Data center power supply and distribution operation state monitoring method and system based on big data analysis

Abstract

The invention discloses a data center power supply and distribution operation state monitoring method and system based on big data analysis, and relates to the technical field of power supply and distribution system operation and maintenance. The method comprises the steps of obtaining a power supply and distribution system topological structure of a data center, constructing a monitoring index system reflecting the subsystem operation state of the power supply and distribution system, obtaining correlation coefficients of monitoring indexes and index influence factors under subsystems, screening main influence factors of the monitoring indexes, and dividing state grades of the main influence factors. Classifying historical data according to the random combination of state levels; constructing a comparison database; determining the state level of the main influence factor real-time data; extracting a comparison data subset corresponding to the state level of the main influence factor real-time data in the comparison database; and according to the obtained outlier factor, judging the abnormal condition of the real-time data of the monitoring index, and further judging the operation state of the power supply and distribution system of the data center.

Description

Technical Field

[0001] The present invention relates to the field of power supply and distribution system operation and maintenance technology, and in particular to a data center power supply and distribution operation status monitoring method and system based on big data analysis. Background Art

[0002] The data center (IDC) serves as the control and dispatch center for the enterprise internet. During information system operation, to prevent power outages that could cause reboots or downtime in servers, minicomputers, network equipment, and other power-consuming devices, momentary power outages exceeding 20ms are unacceptable. Therefore, the power supply to the data center must be secure and reliable. As the core infrastructure for ensuring stable data center operations, real-time monitoring of the data center's power supply and distribution system is essential to ensure stable operations.

[0003] The power supply and distribution system of a data center is complex and has diverse functions. Its main components include the mains power supply subsystem, backup power subsystem, computer power supply system, lighting distribution subsystem, lightning protection subsystem and other auxiliary equipment. The coordination between these parts can ensure the stable operation of the data center.

[0004] Currently, the main methods used to determine the operating status of power supply and distribution systems are statistical methods and machine learning methods. Statistical methods use statistical quantities (such as mean, standard deviation, and variance) of operating status data to determine whether the data is abnormal, and thus whether the system operating status is abnormal. Machine learning methods require training models to identify abnormal data. These methods are usually based on direct analysis of real-time data. In reality, different subsystems have different factors affecting their operating status, and the key indicators reflecting their operating status also vary. Direct data analysis often ignores the mechanism by which real-world factors affect status data. Furthermore, the amount of data to be processed is massive, and the required computing resources increase exponentially, resulting in low monitoring timeliness and difficulty in meeting the actual needs of data centers for power supply and distribution system reliability. To this end, a data center power supply and distribution operating status monitoring method based on big data analysis is proposed. Summary of the Invention

[0005] The main purpose of the present invention is to provide a data center power supply and distribution operation status monitoring method and system based on big data analysis, which can effectively solve the problems in the background technology.

[0006] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0007] The data center power supply and distribution operation status monitoring method based on big data analysis includes:

[0008] Obtain the topology of the power supply and distribution system of the data center to be monitored, build a monitoring indicator system that reflects the operating status of the subsystems of the power supply and distribution system, and obtain historical data on each monitoring indicator and the factors affecting the indicators in the indicator system;

[0009] Perform correlation analysis on the monitoring indicators and the influencing factors of the indicators to obtain the correlation coefficient between the i-th monitoring indicator and the j-th indicator influencing factor under the k-th subsystem Use the correlation coefficient to screen out the main influencing factors of the i-th monitoring indicator under the k-th subsystem;

[0010] Divide the status of each of the main influencing factors into several levels according to historical data, classify the historical data according to random combinations of the status levels, and construct a comparison database, wherein the comparison database contains a number of comparison data sets equal to the number of random combinations of status levels, the comparison data sets contain comparison data subsets equal to the number of monitoring indicators under the k-th subsystem, and the i-th comparison data subset contains data of the i-th monitoring indicator of the k-th subsystem under normal operating conditions;

[0011] Collecting real-time data of the kth subsystem, the real-time data including real-time data of monitoring indicators and real-time data of main influencing factors corresponding to the real-time data of monitoring indicators when collecting the real-time data, and determining its status level according to the real-time data of main influencing factors;

[0012] Extract the comparison data subset corresponding to the real-time data status level of the main influencing factors in the comparison database, calculate the outlier factor of the real-time data of the i-th monitoring indicator in the comparison data subset, and judge the abnormal situation of the real-time data of the i-th monitoring indicator based on the obtained outlier factor.

[0013] On the other hand, this solution also provides a data center power supply and distribution operation status monitoring system based on big data analysis, including:

[0014] An indicator system construction module is used to obtain the topology of the power supply and distribution system of the data center to be monitored and to construct a monitoring indicator system reflecting the operating status of the subsystems of the power supply and distribution system;

[0015] A data acquisition module is used to obtain historical data of each monitoring indicator and its influencing factors in the indicator system and real-time data of the kth subsystem, wherein the real-time data includes the real-time data of the monitoring indicators and the real-time data of the main influencing factors corresponding to the real-time data of the monitoring indicators;

[0016] The data analysis module is used to perform correlation analysis on the historical data of the monitoring indicators and the influencing factors of the indicators, and obtain the correlation coefficient between the i-th monitoring indicator and the j-th indicator influencing factor under the k-th subsystem. Use the correlation coefficient to screen out the main influencing factors of the i-th monitoring indicator under the k-th subsystem;

[0017] A data classification module is used to classify the status of each of the main influencing factors into several levels based on historical data;

[0018] a database construction module, configured to classify historical data according to random combinations of status levels and construct a comparison database, wherein the comparison database contains a number of comparison data sets equal to the number of random combinations of status levels, the comparison data sets contain a number of comparison data subsets equal to the number of monitoring indicators under the k-th subsystem, and the i-th comparison data subset contains data of the i-th monitoring indicator of the k-th subsystem under normal operating conditions;

[0019] The data comparison module is used to determine the status level of the real-time data of the main influencing factors, extract the comparison data subset corresponding to the status level of the real-time data of the main influencing factors in the comparison database, calculate the outlier factor of the real-time data of the i-th monitoring indicator in the comparison data subset, and judge the abnormal situation of the real-time data of the i-th monitoring indicator based on the obtained outlier factor.

