Metering intelligent supervision system and method based on data center

Abstract

The present invention discloses a data center-based intelligent metering supervision system and method, which relates to the field of intelligent metering supervision technology. The system includes: a data acquisition module, a data preprocessing module, an interpolation compensation module, a drift correction module, a time synchronization module, a time-space alignment module, an adaptive adjustment module, and a fusion verification module. Through outlier detection, repair, and normalization processing, data anomalies and differences in metering between manufacturers and equipment are eliminated, thereby improving the level of data standardization. With the help of drift correction and distributed synchronization, a linear drift model is established and dynamically compensated to achieve high-precision synchronization of equipment time and solve the problem of time misalignment. The time scale is unified through interpolation, and a global metering value is generated by combining deviation weighted fusion to improve the accuracy of time-space alignment. Through parameter adaptive adjustment and coefficient of variation testing, parameters are dynamically optimized and intelligent early warning is provided. The output high-precision data supports energy consumption analysis and resource scheduling, providing a reliable basis for refined supervision decision-making.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent metering supervision, and specifically to a data center-based intelligent metering supervision system and method. Background Art

[0002] With the rapid development of cloud computing, big data, and artificial intelligence technologies, data centers, as the core carriers of information infrastructure, continue to grow in size and complexity. Data centers deploy a large number of metering devices from various manufacturers, including power meters, environmental sensors, and network traffic monitors. The metering data generated by these devices is a key basis for energy optimization, equipment operation and maintenance, and security supervision. However, the current field of data center metering supervision still faces the following prominent issues:

[0003] Each metering device in a data center relies on a local clock to generate timestamps. Due to factors such as hardware clock drift, temperature variations, and network latency, time deviations between devices are common. Time synchronization errors can distort spatiotemporal correlation analysis. For example, in power consumption tracing, if the metering data from air conditioning equipment and servers is time-misaligned, it will be impossible to accurately establish a correlation model between energy consumption and load, seriously affecting the scientific nature of regulatory decisions.

[0004] Existing monitoring systems often process multi-source data using simple weighting or direct splicing, lacking effective spatiotemporal alignment mechanisms. Due to differences in device installation locations, signal transmission delays, and measurement principles, even after initial time synchronization, metering data at the same point in time may still exhibit systematic deviations. For example, temperature and humidity sensors in different regions may experience measurement deviations due to environmental disturbances. Direct fusion of these sensors would result in a decrease in data credibility. Furthermore, they lack dynamic adaptive adjustment mechanisms. Fixed algorithm parameters make it difficult to ensure long-term stable fusion accuracy when device performance degrades or the environment changes.

[0005] Traditional data center metering and monitoring relies on manual inspections and regular calibration, which can lead to delayed responses and high costs. For one thing, identifying abnormal data relies on empirical thresholds, which can easily lead to missed or false detections. Furthermore, fixed equipment drift parameters cannot adapt to performance degradation over long periods of operation, resulting in unreasonable calibration cycles. Excessively frequent calibration increases operational costs, while excessively long calibration intervals lead to accumulated errors.

[0006] Therefore, there is an urgent need to build an intelligent supervision system with high-precision time synchronization and dynamic adaptive adjustment to improve the quality of data center metering data and supervision efficiency, and provide technical support for the construction and efficient operation and maintenance of green data centers. Summary of the Invention

[0007] (1) Technical problems solved

[0008] In response to the deficiencies of the existing technology, the present invention provides a data center-based metering intelligent supervision system and method, which solves the problems mentioned in the above background technology.

[0009] (2) Technical solution

[0010] To achieve the above objectives, the present invention is implemented through the following technical solutions: a data center-based metering intelligent supervision system, comprising:

[0011] The data collection module is used to collect the original measurement data of each manufacturer's equipment through the equipment interface of the data center. The original measurement data of each device includes the measurement value and time stamp;

[0012] Data preprocessing module, used to standardize the original measurement data, including outlier processing and data normalization;

[0013] Interpolation compensation module, used to convert non-uniform sampling data into a data sequence of uniform time scale through interpolation calculation, including sampling frequency analysis, interpolation time point generation and adaptive interpolation calculation;

[0014] The drift correction module is used to eliminate the deviation between the local clock of the device and the reference clock and perform unified calibration of the timestamp, including reference clock selection, drift parameter estimation and timestamp correction;

[0015] The time synchronization module is used to perform dynamic time synchronization between various devices, including synchronization node deployment, time message interaction, and dynamic compensation adjustment;

[0016] The spatiotemporal alignment module is used to perform high-precision alignment processing through deviation feedback and dynamic adjustment, including alignment deviation calculation, weight coefficient update, and fusion data generation;

[0017] Adaptive adjustment module, used to dynamically optimize algorithm parameters according to data characteristics and perform adaptive adjustment of algorithm parameters, including parameter monitoring, threshold adjustment and periodic update;

[0018] The fusion verification module is used to perform data fusion verification on the high-precision alignment processing results through consistency verification, including data grouping, statistical testing and result evaluation.

[0019] As a further solution of the present invention, the metering equipment in the data center is marked as M, and the raw data set generated during the collection period is marked as: D i ={(v i（1）， t i（1） ),(v i（2）， t i（2） ),……,(v i（ni）， t i（ni） )}, where i = 1, 2, ..., M;

[0020] Where: v i（j） Indicates the measurement value of the jth data of the i-th device; t i（j） Indicates the timestamp of the jth piece of data on the i-th device; ni indicates the number of valid data pieces on the i-th device during the collection period.

[0021] As a further solution of the present invention: the abnormal value processing method is as follows:

[0022] First, calculate the statistical parameters of the measurement value of the i-th device:

[0023] (Formula 1)

[0024] (Formula 2)

[0025] Among them: vp i represents the arithmetic mean of the measurement value of the i-th device; vb i Indicates the standard deviation of the measurement value of the i-th device; other symbols have the same meanings as before;

[0026] When the measurement value of the jth data satisfies |v i（j） −vp i ∣>3×vb i When , it is determined to be an abnormal value;

[0027] Then the abnormal values ​​are repaired by adjacent effective value interpolation method:

[0028] (Formula 3)

[0029] Where: v1 ij represents the measurement value of the jth data of the i-th device after repair, v i(j−1) represents the measurement value of the j-1th valid data of the i-th device; v i(j+1) Indicates the measurement value of the j+1th valid data of the i-th device.

