Machine room operation and maintenance fault analysis method using full life cycle monitoring

Abstract

The invention relates to the technical field of data processing, in particular to a machine room operation and maintenance fault analysis method utilizing full life cycle monitoring, which comprises the following steps: dividing a machine room into a plurality of monitoring areas, and clustering load stable time periods based on load differences among different load stable time periods of equipment in the monitoring areas to obtain a plurality of class clusters; according to the temperature ranges of all the load stabilization time periods of the class clusters and the temperature stability of the load stabilization time periods, obtaining the correlation between the loads and the temperatures of the class clusters, screening out abnormal class clusters according to the correlation, fusing the correlation between the abnormal class clusters of the equipment and the number of the load stabilization time periods, and combining the number of the abnormal class clusters. And finally, the temperature safety threshold value of the monitoring area is obtained by combining the temperature uniformity of the monitoring area, so that the applicability of the temperature safety threshold value is improved, the problems of low temperature safety monitoring reliability and low safety caused by the improper temperature safety threshold value are avoided, and the operation and maintenance reliability of a machine room is improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically to a method for analyzing data center operation and maintenance faults using full lifecycle monitoring. Background Technology

[0002] Data centers are complex operation and maintenance scenarios comprised of equipment clusters, environmental systems, and supporting facilities. During data center operation and maintenance fault analysis, it's crucial to ensure the data center operates within a suitable temperature range. When the temperature exceeds the safe threshold, alarm signals need to be output promptly to facilitate the initiation of relevant corrective measures. However, the operating status and load capacity of equipment located in different areas of the data center vary, and the aging levels of different devices may also differ. For example, areas with more aged equipment typically have lower temperature resistance and are more sensitive to temperature fluctuations. Therefore, using a fixed and uniform safe temperature threshold for all areas of the data center may not be applicable to the equipment status in all areas, potentially leading to delayed or premature alarms, failing to accurately reflect actual risks, and impacting operational reliability. Summary of the Invention

[0003] To address the technical problem that using fixed and uniform temperature safety thresholds in all areas of a data center during existing data center operation and maintenance fault analysis processes can negatively impact operational reliability, this invention aims to provide a data center operation and maintenance fault analysis method utilizing full lifecycle monitoring. The specific technical solution adopted is as follows: This invention provides a data center operation and maintenance fault analysis method utilizing full lifecycle monitoring, comprising: The computer room was divided into several monitoring areas; Based on the load difference between any two load-stable periods of the equipment within the monitoring area, the load-stable periods are clustered to obtain several clusters. Based on the temperature range of all load stabilization periods of the cluster, and the temperature stability during the load stabilization periods, the correlation between load and temperature of the cluster is obtained; The aging degree of the equipment is obtained by combining the correlation of each abnormal cluster of the fusion equipment and the number of load stable periods contained therein, together with the number of abnormal clusters of the equipment; the abnormal clusters are obtained by filtering the correlation. The temperature safety threshold of the monitoring area is obtained by considering the aging degree of each device and the temperature uniformity within the monitoring area.

[0004] In one exemplary embodiment, dividing the computer room into several monitoring areas includes: The temperature of each monitoring point in the computer room within a historical time period is divided into several temperature sequence segments. The initial temperature uniformity of the monitoring points is obtained based on the difference between the temperature of each temperature sequence segment and the preset standard temperature. Determine the load intensity characteristics of each device within the influence range of the monitoring point at the same time as the temperature sequence segment to obtain the high load period; Determine the extent to which the load intensity characteristics of each device during high-load periods affect the initial temperature uniformity of the monitoring points; Based on the degree of influence and the spatial distance between each device and the monitoring point, the initial temperature uniformity is corrected to obtain the temperature uniformity of the monitoring point; Based on the difference in temperature uniformity between any two monitoring points and the spatial distance between them, the monitoring points are clustered to obtain several monitoring areas.

[0005] In an exemplary embodiment, the process of obtaining the initial temperature uniformity includes: By integrating the differences between the characteristic temperatures of each temperature sequence segment at the monitoring point and the preset standard temperature, the overall temperature deviation of the monitoring point can be obtained. The initial temperature uniformity of the monitoring points is obtained based on the overall temperature deviation and the length ratio of the longest temperature sequence segment at each monitoring point. The initial temperature uniformity is inversely correlated with the overall temperature deviation and positively correlated with the length ratio.

[0006] In an exemplary embodiment, the process of obtaining the load strength characteristics includes: Obtain the minimum load value of the device within the same time period as the temperature sequence segment, as well as the load change trend; The load intensity characteristics are obtained from the minimum load value and the load change trend.

[0007] In an exemplary embodiment, the process of obtaining the degree of influence includes: Based on the load intensity characteristics of the equipment during each high-load period and the correlation between load and temperature, an index of the influence of load on temperature during each high-load period is obtained; the index of the influence of load on temperature is positively correlated with the load intensity characteristics and the correlation between load and temperature. The influence index of temperature on load during each high-load period of the fusion device is combined with the number of high-load periods of the device to obtain the degree of influence; the degree of influence is positively correlated with both the influence index of temperature on load and the number of high-load periods.

[0008] In one exemplary embodiment, the initial temperature uniformity correction process includes: Based on the degree of influence of each device within the influence range of the monitoring point, and the spatial distance between each device and the monitoring point, a correction index for each device is obtained; the correction index is inversely correlated with the degree of influence and positively correlated with the spatial distance between the device and the monitoring point. The initial temperature uniformity of the monitoring points is corrected by the correction coefficients of the monitoring points to obtain the temperature uniformity of the monitoring points; all correction coefficients are obtained by fusing the correction indices of all equipment within the influence range of the monitoring points.

[0009] In an exemplary embodiment, the process of obtaining the temperature stability during the load stabilization period includes: The number of key temperature data points in the operating temperature sequence segment is obtained; the key temperature data points represent temperature data points that indicate temperature fluctuations; the operating temperature sequence segment is the operating temperature data of the equipment during the same period as the load stability period. Obtain the maximum rate of temperature change at the key temperature data points. The temperature stability during the load stabilization period is obtained based on the number of key temperature data points and the maximum temperature change rate; the temperature stability is negatively correlated with both the number of key temperature data points and the maximum temperature change rate.

[0010] In an exemplary embodiment, the temperature range for all load-stable periods of the cluster is the temperature difference between the maximum and minimum device operating temperatures during all load-stable periods in the cluster. The process of obtaining the correlation between the load and temperature of the aforementioned cluster includes: Determine the overall temperature stability level for all load stabilization periods within the cluster; Based on the temperature difference and the overall level of temperature stability, the correlation between the load and temperature of the cluster is obtained; the correlation is positively correlated with the overall level of temperature stability and negatively correlated with the temperature difference.