[0020] The system further includes a memory, a processor, and a computer program stored in the memory and executable on the processor.

[0021] Furthermore, the correlation coefficient The calculation method includes the following process:

[0022] Step S21: Construct the data sequence E of the monitoring indicators and the data sequence F of the influencing factors of the indicators under the k-th subsystem, where E={E1, E2, ..., E m}; F = {F1, F2, ..., F n};E m is the dataset of the mth monitoring indicator; F n is the dataset of factors influencing the nth indicator;

[0023] Step S22: Standardize the data in the data sequence E and the data sequence F using the following formula: Where, e iq Represented as the i-th data set E in the data sequence E i The qth data; for e iq Standardized data of f jq Represented as the jth data set F in the data sequence F j The qth data; f jq Normalized data; Q is the total amount of data; i = 1, 2, ..., m; j = 1, 2, ..., n;

[0024] Step S23: Calculate the correlation coefficient ξ between each data point in the data sequence F and the data sequence E jq , the calculation formula is: Where ρ is the resolution coefficient, and its value range is (0,1); in, It is expressed as first taking when q=1,2,...,Q, Calculate the minimum value among the results, and then take the minimum value of all the calculated minimum values ​​when i = 1, 2, ..., m; It is expressed as first taking when q=1,2,...,Q, Calculate the maximum value among the results, and then take the maximum value of all the calculated maximum values ​​when i = 1, 2, ..., m;

[0025] Step S24: average all calculated correlation coefficients to obtain the correlation coefficient between the i-th monitoring indicator and the j-th indicator influencing factor. The calculation formula is:

[0026] Furthermore, the screening principle of the main influencing factors is: when the correlation coefficient between the i-th monitoring indicator and the j-th indicator influencing factor is When , the influencing factor of the jth indicator is determined to be the main influencing factor of the i-th monitoring indicator.

[0027] Furthermore, the process of classifying the status of the main influencing factors includes the following steps:

[0028] Step S31: normalize the acquired main influencing factor data into a standard normal distribution, wherein the normalization formula is: In the formula, x is the data value of the main influencing factor; z is the standardized value of x; μ is the mean of the main influencing factor data; σ is the standard deviation of the main influencing factor data;

[0029] Step S32: according to the number of status levels to be divided, use the standard normal distribution table to find the corresponding z score;

[0030] Step S33: Convert the z score to the scale of the original data to obtain the classification threshold of the main influencing factor data. The conversion formula is: p =μ+z p ×σ; where t p is the p-th classification threshold; z p is the pth z score;

[0031] Step S34: Classify the status of the main influencing factor data into p levels according to the determined classification threshold, and the classification rules are:

[0032] When the normalized value of the main influencing factor data is in the interval (-∞, t1], it is the first state level;

[0033] When the normalized value of the main influencing factor data is in the interval (t1, t2], it is the second state level;

[0034] And so on,

[0035] When the standardized value of the main influencing factor data is within the interval (t p , +∞), it is the p+1th state level.

[0036] Furthermore, the outlier factor LOF i The calculation process of (t) includes the following steps:

[0037] Step S51: Assume that the real-time data of the i-th monitoring indicator and the r-th data point Df in the comparison data subset i r The kth data point with the smallest distance is Df i k , r=1,2,...,R, R is the total amount of data;

[0038] Step S52: Calculate the rth data point Df i r The k-nearest neighbor distance d k (Df i r ), the calculation formula is:

[0039] Step S53: Calculate the difference between all other data points and the r-th data point Df in the comparison data subset respectively. i r The distance of the calculated distance value is less than the k nearest neighbor distance d k (Df i r ) as the rth data point Df i r k-distance neighborhood N k (Df i r );

[0040] Step S54: Compare k nearest neighbor distances d k (Df i r ) and k-distance neighborhood N k (Df i r )Medium distance data point Df i k The distance to the nearest k-th data point;

[0041] Step S55: Take the maximum value of the two distance values ​​as the rth data point Df i r With the kth data point as Df i k The k-th reachable distance reach_dist k (Df i k , Df i r );

[0042] Step S56: Calculate the rth data point Df i r The local reachable density ρ k (Df i r ), where the calculation formula is:

[0043] Step S57: According to the rth data point Df i r The local outlier factor LOF of the data point is calculated by the local reachability density k (Df i r ), the calculation formula is: Where, ρ k (Df i k ) is the kth data point Df i k The local reachable density is obtained by calculation.

[0044] Furthermore, the rule for judging whether the real-time data of the i-th monitoring indicator is abnormal based on the outlier factor is: when the outlier factor of the real-time data of the monitoring indicator in the comparison data subset is greater than 2 or greater than the 95% quantile of the outlier factors of all data points in the comparison data subset, it is judged as an abnormal point.