[0030] As a further solution of the present invention: the data normalization method is as follows:

[0031] In order to eliminate the influence of the difference in measurement range of different equipment, the repaired data is normalized: (Formula 4)

[0032] Where: v2 i（j） v1 represents the normalized measurement value of the jth data item of the i-th device; i（min） =min{v1 i(1) ,v1 i(2) ,……,v1 i(ni)}, is the minimum value of the measurement value of the i-th device after repair; v1i（max） =max{v1 i(1) ,v1 i(2) ,……,v1 i(ni)}, is the maximum value of the measurement value of the i-th device after repair.

[0033] As a further solution of the present invention: the unified time scale data sequence conversion method is as follows:

[0034] Step B1, sampling frequency analysis:

[0035] pass: (Formula 5)

[0036] Calculate the average sampling period T of the i-th device i ;

[0037] Where: t i（ni） Indicates the timestamp of the last data of the i-th device; t i（1） Indicates the timestamp of the first data of the i-th device; ni indicates the number of valid data of the i-th device;

[0038] Step B2: Generate interpolation time points:

[0039] Extract the unified target sampling period T0 pre-set according to the data center supervision requirements and generate a globally unified interpolation time point sequence;

[0040] The interpolation time range is marked as [t start ,t end ], that is, the interpolation time point set is: T={t1,t2,……,t K};

[0041] Then t1=t start , t k =t k−1 +T0,t K ≤t end (Formula 6)

[0042] Where: t k represents the kth interpolation time point; K represents the total number of interpolation time points; t start and t end Respectively represent the start and end time of the interpolation time range;

[0043] Step B3: Adaptive interpolation calculation:

[0044] For each interpolation time point t k , the corresponding measurement value is calculated using linear interpolation;

[0045] For the i-th device, if there is t i（j） =t k, then directly use v2 i（j） As the interpolation result;

[0046] Otherwise, find the k Two adjacent timestamps t i（a） and t i（b） , and it satisfies t i（a） <t k <t i（b） , the corresponding normalized measurement value is v2 i（a） and v2 i（b） , the interpolation calculation formula is:

[0047] (Formula 7)

[0048] Among them: v3 i（k） Indicates the i-th device at the interpolation time point t k The interpolation result of T i（a） , t i（b） Respectively represent the number of devices i that are less than and greater than t k The most recent timestamp; v2 i（a） 、v2 i（b） They represent the normalized measurement values ​​of the corresponding timestamps respectively;

[0049] After the interpolation is completed, the interpolation data set of the i-th device is generated:

[0050] D1i={(v3 i（1）， t1)、(v3 i（2）， t2),……(v3 i（K）， t K )} (Formula 8).

[0051] As a further solution of the present invention: the unified calibration method is as follows:

[0052] Step C1: Select the reference clock:

[0053] Select a high-precision clock device in the data center as the reference clock and record it as device 0. Its timestamp is t 0（k） As a global time reference;

[0054] Step C2, drift parameter estimation:

[0055] Calculate the time deviation between the i-th device and the reference clock and establish a linear drift model;

[0056] During the calibration period, m groups of synchronous data are collected, and the pth group of data satisfies:

[0057] (Formula 9)

[0058] Where p = 1, 2, ..., m, Δti（p） Indicates the time deviation of device i at time p, in seconds;

[0059] The linear drift model expression is: (Formula 10)

[0060] Where: i（0） represents the initial time deviation of device i, in seconds; β i（1） Indicates the clock drift rate of device i, where a positive value indicates a fast clock.

[0061] Indicates the time difference between the pth moment and the calibration start moment, in seconds;

[0062] The least squares method is used to estimate the model parameters, and the objective function is:

[0063] (Formula 11)

[0064] Taking the partial derivative of the objective function and setting it to zero, we get the parameter solution formula:

[0065] (Formula 12)

[0066] (Formula 13);

[0067] Step C3: timestamp correction:

[0068] The timestamp of device i is corrected according to the estimated drift parameter. The correction formula is:

[0069] (Formula 14)

[0070] where t1 i（k） Indicates the timestamp of the kth data after correction on device i;

[0071] The corrected data set is updated to: D2 i ={(v3 i（k）， t1 i（k） )|k=1, 2, …K} (Formula 15).

[0072] As a further solution of the present invention: the dynamic time synchronization processing method is as follows:

[0073] Step S1: Synchronize node deployment:

[0074] A hierarchical deployment strategy is used to set up synchronization nodes, including 1 root node, N intermediate nodes and M leaf nodes, forming a tree-like communication structure between nodes;

[0075] Step S2: Time message interaction:

[0076] Defines the time synchronization message format, which includes the sending timestamp t s , receiving timestamp t r and node identification ID;

[0077] The root node periodically sends synchronization messages to the intermediate nodes. The interaction process is as follows:

[0078] The root node is at local time t s（1） Send synchronization message; intermediate node at local time t r（1） Receive message; intermediate node at local time t s（2） Returns a response message; the root node at local time t r（2） Receive a response message;

[0079] Calculate the time delay and offset between nodes:

[0080] (Formula 16)

[0081] (Formula 17)

[0082] Where: d represents the one-way delay of message transmission; θ represents the time offset of the intermediate node relative to the root node;

[0083] Step S3: Dynamic compensation adjustment:

[0084] The intermediate node adjusts the local clock according to the time offset θ. The adjustment formula is:

[0085] (Formula 18)

[0086] where t1 local Indicates the adjusted local time, t local Indicates the local time before adjustment;

[0087] The synchronization process between leaf nodes and intermediate nodes is the same as above.

[0088] As a further solution of the present invention: the high-precision alignment processing method is as follows:

[0089] Step D1, calculation of alignment deviation:

[0090] Calculate the deviation degree of each device data after time correction, and define the relative deviation of device i at time k as: (Formula 19)

[0091] in, represents the average measurement value of all devices at the kth moment;

[0092] Calculate the standard deviation of the deviations to assess the overall alignment effect:

[0093] (Formula 20)

[0094] in represents the average relative deviation at the kth moment;

[0095] Step D2, weight coefficient update:

[0096] Dynamically adjust the weight of each device data based on the deviation. The smaller the deviation, the greater the weight:

[0097] (Formula 21)

[0098] where w i（k） Represents the weight coefficient of the i-th device at the k-th moment, satisfying ;

[0099] Step D3: Fusion data generation:

[0100] use (Formula 22)

[0101] Calculate the fusion measurement value v at the kth moment k final .

