Machine room distribution line monitoring method, device and equipment based on machine learning
Through machine learning technology, the distribution lines of the data center computer room are detected online in real time, and abnormalities are diagnosed using hierarchical clustering methods, which solves the problems of timely detection of safety hazards of distribution lines and high manpower investment in the existing technology, and improves the real-time fault diagnosis Sex and intelligence.
Patent Information
- Application Number
- CN202411152073.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-21
- Publication Date
- 2024-11-22
AI Technical Summary
The existing technology cannot detect safety hazards in the distribution lines of data center computer rooms in a timely manner, and the labor investment cost for monitoring is high, and the real-time fault diagnosis is insufficient.
Using machine learning-based methods, the line voltage drop and length data of distribution lines are obtained through the computer room basic environment monitoring platform and the server out-of-band management platform, and fault diagnosis is performed using hierarchical clustering methods to realize online real-time detection and alarm, saving manpower. Reduce hardware investment.
It realizes timely fault diagnosis of power distribution lines in data center computer rooms, reduces manpower and resource investment in operation and maintenance, and improves the real-time and intelligent level of line fault detection.
Smart Images

Figure CN119001328A_ABST
Abstract
Description
Technical Field
[0001] The embodiments of this specification relate to the field of power supply and distribution monitoring and fault diagnosis in a data center computer room, and in particular to a method, device and equipment for monitoring power distribution lines in a computer room based on machine learning. Background Art
[0002] The power supply line from UPS or precision power distribution cabinet to cabinet server is prone to problems such as virtual connection, aging, poor contact, and cable quality of small busbars and cables, which leads to increased local resistance of the power supply line and local overheating. The existing technical solutions are mainly through the following two methods: The first is to measure the voltage drop at both ends of the distribution line when the computer room is put into use, and use a dummy load, and use an infrared measuring instrument to measure the temperature of the distribution cable after a period of load. The second is to use an infrared thermometer to measure the temperature of the distribution cable regularly (generally quarterly or annually) during daily operation and maintenance inspections. Detect local excessive temperatures caused by virtual connection, aging, poor contact, and cable quality of small busbars and cables.
[0003] The first technical solution is not timely enough and cannot detect problems in time. Only after the circuit has already increased resistance due to local problems, resulting in local temperature increases, can problems be discovered through regular (monthly, quarterly, annual) spot checks, or even after the line is unavailable. The second technical solution has a high labor cost and is difficult to conduct regular full coverage inspections.
[0004] How to timely discover the safety hazards in the power distribution lines of the data center computer room, reduce the manpower input cost of monitoring, and improve the real-time performance of line fault diagnosis is a technical problem that needs to be solved urgently. Summary of the invention
[0005] In order to solve the problem that the existing technology cannot timely discover the safety hazards of the power distribution lines in the data center computer room, and the monitoring manpower investment cost is high, the embodiments of this specification provide a computer room power distribution line monitoring method, device and equipment based on machine learning, which obtains monitoring data by monitoring the computer room power distribution (UPS output, precision distribution cabinet) and server out-of-band management monitoring, and after processing the monitoring data, diagnoses the power line fault between the computer room power distribution cabinet and the server through machine learning. This method has online real-time detection and alarm functions, saves manpower, does not require additional hardware investment, and has a high degree of intelligence.
[0006] The specific technical solutions of the embodiments of this specification are as follows:
[0007] On the one hand, an embodiment of this specification provides a method for monitoring power distribution lines in a computer room based on machine learning, the method comprising:
[0008] Obtain the line voltage drop and line length of multiple target distribution lines through the basic environment monitoring platform of the computer room and the server out-of-band management platform;
[0009] Taking the line voltage drop and line length of each target distribution line as a sample of the target distribution line, obtaining sample set data corresponding to the multiple target distribution lines;
[0010] After processing the sample set data, the hierarchical clustering method in machine learning is used to perform fault diagnosis and analysis on the power distribution lines in the data center computer room to obtain abnormal power distribution lines;
[0011] For abnormal distribution lines, on-site staff are prompted to troubleshoot and confirm the abnormal distribution lines.
[0012] Furthermore, through the hierarchical clustering method in machine learning, the fault diagnosis and analysis of the power distribution lines in the data center computer room is carried out, and the abnormal power distribution lines further include:
[0013] Taking each sample in the sample set data as a cluster;
[0014] Calculate the first distance between any two clusters;
[0015] Merging corresponding clusters according to the first distance to obtain new clusters;
[0016] Determine whether all samples are merged into one cluster;
[0017] If not, repeat the step of calculating the first distance between any two clusters according to the obtained new cluster;
[0018] If yes, clustering is completed;
[0019] An abnormal detection sample is determined from multiple samples according to the hierarchical clustering result, and the distribution line corresponding to the abnormal detection sample is taken as the abnormal distribution line.
[0020] Furthermore, the formula for calculating the first distance between any two clusters is:
[0021]
[0022] Among them, d avg Represents cluster C i With C j The first distance between, dist(x,y) represents cluster C i The sample x and cluster C in j The distance between samples y in .
[0023] Furthermore, determining an abnormal detection sample from a plurality of samples according to the hierarchical clustering result further includes:
[0024] Determine whether the first distance between the cluster with a small number of samples and the cluster with a large number of samples in the two clusters of the next level corresponding to the largest first-level cluster exceeds a threshold;
[0025] If so, the samples in the cluster with a small number of samples are used as anomaly detection samples.
[0026] Furthermore, the method further comprises:
[0027] Calculate the average of the second distances between samples in the cluster with a large number of samples as the reference distance;
[0028] Calculating a third distance between each anomaly detection sample in the cluster with a small number of samples and the centroid of the cluster with a large number of samples;
[0029] The abnormal probability value of the abnormal detection sample is determined according to the ratio between the third distance and the reference distance, and the staff is reminded to check and confirm the distribution line corresponding to the abnormal detection sample according to the abnormal probability value.