[0045] A data center power supply and distribution operation status monitoring system based on big data analysis, the system is used to implement the steps of the data center power supply and distribution operation status monitoring method based on big data analysis, including:

[0046] An indicator system construction module is used to obtain the topology of the power supply and distribution system of the data center to be monitored and to construct a monitoring indicator system reflecting the operating status of the subsystems of the power supply and distribution system;

[0047] A data acquisition module is used to obtain historical data of each monitoring indicator and its influencing factors in the indicator system and real-time data of the kth subsystem, wherein the real-time data includes the real-time data of the monitoring indicators and the real-time data of the main influencing factors corresponding to the real-time data of the monitoring indicators;

[0048] The data analysis module is used to perform correlation analysis on the historical data of the monitoring indicators and the influencing factors of the indicators, and obtain the correlation coefficient between the i-th monitoring indicator and the j-th indicator influencing factor under the k-th subsystem. Use the correlation coefficient to screen out the main influencing factors of the i-th monitoring indicator under the k-th subsystem;

[0049] A data classification module is used to classify the status of each of the main influencing factors into several levels based on historical data;

[0050] a database construction module, configured to classify historical data according to random combinations of status levels and construct a comparison database, wherein the comparison database contains a number of comparison data sets equal to the number of random combinations of status levels, the comparison data sets contain a number of comparison data subsets equal to the number of monitoring indicators under the k-th subsystem, and the i-th comparison data subset contains data of the i-th monitoring indicator of the k-th subsystem under normal operating conditions;

[0051] The data comparison module is used to determine the status level of the real-time data of the main influencing factors, extract the comparison data subset corresponding to the status level of the real-time data of the main influencing factors in the comparison database, calculate the outlier factor of the real-time data of the i-th monitoring indicator in the comparison data subset, and judge the abnormal situation of the real-time data of the i-th monitoring indicator based on the obtained outlier factor.

[0052] Preferably, the system also includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the method for monitoring the power supply and distribution operation status of a data center based on big data analysis can be implemented.

[0053] The present invention has the following beneficial effects:

[0054] Compared with the existing technology, by obtaining the topological structure of the power supply and distribution system of the data center to be monitored, a monitoring indicator system reflecting the subsystem operating status of the power supply and distribution system is constructed, correlation analysis is performed on the monitoring indicators and indicator influencing factor data, the correlation coefficient of the monitoring indicators and indicator influencing factors under the subsystem is obtained, the main influencing factors of the monitoring indicators are screened out, the status levels of the main influencing factors are divided, the historical data are classified according to the random combination of the status levels, a comparison database is constructed, the real-time data of the subsystem is collected, its status level is determined according to the real-time data of the main influencing factors, the comparison data subset corresponding to the status level of the real-time data of the main influencing factors in the comparison database is extracted, the outlier factor of the real-time data of the monitoring indicator in the comparison data subset is calculated, and the abnormal situation of the real-time data of the monitoring indicator is judged according to the obtained outlier factor. It can comprehensively consider the action mechanism of realistic factors on the status data, and by establishing a comparison database to classify and store data, it can effectively reduce the calculation complexity of the data in the real-time monitoring process, simplify the calculation process of data comparison analysis, and thus help improve the timeliness of system operation status monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a flow chart of a method for monitoring the power supply and distribution operation status of a data center based on big data analysis according to the present invention;

[0056] Figure 2 This is a schematic diagram of the structure of a data center power supply and distribution operation status monitoring system based on big data analysis of the present invention;

[0057] Figure 3 This is a schematic diagram of the topology of the power supply and distribution system of the data center constructed in the embodiment of this solution;

[0058] Figure 4 This is a diagram of the monitoring indicator system of the mains power supply subsystem in the embodiment of this solution;

[0059] Figure 5 This is a monitoring indicator system diagram of the backup power subsystem in the embodiment of this solution;

[0060] Figure 6 This is a monitoring indicator system diagram of the computer power supply system in this embodiment;

[0061] Figure 7 This is a monitoring indicator system diagram of the lighting distribution subsystem in the embodiment of this solution;

[0062] Figure 8 This is a diagram of the monitoring indicator system of the lightning protection subsystem in the embodiment of this solution;

[0063] Figure 9 This is the login interface of the data center power supply and distribution operation status monitoring system based on big data analysis of the present invention;

[0064] Figure 10 This is the submodule interface of the data center power supply and distribution operation status monitoring system based on big data analysis of the present invention;

[0065] Figure 11 This is a schematic diagram of the operation of the indicator system construction module of the present invention;

[0066] Figure 12 This is a schematic diagram of the operation of the data acquisition module of the present invention;

[0067] Figure 13 This is a schematic diagram of the operation of the data analysis module of the present invention;

[0068] Figure 14 This is a schematic diagram of the operation of the data classification module of the present invention;

[0069] Figure 15 This is a schematic diagram of the operation of the database construction module of the present invention;

[0070] Figure 16 Schematic diagram of the operation of the data comparison module of the present invention. DETAILED DESCRIPTION

[0071] The present invention will be further described below in conjunction with specific embodiments. The accompanying drawings are for illustrative purposes only and represent only schematic diagrams rather than actual drawings. They should not be understood as limiting the present invention. In order to better illustrate the specific embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product.

[0072] The specific implementation process of the technical solution of the present invention includes the following steps:

[0073] Step 1: Obtain the power supply and distribution system topology of the data center to be monitored.

