts and the reference parts. Based on the data changes at each calibration point during the test, the aging level of each calibration point was obtained. Based on the location of each calibration point, the second aging degree of the replacement parts was calculated.

[0011] The total aging degree of the replacement parts is obtained based on the preset weights of the first and second aging degrees.

[0012] The components currently installed on the heat dissipation equipment are taken as the components to be tested. Several replacement components are extracted, and the component weight of each replacement component is obtained based on the environmental factors of the environment in which the components to be tested and the replacement components are located.

[0013] By combining the working time of replacing components, the total degree of aging, and the component weight, the fault warning time of the component to be tested is obtained;

[0014] The system intelligently determines the warning time for the component to be tested and promptly alerts relevant personnel when the warning time is reached.

[0015] Preferably, obtaining the working time of the replacement component includes the following steps: extracting the replacement component B corresponding to a certain replacement record, and taking the time period between the installation time and the replacement time of the replacement component B as the application time; obtaining the real-time temperature data of component X during the application time, wherein when the temperature of component X exceeds Tem, the heat dissipation device will perform heat dissipation work, taking the moment when the temperature of component X exceeds Tem during the application time as the target moment, and collecting consecutive target moments to obtain the target time period. If there is a time period between two adjacent target time periods whose duration is less than a preset time period threshold, then the time period between them is also taken as the target time period; adding the durations of all target time periods to obtain the working time of the replacement component B.

[0016] Preferably, the aging level of each region is obtained by the following steps:

[0017] The aging level of each reference component is obtained and normalized. The higher the aging level, the more severe the aging of the reference component. The reference component with an aging level of 0 is regarded as a brand new component. The surface grayscale image of each reference component is captured, the grayscale value of each pixel in the surface grayscale image is obtained, several grayscale ranges are obtained, the grayscale values ​​are classified into the corresponding grayscale ranges, the pixel ratio of each grayscale range is obtained, and a grayscale histogram of each reference component is established.

[0018] Divide a replacement component B into several regions, capture a grayscale image of the surface of each region, obtain the grayscale value of each pixel in the surface grayscale image, and construct a grayscale histogram for each region; obtain the grayscale histogram H of a certain region Q. Q And the grayscale histogram H0 of the new component, calculate the grayscale histogram H. QThe histogram similarity between region Q and grayscale histogram H0 is calculated. If the histogram similarity is greater than a preset similarity threshold, the aging level of region Q is set to 0; if it is not greater than the preset similarity threshold, the grayscale histogram H0 is calculated. Q The histogram similarity between the grayscale histograms of each reference component and the aging level of the reference component with the largest histogram similarity is taken as the aging level of region Q.

[0019] In this scheme, the greater the histogram similarity, the stronger the grayscale histogram H. Q The more similar the region is to the grayscale histogram H0, the aging level of the region can be calculated based on the histogram similarity.

[0020] Preferably, calculating the first degree of aging of the replacement component includes the following steps:

[0021] Establish a 3D model of the heat dissipation device, obtain several coordinate points on a certain area Q, calculate the center coordinates between the coordinate points, and use the center coordinates as the calibration point P. Q Obtain the position of component X, and set the calibration point P. Q The distance between component X and the target distance of region Q is used as the target distance of region Q. According to the condition that the smaller the target distance, the greater the region weight, the region weight of each region is obtained, and the sum of all region weights is 1. Based on the region weight and aging level of each region, the first aging degree of replacement component B is obtained.

[0022] Since the thermal conductivity is more important for areas closer to component X, this solution should assign higher weights to areas closer to component X and lower weights to areas farther away. Then, based on the area weights and aging levels, the first aging degree of replacement component B is calculated.

[0023] Preferably, obtaining the aging level for each calibration point includes the following steps:

[0024] Obtain all calibration points on replacement component B, heat the parts of replacement component B and reference component that are in contact with component X, complete the performance test of replacement component B and reference component, and monitor the temperature of different calibration points in real time during the heating process.

[0025] Calculate the temperature change slope K0 on the brand-new component and the temperature change slope K on the replaced component B during the heating process at a certain calibration point P. B If the slope K0 and the slope K B If the absolute value of the difference is less than a preset threshold, the aging level of calibration point P is set to 0; if it is not less than the preset threshold, the slope of the temperature change on each reference component during the heating process of calibration point P is obtained, and the slope K is used as the aging level. BThe aging level of the reference component with the smallest absolute value of the difference between the two values ​​is taken as the aging level of the calibration point P.