[0030] Furthermore, obtaining the line voltage drop and line length of multiple target power distribution lines monitored through the computer room basic environment monitoring platform and the server out-of-band management platform further includes:
[0031] Obtaining line lengths of multiple target power distribution lines from the configuration management submodule of the computer room basic environment monitoring platform;
[0032] Obtaining the power supply end voltage value of the target power distribution line from the computer room basic environment monitoring platform;
[0033] Obtaining the voltage value of the power consumption end of the target power distribution line from the server out-of-band management platform;
[0034] The difference between the voltage value at the power supply end and the voltage value at the power consumption end is calculated to obtain the line voltage drop.
[0035] Furthermore, the method further comprises:
[0036] In the process of obtaining the power terminal voltage value of the target distribution line, if the difference between the power terminal voltage values obtained at two adjacent voltage acquisition time points exceeds the predetermined voltage fluctuation rate maximum parameter, it is determined whether the power terminal voltage value corresponding to the later voltage acquisition time point of the two adjacent voltage acquisition time points exceeds the first predetermined range of the normal voltage value and whether the ratio of the difference between the power terminal voltage values obtained at these two voltage acquisition time points to the power terminal voltage value corresponding to the previous voltage acquisition time point exceeds the second predetermined range. If so, the power terminal voltage value corresponding to the later voltage acquisition time point is set as the standard voltage value.
[0037] Furthermore, the method further comprises:
[0038] In the process of obtaining the voltage value of the power consumption terminal of the target distribution line, if the difference between the power consumption terminal voltage values obtained at two adjacent voltage acquisition time points exceeds a predetermined voltage fluctuation rate maximum parameter, the continuous stable working time of the power consumption terminal is determined according to the power consumption terminal voltage value obtained at the voltage acquisition time point before the two adjacent voltage acquisition time points;
[0039] Taking the line voltage drop and the line length of each target distribution line as a sample of the target distribution line further includes:
[0040] The line voltage drop, line length of each target distribution line and the continuous stable working time of the power consumption end corresponding to the target distribution line are used as samples of the target distribution line.
[0041] Furthermore, the power supply end of the target power distribution line is the voltage output end of the power distribution cabinet, and the power consumption end of the target power distribution line is the voltage input end of the cabinet.
[0042] Furthermore, the multiple target distribution lines include multiple distribution lines between multiple distribution cabinets and the same cabinet.
[0043] Furthermore, obtaining the power consumption end voltage value of the target power distribution line from the server out-of-band management platform further includes:
[0044] Obtaining input voltage values of multiple servers in the cabinet from the server out-of-band management platform;
[0045] The average value of the input voltages of the multiple servers is calculated to obtain the power consumption terminal voltage value.
[0046] On the other hand, the embodiment of this specification also provides a device for monitoring power distribution lines in a computer room based on machine learning, the device comprising:
[0047] The power distribution line data acquisition unit is used to obtain the line voltage drop and line length of multiple target power distribution lines monitored through the computer room basic environment monitoring platform and the server out-of-band management platform;
[0048] A sample construction unit, used to use the line voltage drop and line length of each target distribution line as a sample of the target distribution line, and obtain sample set data corresponding to the multiple target distribution lines;
[0049] An abnormal power distribution line detection unit is used to perform fault diagnosis and analysis on the power distribution lines in the data center computer room through a hierarchical clustering method in machine learning to obtain abnormal power distribution lines;
[0050] The processing unit is used to prompt on-site staff to troubleshoot and confirm abnormal distribution lines.
[0051] On the other hand, an embodiment of the present specification further provides a computer device, including a memory, a processor, and a computer program stored in the memory, and the processor implements the above method when executing the computer program.
[0052] On the other hand, an embodiment of the present specification further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above method when executed by a processor.
[0053] Finally, the embodiments of this specification also provide a computer program product, wherein the computer program product includes a computer program, and the computer program implements the above method when executed by a processor.
[0054] In the embodiments of this specification, the power distribution line failure in the computer room of the data center mainly includes the virtual connection, aging, poor contact, cable quality, etc. of the small busbar and cable from the UPS / precision distribution cabinet to the server, which causes the current to be unable to flow smoothly through the distribution line, and the corresponding distribution line resistance value increases, resulting in local overheating of the line and generating risks. The UPS / precision distribution cabinet can be regarded as a constant voltage source. Under normal circumstances, the voltage of the UPS / precision distribution cabinet is constant, and the cable resistance between the UPS / precision distribution cabinet and the cabinet is also a small and stable value. If a virtual connection or other fault occurs in the circuit, the resistance value of the middle distribution line will increase to N times the normal value. The power at the load end does not fluctuate much within a certain period of time. Therefore, the embodiment of this specification calculates the line voltage drop of each distribution line, and the line voltage drop of the distribution line is positively correlated with the resistance value of the distribution line, thereby indirectly obtaining the change in the resistance value of the distribution line and diagnosing line abnormalities.