[0074] In this embodiment, the main components of the power supply and distribution system include five subsystems: mains power supply subsystem, backup power subsystem, computer power supply system, lighting distribution subsystem and lightning protection subsystem. For each subsystem, the main equipment of the mains power supply subsystem includes: high-voltage incoming line equipment, transformer, low-voltage distribution equipment and power quality improvement equipment; the main equipment of the backup power subsystem includes: diesel generator set, automatic transfer switch (ATS) and UPS (uninterruptible power supply); the main equipment of the computer power supply system includes: UPS system, distribution equipment and monitoring equipment; the main equipment of the lighting distribution subsystem includes lighting fixtures, lighting control equipment and distribution equipment; the main equipment of the lightning protection subsystem includes external lightning protection equipment, internal lightning protection equipment and grounding equipment. Then it can be constructed as follows Figure 3 The power supply and distribution system topology of the data center is shown.

[0075] Step 2: Construct a monitoring indicator system that reflects the operating status of the subsystems of the power supply and distribution system.

[0076] Based on the topology constructed in the above steps, the monitoring indicators for the operating status of the main equipment in each subsystem of the power supply and distribution system mainly include:

[0077] Mains power supply subsystem

[0078] High voltage incoming line equipment:

[0079] Voltage monitoring indicators: such as high-voltage incoming line voltage (three-phase voltage).

[0080] Current monitoring indicators: such as high-voltage incoming line current (three-phase current).

[0081] Switch status indicators: such as the opening and closing status of high-voltage switchgear.

[0082] Insulation monitoring indicators: such as insulation resistance of high-voltage equipment.

[0083] Temperature monitoring indicators: such as the temperature of high-voltage switchgear and high-voltage busbar.

[0084] transformer:

[0085] Winding temperature index: such as the operating temperature of the transformer winding.

[0086] Oil temperature monitoring (for oil-immersed transformers) indicators: such as transformer oil temperature.

[0087] Oil level monitoring (for oil-immersed transformers) indicators: such as transformer oil level.

[0088] Load factor indicators: such as the transformer load factor (the ratio of actual load to rated load). Harmonic content indicators: such as the harmonic voltage and current content of the transformer input and output. Cooling system status indicators: such as the operating status of the transformer cooling fan and oil pump.

[0089] Low voltage distribution equipment:

[0090] Voltage monitoring indicators: such as low-voltage bus voltage (three-phase voltage).

[0091] Current monitoring indicators: such as low-voltage bus current (three-phase current).

[0092] Switch status indicators: such as the opening and closing status of low-voltage circuit breakers.

[0093] Power factor index: such as the power factor of the low-voltage system.

[0094] Harmonic content indicators: such as the harmonic voltage and current content of the low-voltage system.

[0095] Temperature monitoring indicators: such as the temperature of key components in the low-voltage distribution cabinet.

[0096] Power quality improvement equipment:

[0097] Harmonic filter status indicators: such as the on / off status of the harmonic filter.

[0098] Active filter status indicators: such as the operating status and compensation effect of the active filter. Voltage regulator status indicators: such as the output voltage stability and regulation accuracy of the voltage regulator.

[0099] Backup power subsystem

[0100] Diesel generator sets:

[0101] Startup status indicators: such as the startup status of the generator set (start / stop). Run time indicators: such as the cumulative run time of the generator set.

[0102] Oil level monitoring indicators: such as the oil level in the fuel tank of a diesel generator set.

[0103] Coolant temperature indicator: such as the coolant temperature of the diesel generator set. Battery voltage indicator: such as the voltage of the starting battery.

[0104] Output voltage and frequency indicators: such as the output voltage and frequency of the generator set.

[0105] Automatic Transfer Switch (ATS):

[0106] Switching status indicators: such as the switching status of the ATS (mains power / backup power). Switching time indicators: such as the switching time of the ATS.

[0107] Fault status indicators: such as the fault alarm status of ATS.

[0108] UPS (Uninterruptible Power Supply):

[0109] Input voltage and frequency indicators: such as the voltage and frequency at the UPS input terminal. Output voltage and frequency indicators: such as the voltage and frequency at the UPS output terminal.

[0110] Battery voltage indicator: such as the voltage of the UPS battery pack.

[0111] Battery remaining capacity indicator: such as the remaining capacity of the UPS battery.

[0112] Load rate indicator: such as UPS load rate.

[0113] Bypass status indicator: such as the UPS bypass on / off status.

[0114] Inverter status indicators: such as the operating status of the UPS inverter.

[0115] Computer power supply system UPS system:

[0116] Input voltage and frequency indicators: such as the voltage and frequency at the UPS input terminal. Output voltage and frequency indicators: such as the voltage and frequency at the UPS output terminal.

[0117] Battery voltage indicator: such as the voltage of the UPS battery pack.

[0118] Battery remaining capacity indicator: such as the remaining capacity of the UPS battery.

[0119] Load rate indicator: such as UPS load rate.

[0120] Bypass status indicator: such as the UPS bypass on / off status.

[0121] Inverter status indicators: such as the operating status of the UPS inverter.

[0122] Power distribution equipment:

[0123] Voltage monitoring indicators: such as the output voltage of the cabinet and plug-in PDU.

[0124] Current monitoring indicators: such as the output current of the terminal cabinet and plug-in PDU.

[0125] Switch status indicators: such as the open and closed status of the power distribution cabinet and plug-in PDU. Temperature monitoring indicators: such as the temperature of key components of the power distribution cabinet and plug-in PDU. Power factor indicators: such as the power factor of the computer power supply system.