[0011] In one exemplary embodiment, the process of obtaining the aging level of the device includes: By integrating the correlations of each abnormal cluster and the number of stable load periods it contains, an overall impact index on equipment aging is obtained; the overall impact index is inversely correlated with the correlations and positively correlated with the number of stable load periods it contains. The aging degree of the equipment is obtained based on the number of abnormal clusters and the overall impact index. The aging degree is positively correlated with both the number of abnormal clusters and the overall impact index.

[0012] In an exemplary embodiment, the process of obtaining the temperature safety threshold of the monitoring area includes: The temperature regulation coefficient of the monitoring area is obtained based on the maximum aging degree and the overall temperature uniformity level. The temperature regulation coefficient is inversely correlated with the maximum aging degree and positively correlated with the overall temperature uniformity level. The maximum aging degree is the maximum value among the aging degrees of each device in the monitoring area. The overall temperature uniformity level is obtained by integrating the temperature uniformity of each monitoring point in the monitoring area. The temperature safety threshold of the monitoring area is obtained by adjusting the preset initial temperature safety threshold using the temperature adjustment coefficient of the monitoring area.

[0013] This invention offers the following advantages: It divides the computer room into several monitoring zones, analyzing each zone independently. Compared to traditional methods that analyze all zones indiscriminately, this improves the reliability and security of temperature safety monitoring, thereby enhancing the reliability of computer room operations. Since there is a correlation between equipment load and temperature, and this correlation determines the degree of equipment aging, analyzing the correlation between load and temperature for each device within the monitoring zone yields the degree of aging for each device, thus determining the temperature safety threshold for that zone. Given the differences in equipment within the monitoring zone, and considering the uneven aging of equipment and temperature distribution across different monitoring zones within the computer room, the temperature safety thresholds for different zones will not be identical. This improves the applicability of the temperature safety thresholds, avoiding low reliability and security due to inappropriate thresholds, thereby enhancing the reliability of computer room operations. Attached Figure Description

[0014] Figure 1 This is a flowchart of a data center operation and maintenance fault analysis method using full lifecycle monitoring, provided by an embodiment of the present invention; Figure 2 This is a flowchart illustrating the division of monitoring areas provided in one embodiment of the present invention; Figure 3 This is a flowchart illustrating the process of obtaining initial temperature uniformity according to an embodiment of the present invention; Figure 4 This is a flowchart illustrating the acquisition of load strength characteristics according to an embodiment of the present invention; Figure 5 This is a flowchart illustrating the process of obtaining the degree of influence according to one embodiment of the present invention; Figure 6 This is a flowchart illustrating the correction of initial temperature uniformity provided in one embodiment of the present invention; Figure 7 This is a flowchart illustrating the process of obtaining temperature stability according to an embodiment of the present invention; Figure 8 This is a flowchart illustrating the process of obtaining relevant relationships according to an embodiment of the present invention; Figure 9 This is a flowchart illustrating the process of obtaining the aging degree of a device according to an embodiment of the present invention; Figure 10 This is a flowchart illustrating the process of obtaining the temperature safety threshold according to an embodiment of the present invention. Detailed Implementation

[0015] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the specific implementation methods, structures, features, and effects of the present invention are described in detail below with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0016] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. All data and information collected in this application have been obtained with full consent.

[0017] This embodiment provides a data center operation and maintenance fault analysis method using full lifecycle monitoring, applicable to data center operation and maintenance fault analysis. The data centers to which this method is applicable typically house several different types of equipment, such as servers, storage devices, and network switches. These devices are assigned different tasks; some devices may operate at full load, while others operate at low load.

[0018] Multiple monitoring points are set up within the computer room. The number of monitoring points and the distance between adjacent monitoring points are determined by the size of the computer room. In this embodiment, the monitoring points are evenly distributed within the computer room. A temperature sensor (such as a digital thermometer, infrared temperature sensor, etc.) is installed at each monitoring point to collect the ambient temperature of the computer room at each monitoring point. In an exemplary embodiment, a three-dimensional coordinate system is constructed using the length, width, and height of the computer room. The projection position of each monitoring point on the horizontal plane (i.e., the xy plane) of the three-dimensional coordinate system must be unique, but there is no limitation on the setting height of each monitoring point in the computer room. The setting height of each monitoring point can be the same or different, and can be determined by the equipment at the corresponding location. This embodiment takes the example of all monitoring points being set at the same height.

[0019] Each device in the computer room also has a built-in temperature sensor, usually located in the core components, to collect the operating temperature of each device.

[0020] This embodiment provides a data processing device for analyzing data center operation and maintenance faults using full lifecycle monitoring. The data processing device is connected to temperature sensors at each monitoring point and within each device to acquire the temperature at each monitoring point and the operating temperature of each device. Simultaneously, the data processing device also acquires the load values ​​of each device.

[0021] It should be understood that this embodiment sets a historical time period for analyzing operational faults in the data center by acquiring data within that period. The length of the historical time period is set according to actual monitoring needs, such as 10 minutes. The sampling frequency of each temperature sensor is set according to actual needs, such as 0.5 seconds per sampling. Therefore, for any given temperature sensor, the temperature at each moment within the historical time period is acquired, thus forming the temperature sequence for that sensor. Furthermore, for any given device, the load value at each moment within the historical time period is acquired, thus forming the load sequence for that device.

[0022] In this embodiment, the temperature of the monitoring points can also be normalized to eliminate dimensions and facilitate data processing. In one exemplary embodiment, the maximum and minimum values ​​of all temperature values ​​from all monitoring points within a historical time period are obtained, and the maximum-minimum-value normalization method is used to normalize each temperature value within the historical time period. Unless otherwise specified, all temperatures mentioned below are normalized temperatures. Similarly, the operating temperatures of all devices within a historical time period are also normalized using the maximum-minimum-value normalization method, that is, the maximum and minimum values ​​of all operating temperature values ​​from all devices within a historical time period are obtained, and the maximum-minimum-value normalization method is used to normalize the operating temperatures of each device within the historical time period.

[0023] like Figure 1 As shown in the figure, the data center operation and maintenance fault analysis method using full lifecycle monitoring provided in this embodiment includes the following steps: Step S1: Divide the computer room into several monitoring areas; Step S2: Based on the load difference between any two load-stable periods of the equipment within the monitoring area, cluster the load-stable periods to obtain several clusters; Step S3: Based on the temperature range of all load stabilization periods for the cluster, and the temperature stability during the load stabilization periods, obtain the correlation between the cluster's load and temperature; Step S4: Combine the correlation between the various abnormal clusters of the equipment and the number of stable load periods contained therein, and combine them with the number of abnormal clusters of the equipment to obtain the aging degree of the equipment; Step S5: Obtain the temperature safety threshold of the monitoring area based on the aging degree of each device and the temperature uniformity within the monitoring area.