[0055] In addition, due to the influence of cabinets in different locations, the line lengths of the power distribution lines in the computer room are different, so the voltage drops of different power supply lines are slightly different. For this reason, the embodiment of this specification introduces the length of the power distribution line on the basis of the line voltage drop as a sample of the power distribution line to obtain a sample set. After data processing of the sample set, the hierarchical clustering method in machine learning is used to perform fault diagnosis and analysis on the power distribution lines in the computer room of the data center. For the distribution lines with abnormal diagnosis, the on-site staff is prompted to troubleshoot and confirm the abnormal distribution lines. The method of the embodiment of this specification can obtain the analyzed data only from the currently deployed monitoring system (the basic environment monitoring platform of the computer room and the server out-of-band management platform), without the need to deploy additional tools or devices, thereby reducing the manpower and resource investment in the operation and maintenance of the computer room, and realizing the online real-time detection capability of the distribution lines. BRIEF DESCRIPTION OF THE DRAWINGS
[0056] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0057] Figure 1 The figure shows a schematic diagram of an implementation system of a method for monitoring power distribution lines in a computer room based on machine learning in an embodiment of this specification;
[0058] Figure 2 The figure shows a flow chart of a method for monitoring power distribution lines in a computer room based on machine learning according to an embodiment of the present specification;
[0059] Figure 3 It is a schematic diagram of a process of obtaining the line voltage drops of multiple target distribution lines to be monitored in an embodiment of this specification;
[0060] Figure 4 It is a schematic diagram of a process of performing hierarchical clustering on the sample set to obtain a hierarchical clustering result in an embodiment of this specification;
[0061] Figure 5 Shown is a schematic diagram of the hierarchical clustering results in the embodiments of this specification;
[0062] Figure 6 It is a schematic diagram of a process of determining an abnormal detection sample from multiple samples according to the hierarchical clustering result in an embodiment of this specification;
[0063] Figure 7 The figure is a schematic diagram of a process of calculating an abnormal probability value of an abnormal detection sample in an embodiment of this specification;
[0064] Figure 8 The figure shows a schematic diagram of the structure of a device for monitoring power distribution lines in a computer room based on machine learning in an embodiment of this specification;
[0065] Fig. 9 The figure is a schematic diagram of the structure of a computer device in an embodiment of the present specification.
[0066]
Description of reference numerals
[0067] 101. Cabinet;
[0068] 1011. Server;
[0069] 102. UPS / precision power distribution cabinet;
[0070] 103. Server out-of-band management platform;
[0071] 104. Computer room basic environment monitoring platform;
[0072] 105. Fault diagnosis platform;
[0073] 106. Monitoring interface
[0074] 801, distribution line data acquisition unit;
[0075] 802, sample construction unit;
[0076] 803. Abnormal distribution line detection unit;
[0077] 804, processing unit;
[0078] 902. Computer equipment;
[0079] 904. Processing equipment;
[0080] 906. Storage resources;
[0081] 908, drive system;
[0082] 910, input / output module;
[0083] 912. Input device;
[0084] 914. Output device;
[0085] 916. Presentation equipment;
[0086] 918. Graphical user interface;
[0087] 920, network interface;
[0088] 922, communication link;
[0089] 924. Communication bus. DETAILED DESCRIPTION
[0090] The following will be combined with the drawings in the embodiments of this specification to clearly and completely describe the technical solutions in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. Based on the embodiments in the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the embodiments of this specification.
[0091] It should be noted that the terms "first", "second", etc. in the description and claims of the embodiments of this specification and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the embodiments of this specification described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, device, product or equipment that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.
[0092] It should be noted that the acquisition, storage, use, and processing of data in the technical solutions of the embodiments of this specification comply with the relevant provisions of national laws and regulations.
[0093] It should be noted that in the embodiments of this specification, certain software, components, models and other existing solutions in the industry may be mentioned, and they should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of this application, but it does not mean that the applicant has or will necessarily use the solution.
[0094] like Figure 1 The figure shows a schematic diagram of an implementation system of a method for monitoring power distribution lines in a computer room based on machine learning in an embodiment of the present specification, including a cabinet 101 and multiple UPS / precision power distribution cabinets 102. Multiple servers 1011 are deployed in the cabinet 101, and the power supply line of each server 1011 is plugged into the plug board of the cabinet 101. Figure 1 Two UPS / precision power distribution cabinets 102 are used to supply power to the same cabinet 101. The two UPS / precision power distribution cabinets 102 are backup power supply methods. For the computer room, these two UPS / precision power distribution cabinets 102 need to supply power to the cabinet 101 normally to avoid power outage in the cabinet when one power supply fails.
[0095] The distribution lines of the embodiments of this specification may refer to a computer room distribution line A between a UPS / precision distribution cabinet 102 and a cabinet 101, and a computer room distribution line B between another UPS / precision distribution cabinet 102 and a cabinet 101. The method of the embodiments of this specification monitors whether the computer room distribution line A and the computer room distribution line B have any safety hazards such as false connection, aging, poor contact, and cable quality.
[0096] In the embodiment of the present specification, the out-of-band management port of the server 1011 in the cabinet 101 is connected to the server out-of-band management platform 103 through the top-of-rack switch, and the server out-of-band management platform 103 sends the collected input voltage of the server 1011 to the fault diagnosis platform 105. The output voltage of the power distribution cabinet is sent to the fault diagnosis platform 105 through the computer room basic environment monitoring platform 104. The configuration data such as the topological structure and line length of each distribution line is static data, which can be obtained from the computer room basic environment monitoring platform 104, or can be entered into the fault diagnosis platform at one time.
[0097] The method of the embodiment of this specification can be applied to the fault diagnosis platform 105, for example, by writing an application program to implement the method of the embodiment of this specification. After the fault diagnosis platform 105 processes the sample set, it uses the hierarchical clustering method in machine learning to perform fault diagnosis and analysis on the power distribution lines in the data center computer room. For the power distribution lines with abnormal diagnosis, the monitoring interface 106 prompts the on-site staff to troubleshoot and confirm the abnormal power distribution lines.
[0098] In addition, it should be noted that Figure 1 What is shown is only one application environment provided by the embodiments of this specification. In actual applications, other application environments may also be included. For example, each server 1011 is directly connected to the busbar of a UPS / precision distribution cabinet 102. The distribution lines between each server 1011 and the UPS / precision distribution cabinet 102 are monitored for safety hazards such as false connection, aging, poor contact, and cable quality through the methods of the embodiments of this specification. This specification does not impose any restrictions.