[0102] As a further solution of the present invention: statistical analysis of the weight coefficient of each device is performed daily, and a calibration reminder is issued for devices whose weight is less than 0.05 for 24 consecutive hours, triggering a recalibration process;

[0103] At the same time, a global alignment check is performed once a month. k When the mean of is greater than 5%, the drift parameter estimation step is re-executed to update the model parameters.

[0104] As a further solution of the present invention, the adaptive adjustment method is as follows:

[0105] Step E1: Parameter monitoring:

[0106] Real-time monitoring of the operating status of key algorithm parameters, including interpolation error, drift rate changes, and synchronization deviation: (Formula 23)

[0107] where e interp represents the interpolation error (unit: original measurement unit), v i（k） true Indicates the real value obtained by high-precision equipment;

[0108] Step E2: Threshold adjustment:

[0109] When the interpolation error e interp Exceeding the threshold Th interpWhen , reduce the target sampling period T0:

[0110] (Formula 24)

[0111] Where T10 represents the adjusted sampling period, ensuring that T10 ≥ 0.1 seconds;

[0112] Step E3: Periodic update:

[0113] The drift parameters are re-estimated every 7 days, and the update formula is:

[0114] (Formula 25)

[0115] (Formula 26)

[0116] Where: i（0） new , β i（1） new represents the updated parameters; β i（0） old , β i（1） old represents the parameters before updating; β i（0） est , β i（1） est Represents the newly estimated parameters; α is the preset smoothing coefficient.

[0117] As a further solution of the present invention, the data fusion verification method is as follows:

[0118] Step F1, data grouping:

[0119] The spatiotemporally aligned data are grouped by device type to form a validation dataset.

[0120] V g ={v g（1）， v g（2）， ……,v g（n）}, where g represents the device type group and n represents the number of devices in the group;

[0121] Step F2, consistency check:

[0122] Calculate the coefficient of variation for each set of data: (Formula 27)

[0123] in: represents the average value of the data within the group; Represents the standard deviation of the data within the group; CV g represents the coefficient of variation, and it is dimensionless;

[0124] Step F3, result evaluation:

[0125] When CV g When the error rate is less than 5%, the data consistency of this group is determined to be qualified; otherwise, the abnormality troubleshooting process is triggered;

[0126] For equipment that fails inspection three times in a row, a calibration work order is automatically generated to prompt maintenance personnel to perform hardware repairs.

[0127] A data center-based intelligent metering supervision method is implemented by a data center-based intelligent metering supervision system, and includes the following steps:

[0128] Step 1: Data Collection:

[0129] Collect raw metering data from various manufacturers' devices through the device interface of the data center. The raw metering data of each device includes the metering value and timestamp.

[0130] Step 2: Data preprocessing:

[0131] Standardize the original measurement data;

[0132] Step 3: Adaptive interpolation compensation:

[0133] The non-uniformly sampled data are converted into a data sequence of uniform time scale through interpolation calculation;

[0134] Step 4: Clock drift correction:

[0135] Eliminate the deviation between the local clock of the device and the reference clock and calibrate the timestamp uniformly;

[0136] Step 5: Distributed time synchronization protocol:

[0137] Perform dynamic time synchronization between various devices;

[0138] Step 6: Time and space alignment optimization:

[0139] High-precision alignment processing through deviation feedback and dynamic adjustment;

[0140] Step 7: Adaptive adjustment of algorithm parameters:

[0141] Dynamically optimize algorithm parameters according to data characteristics and perform adaptive adjustment of algorithm parameters;

[0142] Step 8: Data fusion verification:

[0143] The data fusion verification of the high-precision alignment processing results is carried out through consistency testing.

[0144] (3) Beneficial effects

[0145] The present invention provides a data center-based intelligent metering supervision system and method. Compared with the existing technology, it has the following advantages:

[0146] Through the outlier detection mechanism based on 3 times the standard deviation, abnormal fluctuations in the original data can be efficiently identified and repaired using the adjacent effective value interpolation method (Formula 3), avoiding the interference of abnormal data on subsequent analysis and ensuring the accuracy and continuity of measurement values.

[0147] Through data normalization processing (Formula 4), the impact of differences in the measurement range of equipment from different manufacturers is eliminated, and the measurement values ​​of various types of equipment are mapped to a unified scale. This lays the foundation for cross-device data comparison and integration, and solves the problem of "incomparable" data of equipment from multiple manufacturers.

[0148] The drift correction module establishes a linear drift model (Formula 10), uses the least squares method to estimate the initial deviation and drift rate (Formulas 12-13), and dynamically corrects the timestamp based on the model (Formula 14). This effectively eliminates the accumulated deviation between the device's local clock and the reference clock, ensuring the global consistency of the timestamp.

[0149] A tree-shaped hierarchical synchronization architecture (root node-intermediate node-leaf node) is adopted to calculate the transmission delay and time offset (Formula 16-17) through time message interaction, and dynamic compensation adjustment (Formula 18) is performed to achieve microsecond-level time synchronization between devices, solving the "time misalignment" problem of multi-node data.

[0150] The interpolation compensation module analyzes the sampling frequency and generates a unified time point (Formula 6). Adaptive linear interpolation (Formula 7) is then used to convert the non-uniformly sampled data into a unified time series. This ensures that the data of devices with different sampling frequencies are aligned on the time axis to meet the requirements of timing analysis.

[0151] Based on relative deviation calculation (Formula 19) and dynamic weight update (Formula 21), device data is weightedly fused to generate a global measurement value (Formula 22). Devices with smaller deviations have higher weights, which improves the representativeness and reliability of the fusion results and reduces the impact of single-point device errors on global judgment.

[0152] The adaptive adjustment module monitors key indicators such as interpolation error and drift rate changes (Formula 23) and dynamically optimizes the sampling period (Formula 24) and drift model parameters (Formulas 25-26). This enables the system to autonomously adapt to changes in device status based on data characteristics, avoiding the loss of accuracy caused by fixed parameters.

[0153] By integrating the coefficient of variation test (Formula 27) of the verification module, data consistency is quantitatively evaluated, calibration prompts are automatically triggered for continuously low-weight equipment, and maintenance work orders are generated for equipment that fails multiple inspections, achieving a shift from "passive maintenance" to "active warning" and reducing operation and maintenance costs.