[0024] The following detailed explanation of each step, in conjunction with the accompanying drawings, is provided.

[0025] Step S1: Divide the computer room into several monitoring areas.

[0026] The varying loads and cooling capacities of different devices within a data center lead to uneven temperature distribution. For example, the temperature near the servers may be higher, while the temperature near the air conditioning vents may be lower. This uneven temperature distribution can cause localized overheating or underheating of equipment. Therefore, it is necessary to divide the data center into several monitoring zones and determine corresponding temperature safety thresholds for each zone. This prevents temperature warnings from being issued prematurely or delayed due to excessively high or low temperatures in certain monitoring areas, thereby reducing the reliability of data center operations.

[0027] In one exemplary embodiment, such as Figure 2 As shown below, a specific implementation process for dividing the computer room into several monitoring areas is given: Step S11: Divide the temperature of each monitoring point in the computer room into several temperature sequence segments within the historical time period.

[0028] For any given monitoring point, the temperature of that point over a historical time period is acquired, including the temperature at each moment within that period. Essentially, this is the temperature sequence of that monitoring point. This temperature sequence is then divided into several temperature sequence segments. In one exemplary embodiment, the principle for segmenting these segments is to group consecutive temperatures that are relatively close together into one segment, while maintaining a large temperature difference between adjacent segments. Accordingly, this embodiment uses the APCA (Adaptive Piecewise Constant Approximation) algorithm to adaptively segment the temperature sequence of the monitoring point over the historical time period, resulting in several temperature sequence segments, and obtaining the APCA approximate temperature value for each segment. The APCA approximate temperature value of a temperature sequence segment represents the overall characteristics of that segment, and the differences in APCA approximate temperature values ​​between different segments are significant. Furthermore, the greater the difference in APCA approximate temperature values ​​between different temperature sequence segments, the more drastic the temperature data fluctuations.

[0029] Step S12: Based on the difference between the temperature of each temperature sequence segment and the preset standard temperature, the initial temperature uniformity of the monitoring point is obtained.

[0030] The initial temperature uniformity of the monitoring point is obtained by comparing the temperature of each temperature sequence segment with the preset standard temperature. The greater the difference, the worse the initial temperature uniformity. The preset standard temperature of the monitoring point is set based on actual conditions; in this embodiment, the mode of the temperature sequence of the monitoring point is used as the preset standard temperature.

[0031] In one exemplary embodiment, such as Figure 3 As shown, the following is a specific process for obtaining the initial temperature uniformity: Step S121: Combine the differences between the characteristic temperatures of each temperature sequence segment of the monitoring point and the preset standard temperature to obtain the overall temperature deviation of the monitoring point.

[0032] The characteristic temperature of each temperature sequence segment at the monitoring point is determined. The characteristic temperature represents the overall temperature of the corresponding temperature sequence segment. Based on the APCA algorithm mentioned above, the characteristic temperature of each temperature sequence segment is specifically the APCA approximate temperature value of each temperature sequence segment.

[0033] To obtain the difference between the characteristic temperature of each temperature sequence segment of the monitoring point and the preset standard temperature, specifically: calculate the absolute value of the difference between the characteristic temperature of each temperature sequence segment of the monitoring point and the preset standard temperature, and use this as the difference between the characteristic temperature of each temperature sequence segment of the monitoring point and the preset standard temperature.

[0034] The method integrates the absolute values ​​of the differences between the characteristic temperatures of each temperature sequence segment at the monitoring point and the preset standard temperatures. Specifically, it calculates the average of these absolute values ​​and uses the result as the overall temperature deviation of the monitoring point. The larger the absolute value of the difference between the characteristic temperatures of each temperature sequence segment and the preset standard temperatures, the greater the deviation from the normal temperature, and the higher the overall temperature deviation of the monitoring point. This also indicates a more uneven temperature distribution at the monitoring point over historical periods and lower initial temperature uniformity. Therefore, initial temperature uniformity is inversely correlated with the overall temperature deviation.

[0035] Step S122: Based on the overall temperature deviation and the length ratio of the longest temperature sequence segment at the monitoring point, the initial temperature uniformity of the monitoring point is obtained.

[0036] The time lengths of each temperature sequence segment at the monitoring point are obtained, and the longest time length is selected. The temperature sequence segment corresponding to the longest time length is taken as the longest temperature sequence segment at the monitoring point. Then, the ratio of the time length of the longest temperature sequence segment at the monitoring point to the time length of the historical time period is calculated as the length proportion of the longest temperature sequence segment at the monitoring point. The larger the length proportion of the longest temperature sequence segment at the monitoring point, the more stable the temperature change at the monitoring point, and the higher the initial temperature uniformity of the monitoring point. Therefore, the initial temperature uniformity is positively correlated with the length proportion of the longest temperature sequence segment at the monitoring point. In an exemplary embodiment, a specific quantification method for the initial temperature uniformity of the monitoring point is given below: ; in, This indicates the initial temperature uniformity at the w-th monitoring point. This represents the percentage of the longest temperature sequence segment at the w-th monitoring point. This represents the overall temperature deviation at the w-th monitoring point. As another implementation method, if the temperature has not been normalized beforehand, then the formula above needs to be adjusted. Replace with the overall deviation from temperature The negative correlation normalization results, for example: , This represents an exponential function with the natural constant as its base.

[0037] Step S13: Determine the load intensity characteristics of each device within the influence range of the monitoring point at the same time as the temperature sequence segment to obtain the high load period.

[0038] In a computer room, equipment workload is the core "heat source" of the thermal environment. Equipment loads at different monitoring points may vary; for example, some monitoring points may bear heavier computing tasks, leading to higher local equipment temperatures. Therefore, equipment in high-load areas experiences concentrated heat dissipation, resulting in poor temperature uniformity. Conversely, equipment in other areas has lower loads, resulting in more stable heat dissipation and easier maintenance of regional temperature uniformity. Therefore, it is necessary to determine the load intensity characteristics of each device within the influence range of the monitoring points during the same period as the temperature sequence segment, and to identify the high-load periods.

[0039] First, for any given monitoring point, determine its influence range. The size of this influence range is determined based on the actual situation. As an example, a radius of 2 meters on the xy-plane centered on the monitoring point is defined as its influence range. This allows us to identify the devices located within the influence range of that monitoring point.

[0040] For any device within the influence range of the monitoring point, the load value of that device at each moment in the historical time period is obtained to obtain a load sequence. This embodiment can also normalize the load sequence for easier subsequent processing. In an exemplary embodiment, the maximum and minimum values ​​of the load values ​​of all devices in the computer room at all moments in the historical time period are obtained, and then the maximum and minimum value normalization method is used to normalize the load values ​​of each device in the computer room at each moment. Unless otherwise specified, the load values ​​mentioned below are all normalized load values.