[0099] In view of the problems existing in the prior art, the embodiments of this specification provide a method for monitoring the power distribution line in a computer room based on machine learning, which obtains monitoring data by monitoring the power distribution in the computer room (UPS output, precision power distribution cabinet) and out-of-band management monitoring of the server, and diagnoses the power line fault between the power distribution cabinet in the computer room and the server through machine learning after processing the monitoring data. The method has the functions of online real-time detection and alarm, saves manpower, does not require additional hardware investment, and has a high degree of intelligence. Figure 2The flowchart of the method for monitoring the power distribution line in the computer room of a data center in the embodiment of this specification is shown. This figure describes the process of monitoring the power distribution line in the computer room of a data center. The sequence of steps listed in the embodiment is only one way of executing the sequence of many steps and does not represent the only execution sequence. When the system or device product is executed in practice, it can be executed in the order of the method shown in the embodiment or the accompanying drawings or in parallel.
[0100] Specifically, as shown in the figure, the method may include:
[0101] Step 201: obtaining the line voltage drop and line length of multiple target distribution lines to be monitored through the computer room basic environment monitoring platform and the server out-of-band management platform;
[0102] Step 202: taking the line voltage drop and line length of each target distribution line as a sample of the target distribution line, and obtaining sample set data corresponding to a plurality of target distribution lines;
[0103] Step 203: After processing the sample set data, a hierarchical clustering method in machine learning is used to perform fault diagnosis and analysis on the power distribution lines in the data center computer room to obtain abnormal power distribution lines;
[0104] Step 204: For the distribution line with a diagnosis abnormality, prompt the on-site staff to troubleshoot and confirm the abnormal distribution line.
[0105] In the embodiments of this specification, the power distribution line failure in the computer room of the data center mainly includes the virtual connection, aging, poor contact, cable quality, etc. of the small busbar and cable from the UPS / precision distribution cabinet to the server, which causes the current to be unable to flow smoothly through the distribution line, and the corresponding distribution line resistance value increases, resulting in local overheating of the line and generating risks. The UPS / precision distribution cabinet can be regarded as a constant voltage source. Under normal circumstances, the voltage of the UPS / precision distribution cabinet is constant, and the cable resistance between the UPS / precision distribution cabinet and the cabinet is also a small and stable value. If a virtual connection or other fault occurs in the circuit, the resistance value of the middle distribution line will increase to N times the normal value. The power at the load end does not fluctuate much within a certain period of time. Therefore, the embodiment of this specification calculates the line voltage drop of each distribution line, and the line voltage drop of the distribution line is positively correlated with the resistance value of the distribution line, thereby indirectly obtaining the change in the resistance value of the distribution line and diagnosing line abnormalities.
[0106] In addition, due to the influence of cabinets in different locations, the line lengths of the power distribution lines in the computer room are different, so the voltage drops of different power supply lines are slightly different. For this reason, the embodiment of this specification introduces the length of the power distribution line on the basis of the line voltage drop as a sample of the power distribution line to obtain a sample set. After data processing of the sample set, the hierarchical clustering method in machine learning is used to perform fault diagnosis and analysis on the power distribution lines in the computer room of the data center. For the distribution lines with abnormal diagnosis, the on-site staff is prompted to troubleshoot and confirm the abnormal distribution lines. The method of the embodiment of this specification can obtain the analyzed data only from the currently deployed monitoring system (the basic environment monitoring platform of the computer room and the server out-of-band management platform), without the need to deploy additional tools or devices, thereby reducing the manpower and resource investment in the operation and maintenance of the computer room, and realizing the online real-time detection capability of the distribution lines.
[0107] The method of the embodiment of this specification can obtain the analyzed data only from the currently deployed monitoring system (computer room basic environment monitoring platform and server out-of-band management platform), without deploying additional monitoring tools, thereby reducing the resource investment cost of monitoring.
[0108] In the embodiments of this specification, the data center computer room infrastructure monitoring system can monitor the output voltage of UPS and precision power distribution cabinet in real time. The input voltage of the server can be obtained in the server out-of-band management system (BMC), but currently most server voltages only have real-time values and no historical value records. The embodiments of this specification obtain the real-time value of each server voltage through the server out-of-band management system, and store the historical value of the voltage at each time node; obtain the output voltage of UPS and precision power distribution cabinet through the data center computer room infrastructure monitoring system, and store the historical value of the output voltage of UPS and precision power distribution cabinet at each time node. The circuit diagnosis from the commonly used precision power distribution cabinet (PDC, Precision Distribution Cabinet) to the server (server) is taken as an example, and the lines between other electrical appliances can also be implemented as a reference.
[0109] In an embodiment of the present specification, the sample set data can be preprocessed by removing outliers, and then the processed sample set data can be hierarchically clustered using a hierarchical clustering method in machine learning. Then, based on the hierarchical clustering results, fault diagnosis and analysis can be performed on the power distribution lines in the data center computer room to obtain abnormal distribution lines.
[0110] In the embodiments of this specification, Figure 3 As shown, the line voltage drop and line length of multiple target power distribution lines monitored by the computer room basic environment monitoring platform and the server out-of-band management platform further include:
[0111] Step 301: Obtain the line lengths of multiple target power distribution lines from the configuration management submodule of the basic environment monitoring platform of the computer room;
[0112] Step 302: Obtaining the power supply end voltage value of the target power distribution line from the basic environment monitoring platform of the computer room;
[0113] Step 303: Obtaining the power consumption end voltage value of the target power distribution line from the server out-of-band management platform;
[0114] Step 304: Calculate the difference between the voltage value at the power supply end and the voltage value at the power consumption end to obtain the line voltage drop.