[0154] Through full-process data processing and integration, the high-precision and high-consistency metering data output by the system can be directly used in scenarios such as data center energy consumption analysis, equipment performance evaluation, and resource scheduling optimization, providing managers with objective and reliable data basis, helping to improve the energy efficiency and operational efficiency of data centers and achieve the goal of "data-driven" smart supervision.

[0155] In summary, the present invention effectively solves the core problems of "fragmentation", "heterogeneity" and "asynchrony" of metering data of multi-vendor equipment through a full-chain technical solution of standardized processing, spatiotemporal synchronization, adaptive optimization and intelligent verification, significantly improving the accuracy, efficiency and intelligence level of metering supervision in data centers, and has important practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0156] Figure 1 This is a system block diagram of the metering intelligent supervision system based on the data center of the present invention.

[0157] Figure 2 It is a flow chart of the metering intelligent supervision method based on data center of the present invention.

[0158] Figure 3 This is a system block diagram of electronic equipment in the data center-based metering intelligent monitoring system of the present invention. DETAILED DESCRIPTION

[0159] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0160] See also Figure 1 and Figure 2 As shown, the embodiment of the present invention provides the following technical solutions:

[0161] As embodiment 1 of the present invention:

[0162] The present invention is a data center-based metering intelligent supervision system, comprising:

[0163] The data collection module is used to collect metering data from various manufacturers' devices through the device interface of the data center. The raw data of each device contains the metering value and timestamp;

[0164] Mark the metering equipment in the data center as M units;

[0165] The original data set generated by the i-th device during the collection period is: D i ={(v i（1）， t i（1） ),(v i（2）， t i（2） ),……,(v i（ni）， t i（ni） )}, where i = 1, 2, ..., M;

[0166] Where: v i（j） represents the measurement value of the jth piece of data of the i-th device. In this embodiment, the physical quantities are current, voltage, and power; t i（j） The timestamp of the jth piece of data on the i-th device. In this embodiment, its unit is seconds, recording the local time when the data is generated; ni represents the number of valid data pieces of the i-th device within the collection period;

[0167] The data preprocessing module is used to standardize the original measurement data and provide basic data support for subsequent spatiotemporal alignment, specifically including the steps corresponding to data normalization;

[0168] In order to eliminate the influence of the difference in measurement range of different equipment, the repaired data is normalized: (Formula 4)

[0169] Where: v2 i（j） v1 represents the normalized measurement value of the jth data item of the i-th device; i（min） =min{v1 i(1) ,v1 i(2) ,……,v1 i(ni)}, is the minimum value of the measurement value of the i-th device after repair; v1 i（max） =max{v1 i(1) ,v1 i(2) ,……,v1 i(ni)}, is the maximum value of the measurement value of the i-th device after repair;

[0170] The drift correction module is used to eliminate the deviation between the device's local clock and the reference clock, achieving unified timestamp calibration. It includes three steps: reference clock selection, drift parameter estimation, and timestamp correction:

[0171] Step C1: Select the reference clock:

[0172] Select a high-precision clock device in the data center as the reference clock and record it as device 0. Its timestamp is t 0（k） As a global time reference;

[0173] The reference clock must meet the accuracy requirement of a daily drift rate of less than 1 microsecond to ensure the reliability of the time reference;

[0174] Step C2, drift parameter estimation:

[0175] Calculate the time deviation between device i and the reference clock and establish a linear drift model. Collect m groups of synchronization data during the calibration period, and the pth group of data satisfies:

[0176] (Formula 9)

[0177] Where p = 1, 2, ..., m, Δt i（p） Indicates the time deviation of device i at time p, in seconds;

[0178] The linear drift model expression is: (Formula 10)

[0179] Where: i（0） represents the initial time deviation of device i, in seconds; β i（1） Indicates the clock drift rate of device i, where a positive value indicates a fast clock.

[0180] Indicates the time difference between the pth moment and the calibration start moment, in seconds;

[0181] The least squares method is used to estimate the model parameters, and the objective function is:

[0182] (Formula 11)

[0183] Taking the partial derivative of the objective function and setting it to zero, we get the parameter solution formula:

[0184] (Formula 12)

[0185] (Formula 13);

[0186] The meanings of the symbols are the same as before;

[0187] Step C3: timestamp correction:

[0188] The timestamp of device i is corrected according to the estimated drift parameter. The correction formula is:

[0189] (Formula 14)

[0190] where t1 i（k） Indicates the timestamp of the kth data of device i after correction, in seconds;

[0191] The corrected data set is updated to: D2 i ={(v3 i（k）， t1 i（k） )|k=1, 2, ...K} (Formula 15);

[0192] The time synchronization module is used to achieve dynamic time synchronization between devices and maintain long-term time consistency. It includes three steps: synchronization node deployment, time message exchange, and dynamic compensation adjustment.

[0193] Step S1: Synchronize node deployment:

[0194] A hierarchical deployment strategy is used to set up synchronization nodes, including 1 root node, N intermediate nodes, and M leaf nodes. The root node is the reference clock device, and the leaf node is the common metering device.

[0195] A tree-like communication structure is formed between nodes. The communication delay from root node to intermediate node to leaf node must be less than 10 milliseconds to ensure the real-time nature of time messages.

[0196] Step S2: Time message interaction:

[0197] Define the time synchronization message format including the sending timestamp t s , receiving timestamp t r And node identification ID. The root node periodically sends synchronization messages to the intermediate nodes. The interaction process is as follows:

[0198] The root node is at local time t s（1） Send synchronization message; intermediate node at local time t r（1） Receive message; intermediate node at local time t s（2） Returns a response message; the root node at local time t r（2） Receive a response message;

[0199] Calculate the time delay and offset between nodes:

[0200] (Formula 16)

[0201] (Formula 17)

[0202] Where: d represents the one-way delay of message transmission, and its unit is seconds; θ represents the time offset of the intermediate node relative to the root node, and its unit is seconds;

[0203] Step S3: Dynamic compensation adjustment:

[0204] The intermediate node adjusts the local clock according to the time offset θ. The adjustment formula is:

[0205] (Formula 18)

[0206] where t1 local Indicates the adjusted local time, t local Indicates the local time before adjustment;

[0207] The synchronization process between leaf nodes and intermediate nodes is the same as above. Multi-level synchronization is used to ensure that the time deviation of all devices is controlled within 50 microseconds.