[0041] The boundary points of each temperature sequence segment at the monitoring point are determined, and the load sequence of the equipment is segmented according to each boundary point to obtain the load sequence segment that is in the same period as the temperature sequence segment at the monitoring point.

[0042] Then, the load intensity characteristics of the load sequence segments of each temperature sequence segment at the monitoring point are obtained simultaneously, such as... Figure 4 As shown, the following is a specific process for obtaining load strength characteristics: Step S131: Obtain the minimum load value of the device in the same period as the temperature sequence segment, as well as the load change trend.

[0043] For ease of explanation, any device within the influence range of the w-th monitoring point is taken as the s-th device, and the i-th load sequence segment of the s-th device is taken as an example.

[0044] Since the i-th load sequence segment of the s-th device includes multiple load values ​​in time series, the minimum load value in the i-th load sequence segment of the s-th device is obtained. A larger minimum load value indicates a higher overall load level and a stronger load intensity characteristic in the i-th load sequence segment.

[0045] To obtain the load change trend of the ith load sequence segment of the s-th device, in an exemplary embodiment, the least squares method is used to perform curve fitting on the load value of the ith load sequence segment to obtain the fitted curve of the ith load sequence segment. Then, the slope of the tangent line of the load value at each moment in the fitted curve is obtained, and the average value of the tangent line slope at each moment of the ith load sequence segment is calculated as the overall slope of the ith load sequence segment. It should be understood that the overall slope may be greater than 0, less than 0, or equal to 0. When it is greater than 0, it indicates that the load value of the ith load sequence segment is on an increasing trend, and the larger the value, the more obvious the increasing trend, indicating a higher load intensity characteristic, such as a higher CPU utilization. When it is equal to 0, it indicates that the load value of the ith load sequence segment is generally in a stable state. When it is less than 0, it indicates that the load value of the ith load sequence segment is on a decreasing trend, and the smaller the value, the more obvious the decreasing trend, indicating a lower load intensity characteristic. Therefore, in terms of the overall slope value, the larger the value, the higher the load intensity characteristic of the ith load sequence segment.

[0046] Step S132: Obtain the load intensity characteristics from the minimum load value and the load change trend.

[0047] Based on the minimum load value and load change trend of the ith load sequence segment of the s-th device, the load intensity characteristics of the ith load sequence segment are obtained. A specific quantification method for the load intensity characteristics is given below: ; in, This represents the load intensity characteristic of the i-th load sequence segment of the s-th device. The slope of the i-th load sequence segment of the s-th device is represented by sigmoid, and sigmoid represents the sigmoid function. This represents the minimum load value in the i-th load sequence segment of the s-th device. Alternatively, if the load is not normalized beforehand, the above formula needs to first normalize the load. Normalize it, and then use it in the calculation.

[0048] The load intensity characteristics of each load sequence segment of the s-th device are obtained through the above method. Based on the load intensity characteristics, the high-load period of the s-th device is determined. In an exemplary embodiment, a preset threshold is used to compare with the load intensity characteristics of the load sequence segments, thereby determining the larger load intensity characteristics and thus the high-load period. The value range of the preset threshold is 0-1, and the specific value is set according to actual needs, such as 0.7.

[0049] The load intensity characteristics of each load sequence segment of the s-th device are compared with the preset threshold to determine the time period of the load sequence segment that is greater than the preset threshold. The time period of the load sequence segment that is greater than the preset threshold is taken as the high load time period of the s-th device.

[0050] Step S14: Determine the degree of influence of the load intensity characteristics of each device during the high-load period on the initial temperature uniformity of the monitoring points.

[0051] Different types of equipment generate significantly different loads and heat. For example, storage and computing devices may have higher loads and generate more heat, thus having a greater impact on the ambient temperature of the data center. Network equipment, on the other hand, may have lower loads and generate less heat, thus having a smaller impact on the ambient temperature of the data center. Therefore, it is necessary to determine the extent to which the load intensity characteristics of each device during high-load periods affect the initial temperature uniformity at the monitoring points.

[0052] In one exemplary embodiment, such as Figure 5 As shown, the following is a specific process for obtaining the degree of influence: Step S141: Based on the load intensity characteristics of the equipment during each high-load period and the correlation between load and temperature, obtain the load-related index of temperature during each high-load period.

[0053] For the s-th device, determine the load intensity characteristics of the s-th device during each high-load period. Obtain the temperature sequence segments of the same period during each high-load period of the s-th device, thereby obtaining the correlation between the load of the s-th device during each high-load period and the temperature of the w-th monitoring point during the same period.

[0054] Taking the nth high-load period of the s-th device as an example, the temperature sequence segment of the w-th monitoring point during the same period as the nth high-load period is defined as the nth temperature sequence segment. The correlation between the load of the s-th device during the nth high-load period and the temperature of the nth temperature sequence segment is obtained. Specifically, this correlation is the Pearson correlation coefficient. Since the Pearson correlation coefficient ranges from -1 to 1, it is normalized using the formula: (Pearson correlation coefficient + 1) / 2.

[0055] Based on the load intensity characteristics of the nth high-load period of the s-th device and the correlation between the load of the s-th device during the nth high-load period and the temperature of the nth temperature sequence segment, the load-related influence index of the temperature of the s-th device during the nth high-load period is obtained. The stronger the load intensity characteristics of the s-th device during the nth high-load period, the higher the correlation between the monitoring point of the s-th device and the heat generated by the s-th device, and the greater the load-related influence index of the temperature; the two are positively correlated. Similarly, the higher the correlation between load and temperature, the greater the load-related influence index of the temperature; the two are also positively correlated. In an exemplary embodiment, the product of the load intensity characteristics of the s-th device during the nth high-load period and the normalized Pearson correlation coefficient between the load of the s-th device during the nth high-load period and the temperature of the nth temperature sequence segment is used as the load-related influence index of the temperature of the s-th device during the nth high-load period.

[0056] Step S142: Combine the load-related factors of the temperature during each high-load period of the equipment with the number of high-load periods of the equipment to obtain the degree of influence.

[0057] Calculate the average value of the load-related temperature influence index for all high-load periods of the s-th device. A larger average value indicates a greater impact on the initial temperature uniformity of the monitoring point; the two are positively correlated. Also, obtain the number of high-load periods for the s-th device. A larger number of high-load periods indicates a longer high-load time for the s-th device, resulting in more heat generation, a greater impact on ambient temperature, and a greater impact on the initial temperature uniformity of the monitoring point; the two are also positively correlated.