[0026] Preferably, calculating the second aging degree of the replacement component includes the following steps: calculating the distance between each calibration point and component X, and obtaining the point weight of each calibration point according to the condition that the smaller the distance, the greater the point weight, wherein the sum of all point weights is 1; and obtaining the second aging degree of the replacement component B based on the point weight of each calibration point and the aging level.

[0027] Preferably, obtaining the component weight of each replacement component includes the following steps: acquiring environmental factors that affect the rate of component aging; pre-deploying sensors in the data center to monitor the environmental factor values ​​at the location of the component to be tested in real time, obtaining the average environmental factor value V0; extracting several replacement components; monitoring the environmental factor values ​​at the location of the replacement components in real time, obtaining the average environmental factor value corresponding to each replacement component; and obtaining the component weight of each replacement component according to the condition that the closer the component is to the average environmental factor value V0, the greater its component weight.

[0028] Preferably, obtaining the warning duration for the component to be detected includes the following steps:

[0029] Obtain the working time, total aging degree, and component weight for each replaced component. Pre-set the aging degree threshold for the component to be tested, based on the formula: The fault warning duration of the component under test is obtained, where D is the fault warning duration, Y is the aging threshold, M is the number of components to be replaced, and W... m Let D be the component weight of the m-th replaced component. m Y is the working time for the m-th component replacement. m The total aging degree of the m-th replaced component.

[0030] Preferably, an early warning is issued to relevant personnel, including the following steps: obtaining the installation time of the component to be tested; when the working time of the component to be tested reaches the fault warning time, an early warning is issued, and the staff is promptly instructed to inspect and repair the component to be tested.

[0031] This invention extracts replacement records to obtain replacement parts and baseline parts, and analyzes them by combining grayscale images, performance tests and environmental factors to provide a reliable method for calculating aging degree. It intelligently judges the early warning time and provides timely warning prompts, which helps to improve the reliability of component replacement timing, ensure the continuous and stable operation of data centers and reduce the risk of downtime failures.

[0032] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention provides a data center AI fault probability prevention analysis method, including: retrieving the replacement records of target components to obtain the working duration of the replaced components; dividing the replaced components into several regions to obtain the aging level of each region and calculating the first aging degree of the replaced components; calculating the second aging degree of the replaced components based on the location of each calibration point to obtain the total aging degree of the replaced components; acquiring the component to be tested and several replaced components to obtain the component weight of each replaced component; obtaining the fault warning duration of the component to be tested, intelligently judging the warning time of the component to be tested, and providing timely warning prompts. This invention, by extracting replacement records to obtain replaced components and baseline components, and comprehensively analyzing grayscale images, performance tests, and environmental factors, provides a reliable method for calculating aging degree, intelligently judging the warning time, and providing timely warning prompts, which helps to improve the reliability of component replacement timing, ensure the continuous and stable operation of the data center, and reduce the risk of downtime failures. Attached Figure Description

[0033] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a flowchart illustrating a data center AI fault probability prevention analysis method according to the present invention. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] Example: Figure 1 As shown, this invention provides a technical solution for a data center AI fault probability prevention analysis method, including the following steps:

[0037] (1) Retrieve the historical replacement records of the target component installed on the heat dissipation equipment in the data center. The heat dissipation equipment is used to dissipate heat from component X. Extract the replacement parts in the replacement records, as well as the installation time and replacement time of the replacement parts. Based on the real-time temperature data of component X, obtain the working time of the replacement parts.

[0038] Extract the replacement component B corresponding to a certain replacement record, and use the time period between the installation time and replacement time of the replacement component B as the application time; obtain the real-time temperature data of component X during the application time, wherein when the temperature of component X exceeds Tem, the heat dissipation device will perform heat dissipation work, and use the moment when the temperature of component X exceeds Tem during the application time as the target moment, and collect consecutive target moments to obtain the target time period. If there is a time period between two adjacent target time periods whose duration is less than a preset duration threshold, then the time period between them is also used as the target time period; add up the durations of all target time periods to obtain the working time of replacement component B.

[0039] In this embodiment, the target component on the heat dissipation device is a copper pipe, and component X is a core computing device in the data center, such as a CPU. When the temperature of component X in the data center is too low, the heat dissipation device will not operate. Therefore, in this solution, the target time period is the time period during which the heat dissipation device operates when the temperature of component X is too high. Using the target time period for analysis can effectively improve the reliability of the early warning time calculation in this solution.