[0208] The synchronization period is dynamically adjusted based on the device drift characteristics. The synchronization period for devices with a drift rate greater than 1e-6 is set to 1 minute, and for other devices it is set to 5 minutes.

[0209] The spatiotemporal alignment module is used to further improve cross-device data consistency and achieve high-precision alignment through deviation feedback and dynamic adjustment. It includes three steps: alignment deviation calculation, weight coefficient update, and fused data generation.

[0210] Step D1, calculation of alignment deviation:

[0211] Calculate the deviation degree of each device data after time correction, and define the relative deviation of device i at time k as: (Formula 19)

[0212] in, represents the average measurement value of all devices at the kth moment;

[0213] Calculate the standard deviation of the deviations to assess the overall alignment effect:

[0214] (Formula 20)

[0215] in represents the average relative deviation at the kth moment;

[0216] Step D2, weight coefficient update:

[0217] Dynamically adjust the weight of each device data based on the deviation. The smaller the deviation, the greater the weight:

[0218] (Formula 21)

[0219] where w i（k） Represents the weight coefficient of the i-th device at the k-th moment, satisfying ;

[0220] Step D3: Fusion data generation:

[0221] The weighted average method is used to generate the final fusion data:

[0222] (Formula 22)

[0223] Among them, vk final Represents the fusion measurement value at the kth moment.

[0224] Perform statistical analysis on the weight coefficient of each device every day, and issue a calibration reminder to devices with a weight below 0.05 for 24 consecutive hours, triggering the recalibration process;

[0225] Perform a global alignment check once a month. k When the mean of is greater than 5%, the drift parameter estimation step is re-executed to update the model parameters;

[0226] Adaptive adjustment module for algorithm parameters The adaptive adjustment module is used to dynamically optimize algorithm parameters according to data characteristics to ensure the alignment effect under different working conditions, including three steps: parameter monitoring, threshold adjustment and periodic update.

[0227] Step E1: Parameter monitoring:

[0228] Real-time monitoring of the operating status of key algorithm parameters, including interpolation error, drift rate changes, and synchronization deviation: (Formula 23)

[0229] where e interp represents the interpolation error (unit: original measurement unit), v i（k） true Indicates the real value obtained by high-precision equipment;

[0230] Step E2: Threshold adjustment:

[0231] When the interpolation error e interp Exceeding the threshold Th interp When , reduce the target sampling period T0:

[0232] (Formula 24)

[0233] Where T10 represents the adjusted sampling period, ensuring that T10 ≥ 0.1 seconds;

[0234] Step E3: Periodic update:

[0235] The drift parameters are re-estimated every 7 days, and the update formula is:

[0236] (Formula 25)

[0237] (Formula 26)

[0238] Where: i（0） new , β i（1） new represents the updated parameters; βi（0） old , β i（1） old represents the parameters before updating; β i（0） est , β i（1） est represents the newly estimated parameter; α is the smoothing coefficient, which is 0.3;

[0239] Through adaptive parameter adjustment, the algorithm can maintain stable spatiotemporal alignment accuracy under conditions such as equipment replacement and load changes, ensuring that the cross-device data fusion deviation is controlled within 3%;

[0240] Example 1 establishes a basic framework for metering data supervision through core modules including data normalization, drift correction, time synchronization, spatiotemporal alignment, and adaptive adjustment. The primary contribution of this example is that normalization eliminates the impact of differences in metering ranges across devices, providing a unified data foundation for subsequent fusion analysis. By employing a linear drift model and least squares parameter estimation, combined with a hierarchical tree-based synchronization structure, inter-device time deviation is controlled within 50 microseconds, significantly improving time synchronization accuracy. The spatiotemporal alignment module dynamically adjusts device weights (the smaller the deviation, the greater the weight), effectively reducing the impact of abnormal devices on fusion results and keeping cross-device data fusion deviation within 3%. The synchronization period is dynamically adjusted based on device drift characteristics (devices with a drift rate greater than 1e-6 are synchronized for 1 minute). Smoothly updating drift parameters ensures system stability despite changing operating conditions.

[0241] As the second embodiment of the present invention:

[0242] In the specific implementation of this application, compared with Example 1, the technical solution of this embodiment differs from that of Example 1 only in that, in this embodiment, the normalization process further includes steps corresponding to outlier processing:

[0243] Use threshold method to identify and process outliers in raw data;

[0244] First, calculate the statistical parameters of the measurement value of the i-th device:

[0245] (Formula 1)

[0246] (Formula 2)

[0247] Among them: vp i represents the arithmetic mean of the measurement value of the i-th device; vb i Indicates the standard deviation of the measurement value of the i-th device; other symbols have the same meanings as before;

[0248] When the measurement value of the jth data satisfies |vi（j） −vp i ∣>3vb i When , it is determined to be an abnormal value;

[0249] Use adjacent valid value interpolation method to repair outliers:

[0250] (Formula 3)

[0251] Where: v1 ij represents the measurement value of the jth data of the i-th device after repair, v i(j−1) represents the measurement value of the j-1th valid data of the i-th device; v i(j+1) Indicates the measurement value of the j+1th valid data of the i-th device;

[0252] The interpolation compensation module is used to solve the problem of sampling frequency differences between different devices. It converts non-uniformly sampled data into a data sequence with a uniform time scale through interpolation calculation. It includes three steps: sampling frequency analysis, interpolation time point generation, and adaptive interpolation calculation.