[0115] In the embodiment of this specification, the power supply end of the target distribution circuit is the voltage output end of the distribution cabinet, and the power consumption end of the target distribution circuit is the voltage input end of the cabinet. The multiple target distribution circuits include multiple distribution circuits between multiple distribution cabinets and the same cabinet.
[0116] Therefore, the embodiments of this specification can monitor the power distribution line between the cabinet and the power distribution cabinet. In some other embodiments of this specification, the power consumption end of the target power distribution line can also be the voltage input end of the server, so as to monitor the power distribution line between the target power distribution cabinet and the server, which is not limited by the embodiments of this specification.
[0117] When the multiple target distribution lines include multiple distribution lines between multiple distribution cabinets and the same cabinet, the calculation formula for the line voltage drop is:
[0118] U ▽ =U PDC -U server
[0119] Among them, U ▽ Indicates line voltage drop, U PDC Indicates the busbar voltage of the precision distribution cabinet, that is, the input side voltage U of the distribution line in , U server Indicates the average voltage of a single circuit corresponding to all servers on the same socket strip in the cabinet, that is, the output voltage U of the distribution line out The topological relationship between precise power distribution cabinets, cabinets, servers and power supply lines can be obtained from the data center configuration management system.
[0120] In an embodiment of the present specification, because there is a brief voltage fluctuation during the server powering on and off, in order to avoid being misjudged as an abnormality due to the voltage fluctuation during the server powering on and off, according to an embodiment of the present specification, the difference impact of the brief voltage fluctuation during the server powering on and off is considered.
[0121] Specifically, according to an embodiment of the present specification, in the process of obtaining the power consumption terminal voltage value of the target distribution line, if the power consumption terminal voltage values (U server(T) and U server(T-1) ) exceeds the predetermined voltage fluctuation rate maximum parameter Then determine the power terminal voltage value U corresponding to the later voltage acquisition time point T among the two adjacent voltage acquisition time points T and T-1. server(T) Whether it exceeds the first predetermined range of the normal voltage value (for example, +7% to -10%) and the two voltage acquisition time points T and T-1 obtained at the power terminal voltage value U server(T) and U server(T-1) The difference between server(T) -U server(T-1) The voltage value U of the power consumption terminal corresponding to the previous voltage acquisition time point T-1 server(T-1) Whether the ratio between exceeds the second predetermined range (for example, +7% to -10%), if so, the power terminal voltage value U corresponding to the subsequent voltage acquisition time point T is obtained server(T) The voltage is set to a standard value (for example, = 220V / 380V, depending on the actual situation). Then the line voltage drop is calculated based on the voltage value at the power consumption end set to the voltage standard value.
[0122] It should be noted that the maximum voltage fluctuation parameter It can be an empirical value, and the first predetermined range and the second predetermined range can also be empirical values. The embodiments of this specification do not limit their specific values.
[0123] According to another embodiment of the present specification, if the voltage values (U server(T) and U server(T-1) ) exceeds the predetermined voltage fluctuation rate maximum parameter The continuous stable working time T of the power consumption terminal is determined according to the voltage values of the power consumption terminal obtained at the two adjacent voltage acquisition time points T and the voltage acquisition time point before T-1 (the voltage acquisition time point before T-1). Δ ; Continuous stable operation time T Δ The definition is: within the time window T Δ Within any two adjacent cycles U server(t) -U server(t-1) Where t<T-1.
[0124] Taking the line voltage drop and the line length of each target distribution line as a sample of the target distribution line further includes:
[0125] The line voltage drop, line length and continuous stable working time T of the power consumption end corresponding to each target distribution line are calculated. Δ As the sample of the target power distribution line. Thus, the sample in step 202 is expanded from a two-dimensional vector (including line voltage drop and line length) to a three-dimensional vector (line voltage drop, line length and continuous stable working time T Δ ), and then perform hierarchical clustering.
[0126] In the embodiment of this specification, hierarchical clustering divides data sets at different levels to form a tree-shaped clustering structure. The division of data sets can have a "bottom-up" aggregation strategy and a "top-down" splitting strategy, and there is no need to specify the number of clusters k. The embodiment of this specification adopts agglomerative hierarchical clustering with an aggregation strategy. This algorithm conforms to the circuit diagnosis characteristics of the data center computer room of the embodiment of this specification, that is, under normal circumstances, the vast majority of lines are normal state samples and should be included in the same large cluster group. Only in a few cases (such as circuit failures), circuit failure samples and normal (the vast majority) samples should not be in the same cluster.
[0127] Algorithm input: Sample set D = {U ▽ , L}, where L represents the line length. In actual use, the number of circuit samples in the fault state is relatively small compared to the number of circuit samples in the normal state. In order to increase the sample set of fault state, some fault states can be supplemented through fault simulation, such as the voltage drop and length data at both ends of the cable under fault conditions such as virtual connection, aging, poor contact, and cable quality of small busbars or cables of different lengths, so as to facilitate the verification of fault diagnosis algorithms and parameter optimization. At the same time, for fault instances that occur in actual production operations, they can be used as typical fault samples and retained in the online analysis sample set to optimize the machine learning algorithm.
[0128] The linkage method selects average linkage, that is, the distance between clusters is the average of the distances of all points. i With C j , the average distance d avg :
[0129]
[0130] Among them, d avg Represents cluster C i With C j The first distance between, dist(x,y) represents cluster C i The sample x and cluster C in j The distance between samples y in .
[0131] Among them, dist can use Hausdorff distance dist H :
[0132] dist H (X,Y)=max(dist h (X,Y),dist h (Y,X)
[0133] dist h (X,Y)=max x∈X min y∈Y ||xy||2
[0134] X, Y represent sample sets.