[0058] In one exemplary embodiment, a specific method for quantifying the degree of influence is given below: ; in, This indicates the degree of influence of the load intensity characteristics of the s-th device during the high-load period on the initial temperature uniformity of the w-th monitoring point. This indicates the number of high-load periods for the s-th device. This represents the total number of load sequence segments for the s-th device. This represents the percentage of high-load periods for the s-th device, i.e., the proportion of time the s-th device is in a high-load operating state. This represents the load intensity characteristics of the s-th device during the n-th high-load period. This represents the normalized result of the Pearson correlation coefficient between the load of the s-th device during the nth high-load period and the temperature of the nth temperature sequence segment.

[0059] Step S15: Based on the degree of influence and the spatial distance between each device and the monitoring point, correct the initial temperature uniformity to obtain the temperature uniformity of the monitoring point.

[0060] Determine the spatial distance between the s-th device and the w-th monitoring point within the influence range of the w-th monitoring point. The spatial distance is obtained by: obtaining the projection point of the w-th monitoring point on the xy plane, obtaining the projection point of the center point of the s-th device on the xy plane, and obtaining the distance between these two projection points on the xy plane, which is used as the spatial distance between the s-th device and the w-th monitoring point.

[0061] In one exemplary embodiment, such as Figure 6 As shown, the following is a specific correction process for initial temperature uniformity: Step S151: Based on the degree of influence of each device within the influence range of the monitoring point, and the spatial distance between each device and the monitoring point, obtain the correction index of each device.

[0062] The higher the influence of the s-th device, the greater its impact on temperature uniformity. The lower the temperature uniformity, the smaller the correction index of the s-th device. The correction index is inversely correlated with the influence level. The smaller the spatial distance between the s-th device and the w-th monitoring point, the deeper the influence of the w-th monitoring point on the s-th device, and the greater its impact on temperature uniformity. The smaller the correction index of the s-th device, the positively correlated with the spatial distance.

[0063] In an exemplary embodiment, the following is a specific method for calculating the correction index of the s-th device: ; in, This represents the correction index for the s-th device within the influence range of the w-th monitoring point. This represents the spatial distance between the s-th device and the w-th monitoring point. Indicates to negative correlation normalization, As The weight, The smaller, The larger the value, the more directly the s-th device affects the ambient temperature of the w-th monitoring point.

[0064] Step S152: Correct the initial temperature uniformity of the monitoring points using the correction coefficient to obtain the temperature uniformity of the monitoring points.

[0065] The correction coefficient of the wth monitoring point is obtained by fusing the correction indices of each device within the influence range of the wth monitoring point. In an exemplary embodiment, the average value of the correction indices of each device within the influence range of the wth monitoring point is calculated as the correction coefficient of the wth monitoring point.

[0066] The initial temperature uniformity of the w-th monitoring point is corrected according to the correction factor of the w-th monitoring point to obtain the temperature uniformity of the w-th monitoring point. In an exemplary embodiment, the product of the correction factor of the w-th monitoring point and the initial temperature uniformity of the w-th monitoring point is calculated to obtain the temperature uniformity of the w-th monitoring point. Thus, the temperature uniformity of each monitoring point is obtained.

[0067] Step S16: Based on the difference in temperature uniformity between any two monitoring points and the spatial distance between any two monitoring points, cluster the monitoring points to obtain several monitoring areas.

[0068] Through the above steps, the temperature uniformity of each monitoring point is obtained. To obtain the difference in temperature uniformity between any two monitoring points, the absolute value of the difference is calculated. The spatial distance between any two monitoring points is also obtained by: obtaining the projection points of each monitoring point on the xy-plane, and then calculating the distance between any two projection points on the xy-plane, which is taken as the spatial distance between the two monitoring points.

[0069] The difference in temperature uniformity between any two monitoring points and the spatial distance between them are used as clustering factors to cluster the monitoring points, resulting in several monitoring areas. In an exemplary embodiment, for any two monitoring points, the spatial distance between them is normalized, and the normalized spatial distance is set as parameter A. The absolute value of the difference in temperature uniformity between the two monitoring points is set as parameter B, and the average of parameters A and B is used as the cluster distance between the two monitoring points. This yields the cluster distances for all any two monitoring points. Then, the K-means clustering algorithm is used to cluster the data based on these cluster distances, where the K value can be obtained using the elbow method or the silhouette coefficient. This results in several clusters, each representing an area within the computer room, thus creating several monitoring areas.

[0070] Step S2: Based on the load difference between any two load-stable periods of the equipment within the monitoring area, cluster the load-stable periods to obtain several clusters.

[0071] Since monitoring areas with higher loads generate more heat, their temperature safety thresholds need to be appropriately increased, while those with lower loads can have their temperature safety thresholds appropriately reduced. This allows for early warnings when the temperature in a local monitoring area reaches or approaches the set temperature safety threshold, preventing problems such as equipment overheating and system failures caused by abnormal temperatures in local monitoring areas.

[0072] As equipment is used for longer periods, its hardware components (such as cooling systems and processors) may age or wear out, leading to a decrease in heat dissipation performance and affecting the equipment's temperature resistance. If only the initial temperature safety threshold is relied upon, while ignoring the equipment's health condition, and the initial temperature safety threshold is continued to be used as the judgment standard, the equipment may overheat, thereby accelerating damage and reducing the accuracy of fault detection.

[0073] For any given monitoring area, taking the q-th monitoring area as an example, we will analyze and determine the various devices within the q-th monitoring area.

[0074] This invention aims to adjust a preset initial temperature safety threshold, therefore, it is necessary to determine the preset initial temperature safety threshold. The preset initial temperature safety threshold can be a known preset value. In an exemplary embodiment, the normal temperature range of each device within the q-th monitoring area is obtained. The normal temperature range represents the temperature range within which the corresponding device can operate normally and stably, and different devices may have different normal temperature ranges. The upper limit of the normal temperature range of each device within the q-th monitoring area is obtained, and the minimum value is determined from the upper limit values ​​corresponding to each device within the q-th monitoring area. This minimum value is used as the preset initial temperature safety threshold for the q-th monitoring area.

[0075] Similarly, let the s-th device represent any device within the q-th monitoring area. Based on the load difference between any two load-stable periods of the s-th device within the q-th monitoring area, cluster the load-stable periods to obtain several clusters.

[0076] First, the load sequence of the s-th device within a historical time period is divided into several stable load periods. Specifically, the absolute value of the difference between the load values ​​of any two adjacent moments in the load sequence of the s-th device is calculated. This absolute value is then compared to a preset load difference threshold. The absolute values ​​of the load difference that are less than or equal to the preset threshold are identified, and the consecutive moments corresponding to these thresholds constitute a stable load period. This process yields several stable load periods for the s-th device. The preset load difference threshold is set according to actual needs, for example, 0.3.