[0253] Step B1, sampling frequency analysis:

[0254] Calculate the actual sampling period of each device and determine the sampling characteristics of the device;

[0255] The formula for calculating the average sampling period of the i-th device is:

[0256] (Formula 5)

[0257] Where: Ti represents the average sampling period of the i-th device, in seconds; t i（ni） Indicates the timestamp of the last data of the i-th device; t i（1） Indicates the timestamp of the first data of the i-th device; ni indicates the number of valid data of the i-th device;

[0258] Step B2: Generate interpolation time points:

[0259] A unified target sampling period T0 is set according to the data center supervision requirements. In this embodiment, for example, 1 second, 5 seconds, etc., to generate a globally unified interpolation time point sequence;

[0260] Assume the interpolation time range is [t start ,t end ], then the interpolation time point set is: T={t1,t2,……,t K};

[0261] Where: t1 = t start , t k =t k−1 +T0,tK ≤t end (Formula 6)

[0262] Where: t k represents the kth interpolation time point, in seconds; K represents the total number of interpolation time points; t start and t end Respectively represent the start and end time of the interpolation time range;

[0263] Step B3: Adaptive interpolation calculation:

[0264] For each interpolation time point t k , use linear interpolation to calculate the corresponding measurement value. For the i-th device, if there is t i（j） =t k , then directly use v2 i（j） as the interpolation result; otherwise, find the value corresponding to t k Two adjacent timestamps t i（a） and t i（b） , and it satisfies t i（a） <t k <t i（b） , the corresponding normalized measurement value is v2 i（a） and v2 i（b） , the interpolation calculation formula is:

[0265] (Formula 7)

[0266] Among them: v3 i（k） Indicates the i-th device at the interpolation time point t k The interpolation result of

[0267] T i（a） , t i（b） Respectively represent the latest timestamp less than and greater than tk in the i-th device; v2 i（a） 、v2 i（b） They represent the normalized measurement values ​​of the corresponding timestamps respectively;

[0268] After the interpolation is completed, the interpolation data set of the i-th device is generated:

[0269] D1i={(v3 i（1）， t1),(v3 i（2）， t2),……,(v3 i（K）， t K )} (Formula 8).

[0270] Example 2 adds an outlier processing and interpolation compensation module on the basis of Example 1, further improving data quality and processing capabilities. Its main advantages are: accurately identifying outliers through the 3σ principle and repairing them using the adjacent effective value interpolation method, effectively avoiding the interference of abnormal data on the analysis results and improving data reliability. The interpolation compensation module calculates the average sampling period of the device, generates a sequence of interpolation points with a unified time scale, and uses linear interpolation to achieve data alignment, solving the problem of difficult data comparison caused by differences in sampling frequencies of different devices. It supports setting the target sampling period according to regulatory requirements, enabling the system to adapt to application scenarios with different accuracy requirements, enhancing the flexibility and applicability of the solution.

[0271] As the third embodiment of the present invention:

[0272] In the specific implementation of this application, compared with Example 1 and Example 2, the technical solution of this embodiment is to combine the solutions of Example 1 and Example 2. The technical solution of this embodiment differs from that of Example 1 and Example 2 only in that this embodiment further includes:

[0273] The fusion verification module is used to verify the actual effect of the spatiotemporal alignment algorithm and ensure data quality through consistency testing, including three steps: data grouping, statistical testing, and result evaluation;

[0274] Step F1, data grouping:

[0275] The data that have been processed by time and space alignment are grouped by device type. Each group contains measurement data of at least three devices of the same type to form a validation data set.

[0276] V g ={v g（1）， v g（2）， ……,v g（n）}, where g represents the device type group and n represents the number of devices in the group;

[0277] Step F2, consistency check:

[0278] Calculate the coefficient of variation for each set of data: (Formula 27)

[0279] in: represents the average value of the data within the group; Represents the standard deviation of the data within the group; CV g represents the coefficient of variation (dimensionless);

[0280] Step F3, result evaluation:

[0281] When CV g When the error rate is less than 5%, the data consistency of this group is determined to be qualified; otherwise, the abnormality troubleshooting process is triggered;

[0282] For equipment that fails inspection three times in a row, a calibration work order is automatically generated to prompt maintenance personnel to perform hardware repairs;

[0283] Example 3 combines the functions of Example 1 and Example 2 and adds a fusion verification module to form a complete "acquisition-processing-verification" closed loop. Its unique value lies in: through a full chain of processing including outlier processing, normalization, time synchronization, interpolation compensation, and spatiotemporal alignment, it ensures the accuracy and consistency of data at all stages. The fusion verification module evaluates the consistency of data from the same type of equipment using the coefficient of variation (CVg). When CVg is less than 5%, it is judged to be qualified, providing a quantitative assessment standard for data quality. A calibration work order is automatically generated for equipment that fails three consecutive inspections, enabling the rapid location and repair of abnormal equipment, forming a closed-loop management mechanism for data quality assurance.

[0284] As the fourth embodiment of the present invention:

[0285] When the present application is implemented, compared with Example 1, Example 2 and Example 3, the technical solution of this embodiment is to combine and implement the solutions of the above-mentioned Example 1, Example 2 and Example 3.

[0286] Example 4 integrates all the functions of the first three examples to build a complete intelligent metering supervision system. Its comprehensive advantages are: covering the entire process of outlier processing, data normalization, time synchronization, interpolation compensation, spatiotemporal alignment, adaptive adjustment and fusion verification, and having the ability to process complex heterogeneous metering data. By real-time monitoring of parameters such as interpolation error and drift rate changes, the sampling period and model parameters are dynamically adjusted, so that the system can still maintain high-precision alignment in scenarios such as equipment replacement and load changes. From outlier repair, time synchronization to final data consistency verification, a multi-level data quality assurance system is formed to ensure that the deviation of fused data is controlled within 3%, providing reliable support for data center metering supervision.

[0287] The present invention also provides a data center-based intelligent metering supervision method, which is implemented by a data center-based intelligent metering supervision system. The method includes the following steps:

[0288] Step 1: Data Collection:

[0289] Collect raw metering data from various manufacturers' devices through the device interface of the data center. The raw metering data of each device includes the metering value and timestamp.

[0290] Step 2: Data preprocessing:

[0291] Standardize the original measurement data;

[0292] Step 3: Adaptive interpolation compensation:

[0293] The non-uniformly sampled data are converted into a data sequence of uniform time scale through interpolation calculation;

[0294] Step 4: Clock drift correction:

[0295] Eliminate the deviation between the local clock of the device and the reference clock and calibrate the timestamp uniformly;

[0296] Step 5: Distributed time synchronization protocol:

[0297] Perform dynamic time synchronization between various devices;

[0298] Step 6: Time and space alignment optimization:

[0299] High-precision alignment processing through deviation feedback and dynamic adjustment;

[0300] Step 7: Adaptive adjustment of algorithm parameters:

[0301] Dynamically optimize algorithm parameters according to data characteristics and perform adaptive adjustment of algorithm parameters;

[0302] Step 8: Data fusion verification:

[0303] The data fusion verification of the high-precision alignment processing results is carried out through consistency testing.