[0135] The algorithm steps are as follows Figure 4 As shown:
[0136] Step 401: taking each sample in the sample set data as a cluster;
[0137] Step 402: Calculate the first distance between any two clusters;
[0138] Step 403: merging corresponding clusters according to the first distance to obtain a new cluster;
[0139] Step 404: determine whether all samples are merged into one cluster;
[0140] Step 405: If not, repeat the step of calculating the first distance between any two clusters according to the obtained new cluster;
[0141] Step 406: If yes, clustering is completed.
[0142] For example, the hierarchical clustering results can be Figure 5 shown. Figure 5 K in it represents the level of clustering, and A, B, C, D, and E are samples respectively.
[0143] According to one embodiment of this specification, Figure 6 As shown, determining an abnormal detection sample from multiple samples according to the hierarchical clustering result further includes:
[0144] Step 601: Determine whether the first distance between the cluster with a small number of samples and the cluster with a large number of samples in the two clusters of the next level corresponding to the largest first-level cluster exceeds a threshold;
[0145] Step 602: If yes, the samples in the cluster with a small number of samples are used as abnormal detection samples.
[0146] For example, Figure 5K=3 is the largest first-level clustering. At this time, samples A to E all belong to one cluster. The two clusters of the next level K=2 are the cluster including samples A to D and the cluster including only sample E. Then, it is determined whether the first distance between the cluster corresponding to sample E and the cluster corresponding to samples A to D exceeds the threshold. If so, it means that the distribution line corresponding to sample E may be faulty, so the staff needs to check the distribution line corresponding to sample E.
[0147] In some other embodiments of this specification, in order to further improve the inspection efficiency of the staff, such as Figure 7 As shown, the method also includes:
[0148] Step 701: Calculate the average second distance between samples in a cluster with a large number of samples as a reference distance;
[0149] Step 702: Calculate a third distance between each abnormal detection sample in the cluster with a small number of samples and the centroid of the cluster with a large number of samples;
[0150] Step 703: Determine the abnormal probability value of the abnormal detection sample according to the ratio between the third distance and the reference distance, and remind the staff to check and confirm the distribution line corresponding to the abnormal detection sample according to the abnormal probability value.
[0151] Exemplarily, the second average distance between each sample in the cluster including samples A to D is calculated (for example, the distance between each two samples is calculated separately, and then the average is taken to obtain the second average distance, or only the distance between a part of the samples is calculated, and then the average is taken to obtain the second average distance). Then the third distance between the centroid of the cluster corresponding to sample E and the cluster corresponding to samples A to D is calculated. If the third distance is much larger than the reference distance, it means that the probability of a fault in the distribution line corresponding to sample E is greater. Therefore, the abnormal probability value of the abnormal detection sample is determined according to the ratio between the third distance and the reference distance, and the staff is reminded to check and confirm the distribution line corresponding to the abnormal detection sample according to the abnormal probability value.
[0152] Based on the same inventive concept, the embodiment of this specification also provides a device for monitoring power distribution lines in a computer room based on machine learning, such as Figure 8 As shown, the device comprises:
[0153] The power distribution line data acquisition unit 801 is used to acquire the line voltage drop and line length of multiple monitored target power distribution lines through the computer room basic environment monitoring platform and the server out-of-band management platform;
[0154] A sample construction unit 802 is used to use the line voltage drop and line length of each target distribution line as a sample of the target distribution line to obtain sample set data corresponding to the multiple target distribution lines;
[0155] The abnormal power distribution line detection unit 803 is used to perform fault diagnosis and analysis on the power distribution line in the data center computer room by using the hierarchical clustering method in machine learning to obtain the abnormal power distribution line;
[0156] The processing unit 804 is used to prompt the on-site staff to troubleshoot and confirm the abnormal distribution line for the distribution line diagnosed as abnormal.
[0157] Furthermore, through the hierarchical clustering method in machine learning, the fault diagnosis and analysis of the power distribution lines in the data center computer room is carried out, and the abnormal power distribution lines further include:
[0158] Treat each sample in the sample set as a cluster;
[0159] Calculate the first distance between any two clusters;
[0160] Merging corresponding clusters according to the first distance to obtain new clusters;
[0161] Determine whether all samples are merged into one cluster;
[0162] If not, repeat the step of calculating the first distance between any two clusters according to the obtained new cluster;
[0163] If yes, clustering is completed;
[0164] An abnormal detection sample is determined from multiple samples according to the hierarchical clustering result, and the distribution line corresponding to the abnormal detection sample is taken as the abnormal distribution line.
[0165] Furthermore, the formula for calculating the first distance between any two clusters is:
[0166]
[0167] Among them, d avg Represents cluster C i With C j The first distance between, dist(x,y) represents cluster C i The sample x and cluster C in j The distance between samples y in .
[0168] Furthermore, determining an abnormal detection sample from a plurality of samples according to the hierarchical clustering result further includes:
[0169] Determine whether the first distance between the cluster with a small number of samples and the cluster with a large number of samples in the two clusters of the next level corresponding to the largest first-level cluster exceeds a threshold;
[0170] If so, the samples in the cluster with a small number of samples are used as anomaly detection samples.
[0171] Furthermore, the processing unit 804 is further configured to:
[0172] Calculate the average of the second distances between samples in the cluster with a large number of samples as the reference distance;
[0173] Calculating a third distance between each anomaly detection sample in the cluster with a small number of samples and the centroid of the cluster with a large number of samples;
[0174] The abnormal probability value of the abnormal detection sample is determined according to the ratio between the third distance and the reference distance, and the staff is reminded to check and confirm the distribution line corresponding to the abnormal detection sample according to the abnormal probability value.