[0077] For any given stable load period, the average load value within that period is used as the load reference value for that period. The absolute value of the difference between the load reference values ​​of any two stable load periods for the s-th device is obtained as the cluster distance between those two periods. The K-means clustering algorithm is then used to cluster the stable load periods of the s-th device, where the K value can be obtained using the elbow method or the silhouette coefficient. This results in several clusters, each representing several periods with similar loads.

[0078] Step S3: Based on the temperature range of all load stabilization periods of the cluster and the temperature stability during the load stabilization periods, obtain the correlation between the load and temperature of the cluster.

[0079] Under normal circumstances, when the equipment load remains constant, the equipment temperature should remain within a relatively stable range. However, as the equipment ages, its heat dissipation efficiency decreases, which may cause the temperature to gradually rise. Continuing to use the initial temperature safety threshold for fault warning is prone to false alarms and missed alarms.

[0080] Obtain the operating temperature sequence of the s-th device within a historical time period. For any cluster of the s-th device, obtain the operating temperature sequence segment of the s-th device that corresponds to the operating temperature sequence segment of each load stable period in that cluster, thus obtaining the operating temperature sequence segment that corresponds one-to-one with each load stable period. Obtain the maximum and minimum operating temperatures of the devices among the several operating temperature sequence segments contained in that cluster, and calculate the temperature difference between the maximum and minimum operating temperatures, which is taken as the temperature range of that cluster.

[0081] Then, the temperature stability of each load stabilization period in this cluster is determined. In an exemplary embodiment, such as... Figure 7 As shown, the process of obtaining temperature stability includes: Step S31: Obtain the number of key temperature data points in the working temperature sequence segment.

[0082] For any given period of stable load, obtain the operating temperature data of the equipment during the same period of stable load, i.e., the operating temperature sequence segment during the same period of stable load.

[0083] Key temperature data points are obtained from the operating temperature sequence segment. These key temperature data points represent temperature fluctuations. Specifically, each temperature maximum in the operating temperature sequence segment is obtained, and the difference between each temperature maximum and its adjacent minimum is obtained as the temperature range of each temperature maximum. The time interval between each temperature maximum and its adjacent minimum is determined as the temperature change time interval.

[0084] For any given temperature maximum, if the temperature range corresponding to that maximum is greater than a preset temperature difference, and the time interval for temperature change corresponding to that maximum is less than a preset time interval, it indicates that the temperature change at that maximum is relatively drastic, and thus that maximum is considered a critical temperature data point. All temperature maximums in the working temperature sequence segment are iterated through to obtain each critical temperature data point, and the total number of critical temperature data points in the working temperature sequence segment is determined. The preset temperature difference and preset time interval are set according to actual judgment needs.

[0085] Then, for any key temperature data point, the ratio of the temperature range of that key temperature data point to the time interval of temperature change at that key temperature data point is calculated, which is taken as the rate of temperature change at that key temperature data point.

[0086] Step S32: Obtain the maximum temperature change rate of key temperature data points.

[0087] The maximum value of the temperature change rate among the key temperature data points in the working temperature sequence segment is obtained as the maximum temperature change rate of the working temperature sequence segment.

[0088] Step S33: Based on the number of key temperature data points and the maximum rate of temperature change, obtain the temperature stability during the load stabilization period.

[0089] Obtain the number of key temperature data points and the maximum rate of temperature change in the operating temperature sequence segment (i.e., the load stability period corresponding to the operating temperature sequence segment). Based on the number of key temperature data points and the maximum rate of temperature change in the operating temperature sequence segment, obtain the temperature stability of the load stability period corresponding to the operating temperature sequence segment. The more key temperature data points there are, the more unstable temperature data points there are, the more unstable the temperature is, and the lower the temperature stability. Therefore, temperature stability is inversely correlated with the number of key temperature data points. The larger the maximum rate of temperature change, the more unstable the temperature is during the load stability period, and the lower the temperature stability is. Therefore, temperature stability is negatively correlated with the maximum rate of temperature change. The following is a specific quantification method for temperature stability: ; in, This represents the temperature stability of the s-th device within the q-th monitoring area during the 0-th period of load stability. This represents the number of key temperature data points for the s-th device within the q-th monitoring area during the 0-th period of load stability. This represents the number of temperature data points for the s-th device within the q-th monitoring area during the 0-th period of load stability. This represents the percentage of key temperature data points for the s-th device within the q-th monitoring area during the 0-th load stable period. The higher the percentage, the lower the efficiency of the device's heat dissipation system and the more likely it is to age. This represents the maximum temperature change rate of the s-th device within the q-th monitoring area during the 0-th load stabilization period. The larger the value, the more unstable the temperature of the s-th device in the q-th monitoring area during the 0-th load stable period, and the more likely the device is to age.

[0090] Then obtain the correlation between load and temperature for this cluster, such as Figure 8 As shown, the following is a specific process for obtaining the correlation: Step S34: Determine the overall temperature stability level for all load stabilization periods in the cluster.

[0091] The temperature stability of each load during a stable period within the cluster is obtained, and then the average temperature stability of each load during a stable period within the cluster is calculated as the overall temperature stability level of the cluster. The higher the overall temperature stability level of the cluster, the stronger the positive correlation between the load and temperature within the cluster, i.e., the stronger the correlation between the load and temperature within the cluster, and the two are positively correlated.

[0092] Step S35: Based on the temperature difference and the overall level of temperature stability, obtain the correlation between the cluster's load and temperature.

[0093] The larger the temperature difference between the maximum and minimum operating temperatures of the equipment in this cluster, i.e., the wider the temperature range during all stable load periods for this cluster, the greater the temperature difference within the cluster, the more unstable the temperature, and the less it conforms to the positive correlation between load and temperature. The correlation between load and temperature in this cluster is weaker, and the two are inversely correlated. Based on the temperature difference and overall temperature stability level of this cluster, the correlation between load and temperature for this cluster is obtained. In an exemplary embodiment, the following quantification method is given: ; in, This represents the s-th device within the q-th monitoring area. The correlation between load and temperature for each cluster. This represents the s-th device within the q-th monitoring area. The temperature difference between the maximum and minimum operating temperatures of each cluster of devices. This represents the s-th device within the q-th monitoring area. The overall temperature stability level of each cluster. Similarly, if the equipment operating temperature has not been normalized beforehand, the above formula needs to include... Replace with The negative correlation normalization results, for example: .