[0304] It should be stated that all user data collected in this application is collected with the user's consent and authorization, and the use of user data is legal and compliant, and the use and processing of user data complies with the relevant laws, regulations and standards of the relevant regions.

[0305] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0306] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0307] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

[0308] Corresponding to the aforementioned application function implementation method embodiments, the present application also provides electronic devices and corresponding embodiments.

[0309] See also Figure 3 , the electronic device includes a memory and a processor.

[0310] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0311] Memory can include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage. ROM can store static data or instructions required by the processor or other modules of the computer. Permanent storage can be a read-write memory device. Permanent storage can be a non-volatile storage device that retains stored instructions and data even when the computer is powered off. In some embodiments, the permanent storage device uses a mass storage device (such as a magnetic or optical disk, or flash memory). In other embodiments, the permanent storage device can be a removable storage device (such as a floppy disk or optical drive). System memory can be a read-write memory device or a volatile read-write memory device, such as dynamic random access memory. System memory can store some or all instructions and data required by the processor during operation. Furthermore, memory can include any combination of computer-readable storage media, including various types of semiconductor memory chips (DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), as well as magnetic disks and / or optical disks. In some embodiments, the memory may include a readable and / or writable removable storage device, such as a compact disc (CD), a read-only digital versatile disc (e.g., DVD-ROM, dual-layer DVD-ROM), a read-only Blu-ray disc, an ultra-density optical disc, a flash memory card (e.g., SD card, minSD card, Micro-SD card, etc.), a magnetic floppy disk, etc. Computer-readable storage media do not include carrier waves and transient electronic signals transmitted wirelessly or by wire.

[0312] The memory stores executable codes, which, when processed by the processor, can enable the processor to execute part or all of the above-mentioned methods.

[0313] The scheme of the present application has been described in detail above with reference to the accompanying drawings. In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. Those skilled in the art should also be aware that the actions and modules involved in the description are not necessarily required for this application. In addition, it is understood that the steps in the method of the embodiment of the present application can be adjusted in order, combined, and deleted according to actual needs, and the modules in the device of the embodiment of the present application can be combined, divided, and deleted according to actual needs.

[0314] In addition, the method according to the present application may also be implemented as a computer program or a computer program product, which includes computer program code instructions for executing some or all of the steps in the above method of the present application.

[0315] Alternatively, the present application can also be implemented as a non-transitory machine-readable storage medium (or computer-readable storage medium, or machine-readable storage medium) on which executable code (or computer program, or computer instruction code) is stored. When the executable code (or computer program, or computer instruction code) is executed by a processor of an electronic device (or electronic device, server, etc.), the processor executes part or all of the steps of the above-mentioned method according to the present application.

[0316] Those skilled in the art will further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the application herein may be implemented as electronic hardware, computer software, or combinations of both.

[0317] The flow charts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the system and method according to multiple embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a part for a module, program segment or code, and the part for the module, program segment or code comprises one or more executable instructions for realizing the logical function of the specification. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous boxes can actually be executed substantially in parallel, and they can sometimes also be executed in the opposite order, depending on the functions involved.

[0318] It should also be noted that each box in the block diagrams and / or flowcharts, and combinations of boxes in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0319] The embodiments of the present application have been described above. The above description is illustrative and not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or improvements to the technology in the market, or to enable other persons skilled in the art to understand the embodiments disclosed herein.

Claims

1. The metering intelligent supervision system based on the data center is characterized by: include: The data collection module is used to collect the original metering data of each manufacturer's equipment through the equipment interface of the data center. The original metering data of each device includes the metering value and timestamp. The metering devices in the data center are marked as M, and the original data set generated during the collection period is marked as: D i ={(v i（1）， t i（1） ),(v i（2）， t i（2） ),……,(v i（ni）， t i（ni） )}, where i = 1, 2, ... M; where: v i（j） Indicates the measurement value of the jth data of the i-th device; t i（j） Indicates the timestamp of the jth data item of the i-th device; ni indicates the number of valid data items of the i-th device within the collection period; The data preprocessing module is used to standardize the original measurement data; the standardization processing method is as follows: Step A1: Outlier processing: First, calculate the statistical parameters of the measurement value of the i-th device: (Formula 1); (Formula 2); Among them: vp i represents the arithmetic mean of the measurement value of the i-th device; vb i Indicates the standard deviation of the measurement value of the i-th device; other symbols have the same meanings as before; When the measurement value of the jth data satisfies |v i（j） −vp i ∣>3×vb i When , it is determined to be an abnormal value; Then the abnormal values ​​are repaired by adjacent effective value interpolation method: (Formula 3); Where: v1 ij represents the measurement value of the jth data of the i-th device after repair, v i(j−1) represents the measurement value of the j-1th valid data of the i-th device; v i(j+1) Indicates the measurement value of the j+1th valid data of the i-th device; Step A2: Data normalization: In order to eliminate the influence of the difference in measurement range of different equipment, the repaired data is normalized: (Formula 4); Where: v2 i（j） v1 represents the normalized measurement value of the jth data item of the i-th device; i（min） =min{v1 i(1) ,v1 i(2) ,……,v1 i(ni) }, is the minimum value of the measurement value of the i-th device after repair; v1 i（max） =max{v1 i(1) ,v1 i(2) ,……,v1 i(ni) }, is the maximum value of the measurement value of the i-th device after repair; An interpolation compensation module is used to convert non-uniform sampling data into a data sequence of uniform time scale through interpolation calculation; The unified time scale data series conversion method is as follows: Step B1: (Formula 5); Calculate the average sampling period T of the i-th device i ; Where: t i（ni） Indicates the timestamp of the last data of the i-th device; t i（1） Indicates the timestamp of the first data of the i-th device; ni indicates the number of valid data of the i-th device; Step B2: extracting a unified target sampling period T0 pre-set according to the data center supervision requirements, and generating a globally unified interpolation time point sequence; The interpolation time range is marked as [t start ,t end ], that is, the interpolation time point set is: T={t1,t2,……,t K }; Then t1=t start , t k =t k−1 +T0,t K ≤t end (Formula 6); Where: t k represents the kth interpolation time point; K represents the total number of interpolation time points; t start and t end Respectively represent the start and end time of the interpolation time range; Step B3: For each interpolation time point t k , the corresponding measurement value is calculated using linear interpolation; For the i-th device, if there is t i（j） =t k , then directly use v2 i（j） As the interpolation result; Otherwise, find the k Two adjacent timestamps t i（a） and t i（b） , and it satisfies t i（a） <t k <t i（b） , the corresponding normalized measurement value is v2 i（a） and v2 i（b） , the interpolation calculation formula is: (Formula 7); Among them: v3 i（k） Indicates the i-th device at the interpolation time point t k The interpolation result of T i（a） , t i（b） Respectively represent the number of devices i that are less than and greater than t k The most recent timestamp; v2 i（a） 、v2 i（b） They represent the normalized measurement values ​​of the corresponding timestamps respectively; After the interpolation is completed, the interpolation data set of the i-th device is generated: D1i = {(v3 i（1）， t1), (v3 i（2）， t2), …… (v3 i（K）， t K )} (Formula 8) Drift correction module, used to eliminate the deviation between the device's local clock and the reference clock and perform unified calibration of timestamps; Time synchronization module, used to perform dynamic time synchronization between various devices; Spatiotemporal alignment module, used for high-precision alignment processing through deviation feedback and dynamic adjustment; Adaptive adjustment module, used to dynamically optimize algorithm parameters according to data characteristics and perform adaptive adjustment of algorithm parameters; The fusion verification module is used to verify the data fusion of the high-precision alignment processing results through consistency verification.