[0126] Monitoring equipment:

[0127] Power parameter monitoring indicators: such as real-time monitoring of voltage, current, power, and other parameters. Environmental parameter monitoring indicators: such as real-time monitoring of environmental parameters such as temperature and humidity. Alarm status indicators: such as the alarm status of monitoring equipment.

[0128] Lighting distribution subsystem

[0129] Lighting fixtures:

[0130] Brightness monitoring indicators: such as the brightness of lighting fixtures.

[0131] Fault status indicators: such as the fault alarm status of lighting fixtures.

[0132] Lighting control equipment:

[0133] Switch status indicators: such as the open and closed status of the lighting switch.

[0134] Controller status indicators: such as the operating status of the lighting controller.

[0135] Power distribution equipment:

[0136] Voltage monitoring indicators: such as the output voltage of the lighting distribution box.

[0137] Current monitoring indicators: such as the output current of the lighting distribution box.

[0138] Switch status indicators: such as the opening and closing status of the lighting circuit breaker.

[0139] Power factor index: such as the power factor of the lighting system.

[0140] Temperature monitoring indicators: such as the temperature of key components of the lighting distribution box.

[0141] Lightning protection subsystem

[0142] External lightning protection equipment:

[0143] Lightning rod status indicators: such as the grounding resistance of the lightning rod.

[0144] Lightning protection belt status indicators: such as the grounding resistance of the lightning protection belt.

[0145] Lightning protection network status indicators: such as the grounding resistance of the lightning protection network.

[0146] Internal lightning protection equipment:

[0147] Arrester status indicators: such as the arrester's grounding resistance and number of operations.

[0148] Surge protective device (SPD) status indicators: such as the SPD's grounding resistance and number of operations.

[0149] Grounding equipment:

[0150] Grounding resistance index: such as the grounding resistance of the grounding electrode.

[0151] Ground wire status indicators: such as the integrity and connection status of the ground wire.

[0152] Ground terminal status indicator: such as the connection status of the ground terminal.

[0153] The monitoring indicator system constructed above can accurately reflect the operating status of each subsystem of the power supply and distribution system.

[0154] Step 3: Obtain historical data of each monitoring indicator and its influencing factors in the indicator system.

[0155] It should be noted that, for the factors affecting indicators, different indicator types have different influencing factors, and the data required to be collected is also different. In the initial stage, the type of data to be collected can be determined by analyzing the factors that may affect the operating status of the power supply and distribution system. In this embodiment, the factors that affect the operating data of the data center power supply and distribution system equipment are summarized as follows:

[0156] 1. Environmental factors

[0157] Ambient temperature and humidity: The operating temperature and humidity of data center power distribution equipment have a direct impact on its performance and lifespan. Excessively high or low temperatures can lead to poor heat dissipation or performance degradation, while abnormal humidity can cause corrosion or short circuits.

[0158] Air cleanliness: Dust and other pollutants may affect the heat dissipation and insulation performance of equipment, and even cause equipment failure.

[0159] Electromagnetic interference: The electromagnetic environment in a data center is complex. Electromagnetic interference may affect the normal operation of the power supply and distribution system, resulting in a decrease in the accuracy of data collection and transmission.

[0160] 2. Technical factors

[0161] Equipment aging: Power supply and distribution equipment will gradually age over long-term operation, leading to performance degradation, such as transformer insulation aging and UPS battery capacity degradation. This can be reflected by collecting equipment failure rate data.

[0162] Power quality: Power quality issues such as voltage fluctuations, harmonics, and frequency deviations can affect the operating status of power supply and distribution equipment, potentially leading to equipment overloads and malfunctions of protective devices.

[0163] 3. Load factor

[0164] Load fluctuations: The IT equipment load in a data center is constantly changing, which can cause fluctuations in parameters such as current and voltage in the power distribution system. Sudden increases or decreases in load can affect the stability and efficiency of the equipment.

[0165] 4. Operation and maintenance management factors

[0166] Maintenance: Regular maintenance can promptly identify and address potential equipment issues, reducing the likelihood of failures. Inadequate maintenance can lead to abnormal equipment operating data. This can be reflected by collecting equipment maintenance cycle data.

[0167] Operational errors: Human operational errors (such as misoperation of switches, configuration errors, etc.) may cause abnormal operation of the power supply and distribution system. This can be reflected by the probability of perceived operational errors.

[0168] 5. Other factors

[0169] Mains power quality: Voltage and frequency fluctuations and power outages of the mains power can affect the operating data of the data center's power supply and distribution system, especially when the mains power switches to a backup power source (such as a diesel generator).

[0170] By collecting data on the above possible influencing factors, it is convenient to conduct subsequent correlation analysis between monitoring indicators and indicator influencing factors.