[0094] Using the above process, the correlation between load and temperature for each cluster of the s-th device is obtained. Abnormal clusters are then selected based on these correlations. In an exemplary embodiment, a correlation threshold is preset to determine whether the correlation of each cluster is high. The value range of this correlation threshold is 0-1, and the specific value is set according to actual needs; this embodiment uses 0.5 as an example. The correlation between load and temperature for each cluster of the s-th device is compared with this correlation threshold. Clusters with correlations less than or equal to the threshold are identified as abnormal clusters of the s-th device.

[0095] Step S4: Combine the correlation between the various abnormal clusters of the integrated equipment and the number of stable load periods contained therein, and combine this with the number of abnormal clusters of the equipment to obtain the aging degree of the equipment.

[0096] In one exemplary embodiment, such as Figure 9 As shown below, a specific process for obtaining the aging degree of equipment is presented: Step S41: Integrate the correlation between each abnormal cluster and the number of stable load periods contained therein to obtain the overall impact index of the abnormal cluster on equipment aging.

[0097] For any anomalous cluster of the s-th device, the number of stable load periods contained in that cluster is obtained. Combined with the correlation between load and temperature for that cluster, the overall impact index of that cluster on device aging is derived. The stronger the correlation between load and temperature for that cluster, the weaker the influence of temperature on device aging, i.e., the lower the overall impact index on device aging. In this case, the overall impact index is inversely correlated with the correlation. The more stable load periods contained in that cluster, the higher the reliability of the inverse correlation between the correlation and the overall impact index; i.e., the overall impact index is positively correlated with the number of stable load periods it contains.

[0098] Step S42: Based on the number of abnormal clusters of the equipment and the overall impact index, obtain the aging degree of the equipment.

[0099] Obtain the number of anomalous clusters for the s-th device. The more anomalous clusters there are, the more severe the aging of the s-th device; the two are positively correlated. The stronger the overall impact index of each anomalous cluster of the s-th device, the more severe the aging of the s-th device; the two are also positively correlated.

[0100] In one exemplary embodiment, the following is a specific method for quantifying the aging degree of the s-th device: ; in, This indicates the aging degree of the s-th device within the q-th monitoring area. This represents the number of abnormal clusters for the s-th device within the q-th monitoring area. This represents the total number of clusters for the s-th device within the q-th monitoring area. This represents the percentage of abnormal clusters for the s-th device within the q-th monitoring area. The larger the percentage, the greater the degree of abnormality and the higher the degree of aging of the device. This represents the correlation between load and temperature for the m-th anomaly cluster of the s-th device within the q-th monitoring area. This represents the number of periods of stable load contained in the m-th anomaly cluster of the s-th device within the q-th monitoring area. This represents the total number of stable load periods contained in all anomaly clusters of the s-th device within the q-th monitoring area. This represents the percentage of periods with stable load contained in the m-th anomaly cluster of the s-th device within the q-th monitoring area.

[0101] Using the above process, the aging degree of each device in the q-th monitoring area is obtained, and then the aging degree of each device in each monitoring area is obtained.

[0102] The more aged the equipment, the faster the efficiency of its cooling system declines, and the temperature will gradually rise even under constant load. Therefore, by analyzing the aging level of each device in each monitoring area, the temperature safety threshold for that area can be determined, improving the accuracy of fault early warning in the computer room.

[0103] Step S5: Obtain the temperature safety threshold of the monitoring area based on the aging degree of each device and the temperature uniformity within the monitoring area.

[0104] In one exemplary embodiment, such as Figure 10 As shown below, a specific process for obtaining the temperature safety threshold is given: Step S51: Obtain the temperature regulation coefficient of the monitoring area based on the maximum degree of aging and the overall level of temperature uniformity.

[0105] Obtain the maximum aging degree among all devices in the q-th monitoring area to obtain the maximum aging degree of the q-th monitoring area. Obtain the temperature uniformity of each monitoring point in the q-th monitoring area, and calculate the average temperature uniformity of each monitoring point in the q-th monitoring area as the overall temperature uniformity level of the q-th monitoring area.

[0106] The higher the maximum aging degree of the q-th monitoring area, the more necessary it is to lower the temperature safety threshold; therefore, the smaller the temperature regulation coefficient of the q-th monitoring area. The temperature regulation coefficient is inversely correlated with the maximum aging degree. Conversely, the higher the overall temperature uniformity level of the q-th monitoring area, the less necessary it is to lower the temperature safety threshold; that is, the larger the temperature regulation coefficient of the q-th monitoring area. The temperature regulation coefficient is positively correlated with the overall temperature uniformity level.

[0107] In an exemplary embodiment, a specific quantification method for the temperature regulation coefficient of the q-th monitoring area is given below: ; in, This represents the temperature regulation coefficient for the q-th monitoring area. This represents the overall level of temperature uniformity in the q-th monitoring area. This represents the maximum degree of aging in the q-th monitoring area.

[0108] The above calculation method ensures that the temperature regulation coefficient of the monitoring area ranges from 0 to 1. To ensure that when the temperature regulation coefficient is high, the adjusted temperature safety threshold of the monitoring area is slightly larger than the preset initial temperature safety threshold, and when the temperature regulation coefficient is low, the adjusted temperature safety threshold of the monitoring area is slightly smaller than the preset initial temperature safety threshold, in a more preferred embodiment, the temperature regulation coefficient of the q-th monitoring area can also be calculated as follows: ; Using the above method, the temperature regulation coefficient of the qth monitoring area is in the range of 0.7-1.2.

[0109] Using the above method, the temperature adjustment coefficients for each monitoring area are obtained. It should be understood that of the two methods for obtaining these temperature adjustment coefficients, the first method results in a lower final temperature safety threshold, thus making temperature warnings safer; the second method is more in line with the actual conditions of the computer room. Implementers can choose either method for obtaining the temperature adjustment coefficients as needed.

[0110] Step S52: Adjust the preset initial temperature safety threshold by the temperature adjustment coefficient of the monitoring area to obtain the temperature safety threshold of the monitoring area.

[0111] The temperature safety threshold for each monitoring area is obtained by multiplying its temperature regulation coefficient by the preset initial temperature safety threshold for that area. For example, the temperature safety threshold for the q-th monitoring area is obtained by multiplying its temperature regulation coefficient by the preset initial temperature safety threshold for that area.

[0112] This yields the temperature safety thresholds for each monitoring area. During subsequent data center maintenance and fault analysis, for any given monitoring area, the actual temperature of that area (e.g., the average of the actual temperatures at each monitoring point within the area) is obtained and compared to the temperature safety threshold. If the temperature exceeds the safety threshold, an alarm signal is output to promptly notify maintenance personnel of the abnormal temperature in that monitoring area. Furthermore, the temperature safety thresholds for each monitoring area can be dynamically adjusted in real time.