[0040] (2) Obtain several reference parts with uniform material deterioration but different aging levels, take grayscale images of the surface of the reference parts and the replacement parts respectively, divide the replacement parts into several regions, and obtain the aging level of each region.

[0041] The aging level of each reference component is obtained and normalized. The higher the aging level, the more severe the aging of the reference component. The reference component with an aging level of 0 is regarded as a brand new component. The surface grayscale image of each reference component is captured, the grayscale value of each pixel in the surface grayscale image is obtained, several grayscale ranges are obtained, the grayscale values ​​are classified into the corresponding grayscale ranges, the pixel ratio of each grayscale range is obtained, and a grayscale histogram of each reference component is established.

[0042] Since the target component is a copper pipe, in this embodiment, the material deteriorates to copper rust, with a grayscale value range of 0-255. The closer the value is to 255, the whiter it appears. Normal copper pipes have higher grayscale values, typically around 200, while copper pipes with copper rust have lower grayscale values, typically around 140. Therefore, this solution uses grayscale values ​​for analysis to determine the aging condition of the replacement component. In this embodiment, the grayscale range is divided into 8 levels: 0-31, 32-63, and 224-255. The grayscale values ​​are then assigned to their respective ranges to obtain the pixel ratio for each range, thereby establishing a grayscale histogram.

[0043] Divide a replacement component B into several regions, capture a grayscale image of the surface of each region, obtain the grayscale value of each pixel in the surface grayscale image, and construct a grayscale histogram for each region; obtain the grayscale histogram H of a certain region Q. QAnd the grayscale histogram H0 of the new component, calculate the grayscale histogram H. Q The histogram similarity between region Q and grayscale histogram H0 is calculated. If the histogram similarity is greater than a preset similarity threshold, the aging level of region Q is set to 0; if it is not greater than the preset similarity threshold, the grayscale histogram H0 is calculated. Q The histogram similarity between the grayscale histograms of each reference component and the aging level of the reference component with the largest histogram similarity is taken as the aging level of region Q.

[0044] In this scheme, the greater the histogram similarity, the stronger the grayscale histogram H. Q The more similar the region is to the grayscale histogram H0, the aging level of the region can be calculated based on the histogram similarity.

[0045] (3) Based on the location of each area, obtain the calibration point on the replacement part and calculate the first degree of aging of the replacement part.

[0046] Establish a 3D model of the heat dissipation device, obtain several coordinate points on a certain area Q, calculate the center coordinates between the coordinate points, and use the center coordinates as the calibration point P. Q Obtain the position of component X, and set the calibration point P. Q The distance between component X and the target distance of region Q is used as the target distance of region Q. According to the condition that the smaller the target distance, the greater the region weight, the region weight of each region is obtained, and the sum of all region weights is 1. Based on the region weight and aging level of each region, the first aging degree of replacement component B is obtained.

[0047] Since the thermal conductivity is more important for areas closer to component X, this solution should assign higher weights to areas closer to component X and lower weights to areas farther away. Then, based on the area weights and aging levels, the first aging degree of replacement component B is calculated.

[0048] (4) Perform performance tests on the replacement parts and the reference parts, and obtain the aging level of each calibration point based on the data changes at each calibration point during the test.

[0049] Obtain all calibration points on replacement component B, heat the parts of replacement component B and reference component that are in contact with component X, complete the performance test of replacement component B and reference component, and monitor the temperature of different calibration points in real time during the heating process.

[0050] By establishing a function of temperature change over time, the slope of the temperature change can be obtained; during the heating process at a certain calibration point P, the slope of the temperature change K0 on the brand-new component and the slope of the temperature change K on the replaced component B can be calculated. B If the slope K0 and the slope K BIf the absolute value of the difference is less than a preset threshold, the aging level of calibration point P is set to 0; if it is not less than the preset threshold, the slope of the temperature change on each reference component during the heating process of calibration point P is obtained, and the slope K is used as the aging level. B The aging level of the reference component with the smallest absolute value of the difference between the two values ​​is taken as the aging level of the calibration point P.

[0051] (5) Calculate the second degree of aging of the replacement parts based on the location of each calibration point.

[0052] Calculate the distance between each calibration point and component X. Based on the condition that the smaller the distance, the greater the point weight, obtain the point weight of each calibration point. The sum of all point weights is 1. Based on the point weights of each calibration point and the aging level, obtain the second aging degree of the replacement component B.

[0053] (6) Based on the preset weights of the first aging degree and the second aging degree, the total aging degree of the replacement parts is obtained.