[0171] Step 4: Perform correlation analysis on the monitoring indicators and the influencing factors of the indicators to obtain the correlation coefficient between the i-th monitoring indicator and the j-th indicator influencing factor under the k-th subsystem. The calculation process includes the following steps:

[0172] Step S41: Construct the data sequence E of the monitoring indicators and the data sequence F of the influencing factors of the indicators under the k-th subsystem, where E={E1, E2, ..., E m}; F = {F1, F2, ..., F n};E m is the dataset of the mth monitoring indicator; F n is the dataset of factors influencing the nth indicator;

[0173] Step S42: Standardize the data in the data sequence E and the data sequence F using the following formula: Where, e iq Represented as the i-th data set E in the data sequence E i The qth data; for e iq Standardized data of f jq Represented as the jth data set F in the data sequence F j The qth data; f jq Normalized data; Q is the total amount of data; i = 1, 2, ..., m; j = 1, 2, ..., n;

[0174] Step S43: Calculate the correlation coefficient ξ between each data point in the data sequence F and the data sequence E jq , the calculation formula is: Where ρ is the resolution coefficient, and its value range is (0,1); in, It is expressed as first taking when q=1,2,...,Q, Calculate the minimum value among the results, and then take the minimum value of all the calculated minimum values ​​when i = 1, 2, ..., m; It is expressed as first taking when q=1,2,...,Q, Calculate the maximum value among the results, and then take the maximum value of all the calculated maximum values ​​when i = 1, 2, ..., m;

[0175] Step S44: average all calculated correlation coefficients to obtain the correlation coefficient between the i-th monitoring indicator and the j-th indicator influencing factor. The calculation formula is:

[0176] Step 5: Use the correlation coefficient to screen out the main influencing factors of the i-th monitoring indicator under the k-th subsystem. The screening principle is: when the correlation coefficient of the i-th monitoring indicator and the influencing factor of the j-th indicator is When , the influencing factor of the jth indicator is determined to be the main influencing factor of the i-th monitoring indicator.

[0177] Step 6: Classify the status of each major influencing factor into several levels based on historical data. The classification process includes the following steps:

[0178] Step S61: normalize the acquired main influencing factor data into a standard normal distribution, wherein the normalization formula is: In the formula, x is the data value of the main influencing factor; z is the standardized value of x; μ is the mean of the main influencing factor data; σ is the standard deviation of the main influencing factor data;

[0179] Step S62: according to the number of status levels to be divided, use the standard normal distribution table to find the corresponding z score;

[0180] Step S63: Convert the z score to the scale of the original data to obtain the classification threshold of the main influencing factor data. The conversion formula is: p =μ+z p ×σ; where t p is the p-th classification threshold; z p is the pth z score;

[0181] Step S64: According to the determined classification threshold, the status of the main influencing factor data is classified into p levels. The classification rule is: when the normalized value of the main influencing factor data is in the interval (∞, t1], it is the first status level; when the normalized value of the main influencing factor data is in the interval (t1, t2], it is the second status level; and so on, when the normalized value of the main influencing factor data is in the interval (t p , +∞), it is the p+1th state level.

[0182] Step 7: Classify the historical data according to the random combinations of status levels and construct a comparison database. The comparison database contains a number of comparison data sets equal to the number of random combinations of status levels. The comparison data sets contain a number of comparison data subsets equal to the number of monitoring indicators under the k-th subsystem, and the i-th comparison data subset contains the data of the i-th monitoring indicator of the k-th subsystem under normal operation.

[0183] Assume that through the above steps, the voltage monitoring indicators of the high-voltage incoming line equipment of the main power supply subsystem are analyzed, and the main influencing factors include voltage fluctuation, grid harmonics and ambient temperature. By analyzing the collected historical data, the status of the main influencing factors is classified into three levels according to low, medium and high. Then the comparison database contains 3×3×3, that is, 27 comparison data sets, and the comparison data sets are subdivided according to different monitoring indicators, and contain comparison data subsets equal to the number of monitoring indicators. Each comparison data subset is used to store the data of the monitoring indicator of the subsystem under normal operating conditions.

[0184] By dividing the database, different monitoring indicator data can be classified and stored, which facilitates the subsequent comparison of real-time data and historical data.

[0185] Step 8: Collect real-time data of the kth subsystem. The real-time data includes the real-time data of monitoring indicators and the real-time data of the main influencing factors corresponding to the real-time data of monitoring indicators. Determine its status level based on the real-time data of the main influencing factors. The determination method is as shown in step 6 and will not be repeated here.

[0186] Step 9: Extract the comparison data subset corresponding to the real-time data status level of the main influencing factors from the comparison database. The comparison data subset stores the data of the monitoring indicator under the normal operation of the system.

[0187] Step 10: Calculate the outlier factor of the real-time data of the i-th monitoring indicator in the comparison data subset. The calculation process includes the following steps:

[0188] Step S51: Assume that the real-time data of the i-th monitoring indicator and the r-th data point Df in the comparison data subset i r The kth data point with the smallest distance is Df i k , r=1,2,...,R, R is the total amount of data;

[0189] Step S52: Calculate the rth data point Df i r The k-nearest neighbor distance d k (Df i r ), the calculation formula is:

[0190] Step S53: Calculate the difference between all other data points and the r-th data point Df in the comparison data subset respectively. i r The distance of the calculated distance value is less than the k nearest neighbor distance d k (Df i r) as the rth data point Df i r k-distance neighborhood N k (Df i r );

[0191] Step S54: Compare k nearest neighbor distances d k (Df i r ) and k-distance neighborhood N k (Df i r )Medium distance data point Df i k The distance to the nearest k-th data point;

[0192] Step S55: Take the maximum value of the two distance values ​​as the rth data point Df i r With the kth data point as Df i k The k-th reachable distance reach_dist k (Df i k , Df i r );

[0193] Step S56: Calculate the rth data point Df i r The local reachable density ρ k (Df i r ), where the calculation formula is:

[0194] Step S57: According to the rth data point Df i r The local outlier factor LOF of the data point is calculated by the local reachability density k (Df i r ), the calculation formula is: Where, ρ k (Df i k ) is the kth data point Df i k The local reachable density is obtained by calculation.