[0113] This solution determines the temperature safety threshold for each monitoring area by analyzing the aging degree of the equipment in each monitoring area of ​​the computer room. This avoids the problem that a uniform temperature safety threshold cannot effectively reflect the actual risk due to uneven equipment aging and temperature distribution in different locations within the computer room. The temperature safety threshold for each monitoring area is dynamically adjusted, which improves the accuracy of fault early warning.

[0114] 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. The processes depicted in the accompanying 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.

[0115] 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.

Claims

1. A data center operation and maintenance fault analysis method utilizing full lifecycle monitoring, characterized in that, include: The computer room was divided into several monitoring areas; Based on the load difference between any two load-stable periods of the equipment within the monitoring area, the load-stable periods are clustered to obtain several clusters. Based on the temperature range of all load stabilization periods of the cluster, and the temperature stability during the load stabilization periods, the correlation between load and temperature of the cluster is obtained; The aging degree of the equipment is obtained by combining the correlation of each abnormal cluster of the fusion equipment and the number of load stable periods contained therein, together with the number of abnormal clusters of the equipment; the abnormal clusters are obtained by filtering the correlation. The temperature safety threshold of the monitoring area is obtained by considering the aging degree of each device and the temperature uniformity within the monitoring area.

2. The data center operation and maintenance fault analysis method using full lifecycle monitoring as described in claim 1, characterized in that, The division of the computer room into several monitoring areas includes: The temperature of each monitoring point in the computer room within a historical time period is divided into several temperature sequence segments. The initial temperature uniformity of the monitoring points is obtained based on the difference between the temperature of each temperature sequence segment and the preset standard temperature. Determine the load intensity characteristics of each device within the influence range of the monitoring point at the same time as the temperature sequence segment to obtain the high load period; Determine the extent to which the load intensity characteristics of each device during high-load periods affect the initial temperature uniformity of the monitoring points; Based on the degree of influence and the spatial distance between each device and the monitoring point, the initial temperature uniformity is corrected to obtain the temperature uniformity of the monitoring point; Based on the difference in temperature uniformity between any two monitoring points and the spatial distance between any two monitoring points, the monitoring points are clustered to obtain several monitoring areas.

3. The data center operation and maintenance fault analysis method using full lifecycle monitoring as described in claim 2, characterized in that, The process of obtaining the initial temperature uniformity includes: By integrating the differences between the characteristic temperatures of each temperature sequence segment at the monitoring point and the preset standard temperature, the overall temperature deviation of the monitoring point can be obtained. The initial temperature uniformity of the monitoring points is obtained based on the overall temperature deviation and the length ratio of the longest temperature sequence segment at each monitoring point. The initial temperature uniformity is inversely correlated with the overall temperature deviation and positively correlated with the length ratio.

4. The data center operation and maintenance fault analysis method using full lifecycle monitoring as described in claim 2, characterized in that, The process of obtaining the load strength characteristics includes: Obtain the minimum load value of the device within the same time period as the temperature sequence segment, as well as the load change trend; The load intensity characteristics are obtained from the minimum load value and the load change trend.

5. The data center operation and maintenance fault analysis method using full lifecycle monitoring as described in claim 2, characterized in that, The process of obtaining the degree of influence includes: Based on the load intensity characteristics of the equipment during each high-load period and the correlation between load and temperature, an index of the influence of load on temperature during each high-load period is obtained; the index of the influence of load on temperature is positively correlated with the load intensity characteristics and the correlation between load and temperature. The influence index of temperature on load during each high-load period of the fusion device is combined with the number of high-load periods of the device to obtain the degree of influence; the degree of influence is positively correlated with both the influence index of temperature on load and the number of high-load periods.

6. The data center operation and maintenance fault analysis method using full lifecycle monitoring as described in claim 2, characterized in that, The process for correcting the initial temperature uniformity includes: Based on the degree of influence of each device within the influence range of the monitoring point, and the spatial distance between each device and the monitoring point, a correction index for each device is obtained; the correction index is inversely correlated with the degree of influence and positively correlated with the spatial distance between the device and the monitoring point. The initial temperature uniformity of the monitoring points is corrected by the correction coefficients of the monitoring points to obtain the temperature uniformity of the monitoring points; all correction coefficients are obtained by fusing the correction indices of all equipment within the influence range of the monitoring points.

7. The data center operation and maintenance fault analysis method using full lifecycle monitoring as described in claim 1, characterized in that, The process of obtaining the temperature stability during the load stabilization period includes: The number of key temperature data points in the operating temperature sequence segment is obtained; the key temperature data points represent temperature data points that indicate temperature fluctuations; the operating temperature sequence segment is the operating temperature data of the equipment during the same period as the load stability period. Obtain the maximum rate of temperature change at the key temperature data points. The temperature stability during the load stabilization period is obtained based on the number of key temperature data points and the maximum temperature change rate; the temperature stability is negatively correlated with both the number of key temperature data points and the maximum temperature change rate.

8. The data center operation and maintenance fault analysis method using full lifecycle monitoring as described in claim 1, characterized in that, The temperature range for all load stabilization periods of the cluster is: the temperature difference between the maximum and minimum device operating temperatures during all load stabilization periods in the cluster. The process of obtaining the correlation between the load and temperature of the aforementioned cluster includes: Determine the overall temperature stability level for all load stabilization periods within the cluster; Based on the temperature difference and the overall level of temperature stability, the correlation between the load and temperature of the cluster is obtained; the correlation is positively correlated with the overall level of temperature stability and negatively correlated with the temperature difference.

9. The data center operation and maintenance fault analysis method using full lifecycle monitoring as described in claim 1, characterized in that, The process of obtaining the aging degree of the equipment includes: By integrating the correlations of each abnormal cluster and the number of stable load periods it contains, an overall impact index on equipment aging is obtained; the overall impact index is inversely correlated with the correlations and positively correlated with the number of stable load periods it contains. The aging degree of the equipment is obtained based on the number of abnormal clusters and the overall impact index. The aging degree is positively correlated with both the number of abnormal clusters and the overall impact index.

10. The data center operation and maintenance fault analysis method using full lifecycle monitoring as described in claim 2, characterized in that, The process of obtaining the temperature safety threshold of the monitoring area includes: The temperature regulation coefficient of the monitoring area is obtained based on the maximum aging degree and the overall temperature uniformity level. The temperature regulation coefficient is inversely correlated with the maximum aging degree and positively correlated with the overall temperature uniformity level. The maximum aging degree is the maximum value among the aging degrees of each device in the monitoring area. The overall temperature uniformity level is obtained by integrating the temperature uniformity of each monitoring point in the monitoring area. The temperature safety threshold of the monitoring area is obtained by adjusting the preset initial temperature safety threshold using the temperature adjustment coefficient of the monitoring area.

Images