[0175] Further,
[0176] The line voltage drop and line length of multiple target distribution lines monitored through the computer room basic environment monitoring platform and the server out-of-band management platform further include:
[0177] Obtaining line lengths of multiple target power distribution lines from the configuration management submodule of the computer room basic environment monitoring platform;
[0178] Obtaining the power supply end voltage value of the target power distribution line from the computer room basic environment monitoring platform;
[0179] Obtaining the voltage value of the power consumption end of the target power distribution line from the server out-of-band management platform;
[0180] The difference between the voltage value at the power supply end and the voltage value at the power consumption end is calculated to obtain the line voltage drop.
[0181] Furthermore, the device also includes: a first voltage fluctuation processing unit, which is used to determine whether the power terminal voltage value corresponding to the later voltage acquisition time point of the two adjacent voltage acquisition time points exceeds the first predetermined range of the normal voltage value and whether the ratio of the difference between the power terminal voltage values obtained at the two voltage acquisition time points to the power terminal voltage value corresponding to the previous voltage acquisition time point exceeds the second predetermined range in the process of obtaining the power terminal voltage value of the target distribution line. If so, the power terminal voltage value corresponding to the later voltage acquisition time point is set as the standard voltage value.
[0182] Furthermore, the device further comprises: a second voltage fluctuation processing unit, configured to determine the continuous stable working time of the power consumption terminal according to the power consumption terminal voltage value obtained at the voltage acquisition time point before the two adjacent voltage acquisition time points, if the difference between the power consumption terminal voltage values obtained at two adjacent voltage acquisition time points exceeds a predetermined voltage fluctuation rate maximum parameter during the process of obtaining the power consumption terminal voltage value of the target distribution line;
[0183] Taking the line voltage drop and the line length of each target distribution line as a sample of the target distribution line further includes:
[0184] The line voltage drop, line length of each target distribution line and the continuous stable working time of the power consumption end corresponding to the target distribution line are used as samples of the target distribution line.
[0185] Furthermore, the power supply end of the target power distribution line is the voltage output end of the power distribution cabinet, and the power consumption end of the target power distribution line is the voltage input end of the cabinet.
[0186] Furthermore, the multiple target distribution lines include multiple distribution lines between multiple distribution cabinets and the same cabinet.
[0187] Furthermore, obtaining the power consumption end voltage value of the target power distribution line from the server out-of-band management platform further includes:
[0188] Obtaining input voltage values of multiple servers in the cabinet from the server out-of-band management platform;
[0189] The average value of the input voltages of the multiple servers is calculated to obtain the power consumption terminal voltage value.
[0190] Since the principle of solving the problem by the above device is similar to that of the above method, the implementation of the above system can refer to the implementation of the above method, and the repeated parts will not be repeated.
[0191] like Fig. 9 The figure is a schematic diagram of the structure of the computer device in the embodiment of this specification. The method in the embodiment of this specification can be applied to the computer device in this embodiment.
[0192] Specifically, Fig. 9As shown, the computer device 902 may include one or more processing devices 904, such as one or more central processing units (CPUs), each of which may implement one or more hardware threads. The computer device 902 may also include any storage resource 906, which is used to store any kind of information such as code, settings, data, etc. Without limitation, for example, the storage resource 906 may include any one or more combinations of the following: any type of RAM, any type of ROM, flash memory device, hard disk, optical disk, etc. More generally, any storage resource may use any technology to store information.
[0193] Further, any storage resource may provide volatile or non-volatile retention of information.
[0194] Further, any storage resource may represent a fixed or removable component of the computer device 902. In one case, when the processing device 904 executes the associated instructions stored in any storage resource or combination of storage resources, the computer device 902 may perform any operation of the associated instructions. The computer device 902 also includes one or more drive systems 908 for interacting with any storage resource, such as a hard disk drive system, an optical disk drive system, etc.
[0195] The computer device 902 may also include an input / output module 910 (I / O) for receiving various inputs (via input devices 912) and for providing various outputs (via output devices 914). A specific output mechanism may include a presentation device 916 and an associated graphical user interface (GUI) 918. In other embodiments, the input / output module 910 (I / O), the input device 912, and the output device 914 may not be included, and the computer device 902 may be used as a computer device in a network. The computer device 902 may also include one or more network interfaces 920 for exchanging data with other devices via one or more communication links 922. One or more communication buses 924 couple the components described above together.
[0196] The communication link 922 may be implemented in any manner, for example, through a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication link 922 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc. governed by any protocol or combination of protocols.
[0197] The embodiments of the present specification also provide a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above method when executed by a processor.
[0198] The embodiments of the present specification also provide a computer-readable instruction, wherein when a processor executes the instruction, the program therein causes the processor to execute the above method.
[0199] It should be understood that in the various embodiments of the present specification, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present specification.
[0200] It should also be understood that in the embodiments of this specification, the term "and / or" is only a description of the association relationship of the associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in the embodiments of this specification generally indicates that the associated objects before and after are in an "or" relationship.
[0201] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the embodiments of this specification can be implemented with electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the embodiments of this specification.
[0202] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0203] In the several embodiments provided in the embodiments of this specification, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or it can be an electrical, mechanical or other form of connection.
[0204] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of this specification.
[0205] In addition, each functional unit in each embodiment of the present specification can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of software functional units.
[0206] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of this specification is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the embodiment of this specification. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code.
[0207] The embodiments of this specification use specific embodiments to illustrate the principles and implementation methods of the embodiments of this specification. The description of the above embodiments is only used to help understand the methods and core ideas of the embodiments of this specification. At the same time, for those skilled in the art, according to the ideas of the embodiments of this specification, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the embodiments of this specification.