[0054] In this embodiment, the first aging degree is set as N1, the second aging degree is set as N2, the weight of the first aging degree is S1, and the weight of the second aging degree is S2. The total aging degree of the replacement component is N = N1*S1 + N2*S2.

[0055] (7) Take the components installed on the current heat dissipation equipment as the components to be tested, extract several replacement components, and obtain the component weight of each replacement component based on the environmental factors of the environment in which the components to be tested and the replacement components are located.

[0056] Environmental factors affecting the aging rate of components are identified. Sensors are pre-deployed in the data center to monitor the environmental factor values ​​at the location of the component under test in real time, obtaining an average environmental factor value V0. Several replacement components are then selected, and the environmental factor values ​​at their locations are monitored in real time to obtain an average environmental factor value for each replacement component. The component weight is determined by the condition that the closer the component's value is to the average environmental factor value V0, the higher its weight. In this embodiment, the environmental factor is humidity, which is one of the key factors leading to copper rust. When the humidity in the air between the component under test and a replacement component is similar, their aging rates are more similar. Therefore, replacement components with similar environmental factors should be assigned higher component weights.

[0057] (8) The fault warning time of the component to be tested is obtained by combining the working time of the replaced component, the total aging degree and the component weight.

[0058] Obtain the working time, total aging degree, and component weight for each replaced component. Pre-set the aging degree threshold for the component to be tested, based on the formula: The fault warning duration of the component under test is obtained, where D is the fault warning duration, Y is the aging threshold, M is the number of components to be replaced, and W... m Let D be the component weight of the m-th replaced component. m Y is the working time for the m-th component replacement. m The total aging degree of the m-th replaced component.

[0059] The following example is given: the number of parts to be replaced M is 2, the working time of the first replaced part is 10, the total aging degree is 1, and the part weight is 0.7; the working time of the first replaced part is 20, the total aging degree is 0.8, the part weight is 0.3, and the aging degree threshold Y is 0.9. Substituting these values ​​into the formula, we can obtain the fault warning time as 13.05. Therefore, based on the method of this scheme, we can obtain the fault warning time of the part to be tested and issue timely warnings.

[0060] (9) Intelligently judge the warning time of the component to be tested, and promptly issue a warning to relevant personnel when the warning time is reached.

[0061] The system obtains the installation time of the component under test. When the working time of the component under test reaches the fault warning time, an early warning is issued, prompting staff to promptly inspect and repair the component under test.

[0062] This invention extracts replacement records to obtain replacement parts and baseline parts, and analyzes them by combining grayscale images, performance tests and environmental factors to provide a reliable method for calculating aging degree. It intelligently judges the early warning time and provides timely warning prompts, which helps to improve the reliability of component replacement timing, ensure the continuous and stable operation of data centers and reduce the risk of downtime failures.