[0195] Step 11: Determine the abnormality of the real-time data of the i-th monitoring indicator based on the obtained outlier factor. The judgment rule is: when the outlier factor of the real-time data of the monitoring indicator in the comparison data subset is greater than 2 or greater than the 95% quantile of the outlier factors of all data points in the comparison data subset, it is determined to be an outlier.

[0196] like Figure 9 As shown, a login diagram of a data center power supply and distribution operation status monitoring system based on big data analysis is provided. The system is used to implement the steps of the data center power supply and distribution operation status monitoring method based on big data analysis, including the following steps: Figure 10 The submodules shown are as follows:

[0197] The indicator system construction module is used to obtain the topology of the power supply and distribution system of the data center to be monitored and to construct a monitoring indicator system that reflects the operating status of the subsystems of the power supply and distribution system, such as Figure 11 As shown;

[0198] The data acquisition module is used to obtain the historical data of each monitoring indicator and the influencing factors of the indicator system and the real-time data of the kth subsystem, wherein the real-time data includes the real-time data of the monitoring indicators and the real-time data of the main influencing factors corresponding to the real-time data of the monitoring indicators. Figure 12 As shown;

[0199] The data analysis module is used to perform correlation analysis on the historical data of the monitoring indicators and the influencing factors of the indicators, and obtain the correlation coefficient between the i-th monitoring indicator and the j-th indicator influencing factor under the k-th subsystem. Use the correlation coefficient to screen out the main influencing factors of the i-th monitoring indicator under the k-th subsystem, such as Figure 13 As shown;

[0200] The data classification module is used to classify the status of each major influencing factor into several levels according to historical data, such as Figure 14 As shown;

[0201] The database construction module is used to classify historical data according to the random combination of status levels and construct a comparison database, wherein the comparison database contains a number of comparison data sets equal to the number of random combinations of status levels, and the comparison data sets contain comparison data subsets equal to the number of monitoring indicators under the k-th subsystem, and the i-th comparison data subset contains the data of the i-th monitoring indicator of the k-th subsystem under normal operation, such as Figure 15 As shown;

[0202] The data comparison module is used to determine the status level of the real-time data of the main influencing factors, extract the comparison data subset corresponding to the status level of the real-time data of the main influencing factors in the comparison database, calculate the outlier factor of the real-time data of the i-th monitoring indicator in the comparison data subset, and judge the abnormality of the real-time data of the i-th monitoring indicator based on the obtained outlier factor, such as Figure 16 As shown;

[0203] Preferably, the system also includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the method for monitoring the power supply and distribution operation status of a data center based on big data analysis can be implemented.

[0204] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A data center power supply and distribution operation status monitoring method based on big data analysis, characterized in that: include: Obtain the topology of the power supply and distribution system of the data center to be monitored, build a monitoring indicator system that reflects the operating status of the subsystems of the power supply and distribution system, and obtain historical data on each monitoring indicator and the factors affecting the indicators in the indicator system; Perform correlation analysis on the monitoring indicators and the influencing factors of the indicators to obtain the correlation coefficient between the i-th monitoring indicator and the j-th indicator influencing factor under the k-th subsystem Use the correlation coefficient to screen out the main influencing factors of the i-th monitoring indicator under the k-th subsystem; Divide the status of each of the main influencing factors into several levels according to historical data, classify the historical data according to random combinations of the status levels, and construct a comparison database, wherein the comparison database contains a number of comparison data sets equal to the number of random combinations of status levels, the comparison data sets contain comparison data subsets equal to the number of monitoring indicators under the k-th subsystem, and the i-th comparison data subset contains data of the i-th monitoring indicator of the k-th subsystem under normal operating conditions; Collecting real-time data of the kth subsystem, the real-time data including real-time data of monitoring indicators and real-time data of main influencing factors corresponding to the real-time data of monitoring indicators when collecting the real-time data, and determining its status level according to the real-time data of main influencing factors; Extract the comparison data subset corresponding to the real-time data status level of the main influencing factors in the comparison database, calculate the outlier factor of the real-time data of the i-th monitoring indicator in the comparison data subset, and judge the abnormal situation of the real-time data of the i-th monitoring indicator based on the obtained outlier factor.

2. The data center power supply and distribution operation status monitoring method based on big data analysis according to claim 1 is characterized in that: Correlation coefficient The calculation method includes the following process: Step S21: Construct the data sequence E of the monitoring indicators and the data sequence F of the influencing factors of the indicators under the k-th subsystem, where E={E1, E2, ..., E m }; F = {F1, F2, ..., F n };E m is the dataset of the mth monitoring indicator; F n is the dataset of factors influencing the nth indicator; Step S22: Standardize the data in the data sequence E and the data sequence F using the following formula: Where, e iq Represented as the i-th data set E in the data sequence E i The qth data; for e iq Standardized data of f jq Represented as the jth data set F in the data sequence F j The qth data; f jq Normalized data; Q is the total amount of data; i = 1, 2, ..., m; j = 1, 2, ..., n; Step S23: Calculate the correlation coefficient ξ between each data point in the data sequence F and the data sequence E jq , the calculation formula is: Where ρ is the resolution coefficient, and its value range is (0,1); in, It is expressed as first taking when q=1,2,...,Q, Calculate the minimum value among the results, and then take the minimum value of all the calculated minimum values ​​when i = 1, 2, ..., m; It is expressed as first taking when q=1,2,...,Q, Calculate the maximum value among the results, and then take the maximum value of all the calculated maximum values ​​when i = 1, 2, ..., m; Step S24: average all calculated correlation coefficients to obtain the correlation coefficient between the i-th monitoring indicator and the j-th indicator influencing factor. The calculation formula is:

3. The data center power supply and distribution operation status monitoring method based on big data analysis according to claim 1 is characterized in that: The screening principle of the main influencing factors is: when the correlation coefficient between the i-th monitoring indicator and the j-th indicator influencing factor is When , the influencing factor of the jth indicator is determined to be the main influencing factor of the i-th monitoring indicator.