Claims
1. A method for monitoring power distribution lines in a computer room based on machine learning, characterized in that: The method comprises: Obtain the line voltage drop and line length of multiple target distribution lines through the basic environment monitoring platform of the computer room and the server out-of-band management platform; Taking the line voltage drop and line length of each target distribution line as a sample of the target distribution line, obtaining sample set data corresponding to the multiple target distribution lines; After processing the sample set data, the hierarchical clustering method in machine learning is used to perform fault diagnosis and analysis on the power distribution lines in the data center computer room to obtain abnormal power distribution lines; For abnormal distribution lines, on-site staff are prompted to troubleshoot and confirm the abnormal distribution lines.
2. The method according to claim 1, characterized in that: Through the hierarchical clustering method in machine learning, the fault diagnosis and analysis of the power distribution lines in the data center computer room is carried out, and the abnormal power distribution lines are further included: Taking each sample in the sample set data as a cluster; Calculate the first distance between any two clusters; Merging corresponding clusters according to the first distance to obtain new clusters; Determine whether all samples are merged into one cluster; If not, repeat the step of calculating the first distance between any two clusters according to the obtained new cluster; If yes, clustering is completed; An abnormal detection sample is determined from multiple samples according to the hierarchical clustering result, and the distribution line corresponding to the abnormal detection sample is taken as the abnormal distribution line.
3. The method according to claim 2, characterized in that The formula for calculating the first distance between any two clusters is: Among them, d avg Represents cluster C i With C j The first distance between, dist(x,y) represents cluster C i The sample x and cluster C in j The distance between samples y in .
4. The method according to claim 2, characterized in that: Determining an abnormal detection sample from a plurality of samples according to the hierarchical clustering result further includes: Determine whether the first distance between the cluster with a small number of samples and the cluster with a large number of samples in the two clusters of the next level corresponding to the largest first-level cluster exceeds a threshold; If so, the samples in the cluster with a small number of samples are used as anomaly detection samples.
5. The method according to claim 4, characterized in that The method further comprises: Calculate the average of the second distances between samples in the cluster with a large number of samples as the reference distance; Calculating a third distance between each anomaly detection sample in the cluster with a small number of samples and the centroid of the cluster with a large number of samples; The abnormal probability value of the abnormal detection sample is determined according to the ratio between the third distance and the reference distance, and the staff is reminded to check and confirm the distribution line corresponding to the abnormal detection sample according to the abnormal probability value.
6. The method according to claim 1, characterized in that The line voltage drop and line length of multiple target distribution lines monitored through the computer room basic environment monitoring platform and the server out-of-band management platform further include: Obtaining line lengths of multiple target power distribution lines from the configuration management submodule of the computer room basic environment monitoring platform; Obtaining the power supply end voltage value of the target power distribution line from the computer room basic environment monitoring platform; Obtaining the voltage value of the power consumption end of the target power distribution line from the server out-of-band management platform; The difference between the voltage value at the power supply end and the voltage value at the power consumption end is calculated to obtain the line voltage drop.
7. The method according to claim 6, characterized in that The method further comprises: In the process of obtaining the power terminal voltage value of the target distribution line, if the difference between the power terminal voltage values obtained at two adjacent voltage acquisition time points exceeds the predetermined voltage fluctuation rate maximum parameter, it is determined whether the power terminal voltage value corresponding to the later voltage acquisition time point of the two adjacent voltage acquisition time points exceeds the first predetermined range of the normal voltage value and whether the ratio of the difference between the power terminal voltage values obtained at these two voltage acquisition time points to the power terminal voltage value corresponding to the previous voltage acquisition time point exceeds the second predetermined range. If so, the power terminal voltage value corresponding to the later voltage acquisition time point is set as the standard voltage value.
8. The method according to claim 6, characterized in that The method further comprises: In the process of obtaining the voltage value of the power consumption terminal of the target distribution line, if the difference between the power consumption terminal voltage values obtained at two adjacent voltage acquisition time points exceeds a predetermined voltage fluctuation rate maximum parameter, the continuous stable working time of the power consumption terminal is determined according to the power consumption terminal voltage value obtained at the voltage acquisition time point before the two adjacent voltage acquisition time points; Taking the line voltage drop and the line length of each target distribution line as a sample of the target distribution line further includes: The line voltage drop, line length of each target distribution line and the continuous stable working time of the power consumption end corresponding to the target distribution line are used as samples of the target distribution line.
9. The method according to claim 6, characterized in that The power supply end of the target power distribution line is the voltage output end of the power distribution cabinet, and the power consumption end of the target power distribution line is the voltage input end of the cabinet.
10. The method according to claim 9, characterized in that The multiple target power distribution lines include multiple power distribution lines between multiple power distribution cabinets and the same cabinet.
11. The method according to claim 9, characterized in that Acquiring the power consumption end voltage value of the target power distribution line from the server out-of-band management platform further includes: Obtaining input voltage values of multiple servers in the cabinet from the server out-of-band management platform; The average value of the input voltages of the multiple servers is calculated to obtain the power consumption terminal voltage value.
12. A machine room power distribution line monitoring device based on machine learning, characterized in that: The device comprises: The power distribution line data acquisition unit is used to obtain the line voltage drop and line length of multiple target power distribution lines monitored through the computer room basic environment monitoring platform and the server out-of-band management platform; A sample construction unit, used to use the line voltage drop and line length of each target distribution line as a sample of the target distribution line, and obtain sample set data corresponding to the multiple target distribution lines; An abnormal power distribution line detection unit is used to perform fault diagnosis and analysis on the power distribution lines in the data center computer room through a hierarchical clustering method in machine learning to obtain abnormal power distribution lines; The processing unit is used to prompt on-site staff to troubleshoot and confirm abnormal distribution lines.
13. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 11 is implemented.
14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 11 is implemented.
15. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 11 is implemented.