2. The data center-based intelligent metering supervision system according to claim 1 is characterized in that: The unified calibration method is as follows: Step C1: Select a high-precision clock device in the data center as the reference clock and record it as device 0. Its timestamp is t 0（k） As a global time reference; Step C2: Calculate the time deviation between the i-th device and the reference clock and establish a linear drift model; During the calibration period, m groups of synchronous data are collected, and the pth group of data satisfies: (Formula 9); Where p = 1, 2, ..., m, Δt i（p） Indicates the time deviation of device i at time p, in seconds; The linear drift model expression is: (Formula 10); Where: i（0） represents the initial time deviation of device i, in seconds; β i（1） Indicates the clock drift rate of device i, where a positive value indicates a fast clock. Indicates the time difference between the pth moment and the calibration start moment, in seconds; The least squares method is used to estimate the model parameters, and the objective function is: (Formula 11); Taking the partial derivative of the objective function and setting it to zero, we get the parameter solution formula: (Formula 12); (Formula 13); Step C3: Correct the timestamp of device i according to the estimated drift parameter. The correction formula is: (Formula 14); where t1 i（k） Indicates the timestamp of the kth data after correction on device i; The corrected data set is updated to: D2 i ={(v3 i（k）， t1 i（k） )|k=1, 2, …K} (Formula 15).

3. The data center-based intelligent metering supervision system according to claim 2 is characterized in that: Dynamic time synchronization is handled as follows: Step S1: Use a hierarchical deployment strategy to set up synchronization nodes, including 1 root node, N intermediate nodes, and M leaf nodes, forming a tree-like communication structure between nodes; Step S2: define the time synchronization message format, which includes the sending timestamp t s , receiving timestamp t r and node identification ID; The root node periodically sends synchronization messages to the intermediate nodes. The interaction process is as follows: The root node is at local time t s（1） Send synchronization message; intermediate node at local time t r（1） Receive message; intermediate node at local time t s（2） Returns a response message; the root node at local time t r（2） Receive a response message; Calculate the time delay and offset between nodes: (Formula 16); (Formula 17); Where: d represents the one-way delay of message transmission; θ represents the time offset of the intermediate node relative to the root node; Step S3: The intermediate node adjusts the local clock according to the time offset θ. The adjustment formula is: (Formula 18); where t1 local Indicates the adjusted local time, t local Indicates the local time before adjustment; The synchronization process between leaf nodes and intermediate nodes is the same as above.

4. The data center-based intelligent metering supervision system according to claim 3 is characterized by: High-precision alignment is handled as follows: Step D1: Calculate the deviation of each device data after time correction, and define the relative deviation of device i at time k as: (Formula 19); in, represents the average measurement value of all devices at the kth moment; Calculate the standard deviation of the deviations to assess the overall alignment effect: (Formula 20); in represents the average relative deviation at the kth moment; Step D2: Dynamically adjust the weight of each device data according to the deviation. The smaller the deviation, the greater the weight. (Formula 21); where w i（k） Represents the weight coefficient of the i-th device at the k-th moment, satisfying ; Step D3: Use (Formula 22); Calculate the fusion measurement value v at the kth moment k final .

5. The data center-based intelligent metering supervision system according to claim 4 is characterized in that: The adaptive adjustment method is as follows: Step E1: Real-time monitoring of the operating status of key algorithm parameters, including interpolation error, drift rate change, and synchronization deviation: (Formula 23); where e interp represents the interpolation error (unit: original measurement unit), v i（k） true Indicates the real value obtained by high-precision equipment; Step E2: When the interpolation error e interp Exceeding the threshold Th interp When , reduce the target sampling period T0: (Formula 24); Where T10 represents the adjusted sampling period, ensuring that T10 ≥ 0.1 seconds; Step E3: Re-estimate the drift parameter every 7 days. The update formula is: (Formula 25); (Formula 26); Where: i（0） new , β i（1） new represents the updated parameters; β i（0） old , β i（1） old represents the parameters before updating; β i（0） est , β i（1） est Represents the newly estimated parameters; α is the preset smoothing coefficient.

6. The data center-based intelligent metering supervision system according to claim 5 is characterized by: The data fusion verification method is as follows: Step F1: Group the spatiotemporally aligned data by device type to form a validation dataset. V g ={v g（1）， v g（2）， ……,v g（n） }, where g represents the device type group and n represents the number of devices in the group; Step F2: Calculate the coefficient of variation for each set of data: (Formula 27); in: represents the average value of the data within the group; Represents the standard deviation of the data within the group; CV g represents the coefficient of variation, and it is dimensionless; Step F3, when CV g When the error rate is less than 5%, the data consistency of this group is determined to be qualified; otherwise, the abnormality troubleshooting process is triggered; For equipment that fails inspection three times in a row, a calibration work order is automatically generated to prompt maintenance personnel to perform hardware repairs.

7. The data center-based intelligent metering supervision method is characterized by: The method is implemented by the data center-based intelligent metering supervision system described in any one of claims 1-6.

Images