[0063] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0064] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0065] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A data center AI failure probability prevention analysis method, characterized by, The method comprises the following steps: obtaining the historical replacement records of the target component installed on the heat dissipation device of the data center, the heat dissipation device being used for dissipating heat of the component X, extracting the replaced component, the installation time and the replacement time of the replaced component, and obtaining the working time of the replaced component according to the real-time temperature data of the component X; obtaining a plurality of reference components with uniform material degradation degree but different aging levels, respectively shooting the surface gray scale images of the reference components and the replaced component, dividing the replaced component into a plurality of regions, and obtaining the aging level of each region; obtaining the calibration points on the replaced component according to the positions of each region, and calculating the first aging degree of the replaced component; performing performance test on the replaced component and the reference component, obtaining the aging level of each calibration point according to the data change of each calibration point during the test; calculating the second aging degree of the replaced component according to the positions of each calibration point; obtaining the total aging degree of the replaced component according to the preset weight values of the first aging degree and the second aging degree; taking the component installed on the current heat dissipation device as a to-be-detected component, extracting a plurality of replaced components, and obtaining the component weight value of each replaced component according to the environmental elements of the environment in which the to-be-detected component and the replaced component are located; comprehensively obtaining the fault warning time of the to-be-detected component by combining the working time, the total aging degree and the component weight value of the replaced component; intelligently judging the warning time of the to-be-detected component, and timely warning and prompting relevant personnel when the warning time is reached; The method comprises the following steps: establishing a three-dimensional model of a heat dissipation device, obtaining a plurality of coordinate points on a certain area Q, calculating the central coordinates between the coordinate points, taking the central coordinates as calibration points P Q , obtaining the position of a component X, taking the distance between the calibration points P Q and the component X as the target distance of the area Q; obtaining the area weight of each area according to the condition that the smaller the target distance is, the greater the area weight is, wherein the sum of all area weights is 1; obtaining the first aging degree of the replacement component B according to the area weight and the aging grade of each area; obtaining the aging level of each calibration point, comprising: obtaining all calibration points on the replaced component B, heating the parts of the replaced component B and the reference component that are in contact with the component X, completing the performance test on the replaced component B and the reference component, and monitoring the temperature of different calibration points in real time during the heating process; calculating a temperature change slope K0 of the new part and a temperature change slope K of the replacement part B during heating of the calibration point P B , if the absolute value of the difference between the slope K0 and the slope K B is less than a preset numerical threshold, then the aging level of the calibration point P is 0; if the absolute value of the difference is not less than the preset numerical threshold, then the temperature change slope of each reference part during heating of the calibration point P is obtained, and the aging level of the reference part corresponding to the smallest absolute value of the difference between the slope K B and the slope K is taken as the aging level of the calibration point P. calculating the second aging degree of the replaced component, comprising: calculating the distance between each calibration point and the component X, obtaining the point weight of each calibration point according to the condition that the smaller the distance is, the greater the point weight is, wherein the sum of all point weights is 1; obtaining the second aging degree of the replaced component B according to the point weight and the aging level of each calibration point; obtaining the component weight value of each replaced component, comprising: obtaining the environmental elements that affect the aging degree of the component, pre-deploying sensors in the data center to monitor the environmental element values at the positions of the to-be-detected components in real time, obtaining the average environmental element value V0, extracting a plurality of replaced components, monitoring the environmental element values at the positions of the replaced components in real time, obtaining the average environmental element value corresponding to each replaced component, and obtaining the component weight of each replaced component according to the condition that the closer the average environmental element value is to the average environmental element value V0, the greater the component weight is; obtaining the warning time of the to-be-detected component, comprising: The working time length, total aging degree and component weight value of each replacement component are acquired, the aging degree threshold of the component to be detected is set in advance, and the fault early warning time length of the component to be detected is obtained according to the formula: , wherein D is the fault early warning time length, Y is the aging degree threshold, M is the number of replacement components, W m is the component weight value of the mth replacement component, D m is the working time length of the mth replacement component, Y m is the total aging degree of the mth replacement component.

2. The data center AI failure probability prevention analysis method of claim 1, wherein, The work duration of the replacement component is obtained by the following steps: extracting the replacement component B corresponding to a replacement record, taking the time period between the installation time and the replacement time of the replacement component B as the application time; obtaining the real-time temperature data of the component X in the application time, wherein when the temperature of the component X exceeds Tem, the heat dissipation device performs heat dissipation work, taking the time when the temperature of the component X exceeds Tem in the application time as the target time, collecting the target time periods of the continuous target times, and if the time duration of the time period between two adjacent target time periods is less than the preset time duration threshold, the time period is also taken as the target time period; adding the time durations of all the target time periods to obtain the work duration of the replacement component B.

3. The data center AI failure probability prevention analysis method of claim 1, wherein, The aging grade of each region is obtained by the following steps: The aging grade of each reference component is obtained and normalized, and the larger the aging grade is, the more serious the aging degree of the reference component is. The reference component with an aging grade of 0 is taken as a brand new component; a surface gray image of each reference component is shot, the gray value of each pixel in the surface gray image is obtained, a plurality of gray ranges are obtained, the gray values are classified into the corresponding gray ranges, the pixel ratio of each gray range is obtained, and the gray histogram of each reference component is established; The replacement part B is divided into several areas, the surface gray scale image of each area is photographed, the gray scale value of each pixel in the surface gray scale image is obtained, and the gray scale histogram of each area is established; the gray scale histogram H Q of a certain area Q and the gray scale histogram H0 of a new part are obtained, the histogram similarity between the gray scale histogram H Q and the gray scale histogram H0 is calculated, if the histogram similarity is greater than a preset similarity threshold, the aging grade of the area Q is 0; if the histogram similarity is not greater than the preset similarity threshold, the histogram similarity between the gray scale histogram H Q and the gray scale histogram of each reference part is calculated, and the aging grade of the reference part corresponding to the maximum histogram similarity is taken as the aging grade of the area Q.

4. The data center AI failure probability prevention analysis method of claim 1, wherein, The related personnel are warned and prompted by the following steps: the installation time of the to-be-detected component is obtained, a warning prompt is issued when the work duration of the to-be-detected component reaches the fault warning duration, and the worker is timely prompted to repair and process the to-be-detected component.

Images