4. The data center power supply and distribution operation status monitoring method based on big data analysis according to claim 1 is characterized in that: The process of classifying the status of major influencing factors includes the following steps: Step S31: normalize the acquired main influencing factor data into a standard normal distribution, wherein the normalization formula is: In the formula, x is the data value of the main influencing factor; z is the standardized value of x; μ is the mean of the main influencing factor data; σ is the standard deviation of the main influencing factor data; Step S32: according to the number of status levels to be divided, use the standard normal distribution table to find the corresponding z score; Step S33: Convert the z score to the scale of the original data to obtain the classification threshold of the main influencing factor data. The conversion formula is: p =μ+z p ×σ; where t p is the p-th classification threshold; z p is the pth z score; Step S34: Classify the status of the main influencing factor data into p levels according to the determined classification threshold, and the classification rules are: When the normalized value of the main influencing factor data is in the interval (-∞, t1], it is the first state level; When the normalized value of the main influencing factor data is in the interval (t1, t2], it is the second state level; And so on, When the standardized value of the main influencing factor data is within the interval (t p , +∞), it is the p+1th state level.

5. The data center power supply and distribution operation status monitoring method based on big data analysis according to claim 1 is characterized in that: Outlier Factor LOF i The calculation process of (t) includes the following steps: Step S51: Assume that the real-time data of the i-th monitoring indicator and the r-th data point Df in the comparison data subset i r The kth data point with the smallest distance is Df i k , r=1,2,...,R, R is the total amount of data; Step S52: Calculate the rth data point Df i r k nearest neighbor distance The calculation formula is: Step S53: Calculate the difference between all other data points and the r-th data point Df in the comparison data subset respectively. i r The distance of the calculated distance value is less than the k nearest neighbor distance d k (Df i r ) as the rth data point Df i r k-distance neighborhood N k (Df i r ); Step S54: Compare k nearest neighbor distances d k (Df i r ) and k-distance neighborhood N k (Df i r )Medium distance data point Df i k The distance to the nearest k-th data point; Step S55: Take the maximum value of the two distance values ​​as the rth data point Df i r With the kth data point as Df i k The k-th reachable distance reach_dist k (Df i k , Df i r ); Step S56: Calculate the rth data point Df i r The local reachable density ρ k (Df i r ), where the calculation formula is: Step S57: According to the rth data point Df i r The local outlier factor LOF of the data point is calculated by the local reachability density k (Df i r ), the calculation formula is: Where, ρ k (Df i k ) is the kth data point Df i k The local reachable density is obtained by calculation.

6. The data center power supply and distribution operation status monitoring method based on big data analysis according to claim 1 is characterized in that: The rule for judging whether the real-time data of the i-th monitoring indicator is abnormal based on the outlier factor is: when the outlier factor of the real-time data of the monitoring indicator in the comparison data subset is greater than 2 or greater than the 95% quantile of the outlier factors of all data points in the comparison data subset, it is judged as an abnormal point.

7. A data center power supply and distribution operation status monitoring system based on big data analysis is characterized by: The system is used to implement the steps of the data center power supply and distribution operation status monitoring method based on big data analysis according to any one of claims 1 to 6, including: An indicator system construction module is used to obtain the topology of the power supply and distribution system of the data center to be monitored and to construct a monitoring indicator system reflecting the operating status of the subsystems of the power supply and distribution system; A data acquisition module is used to obtain historical data of each monitoring indicator and its influencing factors in the indicator system and real-time data of the kth subsystem, wherein the real-time data includes the real-time data of the monitoring indicators and the real-time data of the main influencing factors corresponding to the real-time data of the monitoring indicators; The data analysis module is used to perform correlation analysis on the historical data of the monitoring indicators and the influencing factors of the indicators, and obtain the correlation coefficient between the i-th monitoring indicator and the j-th indicator influencing factor under the k-th subsystem. Use the correlation coefficient to screen out the main influencing factors of the i-th monitoring indicator under the k-th subsystem; A data classification module is used to classify the status of each of the main influencing factors into several levels based on historical data; a database construction module, configured to classify historical data according to random combinations of status levels and construct a comparison database, wherein the comparison database contains a number of comparison data sets equal to the number of random combinations of status levels, the comparison data sets contain a number of comparison data subsets equal to the number of monitoring indicators under the k-th subsystem, and the i-th comparison data subset contains data of the i-th monitoring indicator of the k-th subsystem under normal operating conditions; The data comparison module is used to determine the status level of the real-time data of the main influencing factors, extract the comparison data subset corresponding to the status level of the real-time data of the main influencing factors in the comparison database, calculate the outlier factor of the real-time data of the i-th monitoring indicator in the comparison data subset, and judge the abnormal situation of the real-time data of the i-th monitoring indicator based on the obtained outlier factor.

8. The data center power supply and distribution operation status monitoring system based on big data analysis according to claim 7 is characterized in that: The system also includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, is capable of implementing the steps of the method for monitoring the power supply and distribution operation status of a data center based on big data analysis as described in any one of claims 1 to 6.

Images